The long-term aim of this subproject is to evaluate computational grammars of Polish, both quantitatively (what percent of constructions covered by the test-suite are also covered by the grammar?) and qualitatively (does the resulting parse contain all the information available in the corresponding test-suite annotation?).
We describe the design of this test-suite, discussing various conceptual decisions we made, including the employment of HPSG-like AVM representations as the annotation formalism. We also present the classification of grammatical phenomena of Polish used for indexing the sentences in the test-suite. Finally, we present the graphical user interface constructed in order to facilitate entering and viewing test sentences, their indices and their AVM annotations. (We are ready to provide a demo.)