A hand-corrected syntactic annotation of the Dundee eye-tracking corpus

Authors

Cory Shain, Marten van Schijndel and William Schuler
The Ohio State University
Eliminate the risk of parser noise in syntax-based reading time studies using the Dundee eye-tracking corpus. This annotation distributes by default with the Modelblocks repository, which you can access at the link below.
Because of licensure restrictions, we only distribute syntactic trees, not the Dundee source texts. To generate the complete source trees:
  1. Git clone Modelblocks and navigate into the golddundee directory.
  2. Run make gold. The build will exit with a warning about the contents of ../config/user-dundee-directory.txt.
  3. Edit ../config/user-dundee-directory.txt to contain a path to your Dundee directory.
  4. Run make gold again. This should generate the files srcmodel/dundee.gold.linetrees (which contains computer-readable single-line trees) and srcmodel/dundee.gold.edit.editabletrees (which contains human-readable indented trees).
If you have questions or comments, please email Cory Shain (shain.3@osu.edu).

Works cited: