Include {} as punct words, as some parsers don't produce LCB/RCB

Author: AngledLuffa

src/edu/stanford/nlp/trees/PennTreebankLanguagePack.java

@@ -29,7 +29,7 @@ public PennTreebankLanguagePack() {
   private static final String[] collinsPunctTags = {"''", "``", ".", ":", ","};
 
   // new tokenizers should return (), old tokenizers return -LRB- -RRB-.  so we anticipate both
-  private static final String[] pennPunctWords = {"''", "'", "``", "`", "-LRB-", "-RRB-", "(", ")", "-LCB-", "-RCB-", ".", "?", "!", ",", ":", "-", "--", "...", ";"};
+  private static final String[] pennPunctWords = {"{", "}", "''", "'", "``", "`", "-LRB-", "-RRB-", "(", ")", "-LCB-", "-RCB-", ".", "?", "!", ",", ":", "-", "--", "...", ";"};
 
   private static final String[] pennSFPunctWords = {".", "!", "?"};