Welcome to ELIA - Eyegaze Language Integration Analysis
ELIA is a free software tool that supports the analysis of eye-tracking data recorded for studies in human language processing. It allows for rapid integration of gaze data with spoken language input (either live or pre-recorded).
Financial support for the creation of the project was provided by a Social Sciences and Humanities Research Council of Canada operating grant awarded to S. Graham and C. Chambers as well as through funding provided by the University of Calgary.
If you are a researcher who uses an eye-tracking system for experimental research, you have surely struggled with issues pertaining to data collection and analysis. While newer eye-tracking and experimental design software now offer support for synchronizing the timing of recorded audio prompts with a test subject's gaze response, there is still little support for subsequent statistical analysis. A typical experiment involves many hours of error-prone manual manipulation of recorded data within a spreadsheet before a simple time-course analysis plot can be viewed or before data can be imported into a statistical package for further analysis. ELIA aims to hit the sweet spot between data collection software such as E-Prime and statistical analysis software such as SPSS, to dramatically reduce the manual effort of transforming your data to prepare for statistical analysis.
With ELIA, you'll be able to do a time-course analysis quickly and easily at any point during your data collection. If an experimental problem is revealed after collecting data from the first few test subjects, then you'll have the opportunity to fix the problem before having invested enormous effort in data collection and manipulation.
The Language and Cognitive Development Lab at the University of Calgary has recently been using a Tobii eye tracker with E-Prime Extensions for Tobii. Dr. Susan Graham and PhD student Jared Berman found that an enormous amount of time and effort was required to transform the recorded data before they could view a time-course analysis or import the data into SPSS for further analysis. As a consequence, we began developing ELIA with the goal of contributing it as an open-source project for the benefit of the larger community of eye-tracking researchers in psychology and linguistics. It is the hope that others in the community will view ELIA as a base for ongoing development and contribute further towards making it more widely useful. For information about how you can contribute, see Contributing to the ELIA Project.
- Ian G. Graham, B.Sc.: Software Developer
- Jared M.J. Berman, M.Sc.
- Susan A. Graham, Ph.D.
Current features of ELIA
Currently, ELIA has only been used to process data recorded by a Tobii E-Prime Eye Tracking Lab. To gain support for your eye-tracking system, please contribute by attaching samples of your data to an enhancement request, using this example request.
- Import of Tobii/E-Prime .gazedata files, although other comma or tab separated text files may also work
- Establish window of interest within each trial, defined by the start and end of specific spoken words(or other temporal events)
- Align time 0 within a trial to the start of a specific spoken word(or other temporal event)
- Export of overall time-course analysis in .csv file
- Export of per-subject time-course analysis in .csv file
- Support for removal of "looks"/gazes that occur before a specific time within a trial
- Support for proportioning of looks to exclude periods in which no gaze was associated with the defined areas of interest
- Integration of live speaker input with gaze data
The ELIA Wiki
ELIA uses Trac for its public face. Trac provides both a Wiki and bug-report/enhancement-request system.
For a complete list of local wiki pages, which includes extensive documentation about using Trac, see TitleIndex.
Bug Reports and Enhancement Requests
When creating and editing bug reports, all WikiFormatting is supported, including links within Trac as described in TracLinks. Referencing another bug report within Wiki-formatted text can be done simply by preceding the bug number with a # sign.