From this page you can download the tool to convert the Penn WSJ Treebank into Tesnière Dependency Structure (TDS).
The tool is able to convert the mrg files of the Penn WSJ Treebank or any file containing structures with the same annotation style.
It is strongly recommended (but not required) that you preprocess the Penn Treebank structures with Vadas&Curran patch which adds extra bracketing levels to noun phrases.
You can find more information about the conversion procedure and the annotation style in our TLT paper.
The first attempt to parse TDS structures is described in my ACL SRW paper.
In agreement with the Linguistic Data Consortium, you can download a pdf file containing the first 100 converted structures of the Penn Treebank.

Downloads: TDS_1.2.jar   (last update 01.09.2011) readme file wsj00_TDS_1.2_first100.pdf changes in version 1.2 all versions Instructions: Download the jar file in a folder of your computer. From the terminal go to the the folder with the jar file and type:           java [-Xmx100M] -jar TDS_1.1.jar [-log logFile]     (arguments in square brackets are optional). Requirements: Java version 1.5 or above     (check with java -version from a terminal). Feedback: Comments, mistakes, suggestions are always welcomed at: f (dot) sangati (at) uva (dot) nl