Using the Floresta Sintática corpus, or converting it to Treebank format, it's possible to use ClearNLP, since in TLanguage "Portuguese" isn't listed?