//key: a single sentence in both languages and alignment
//ignore value. each key is parallel sentence and its alignment, in xml format
ParallelChunk c = pcr.parseString(key.toString());
ok.set(c.idString());
//Chunk is an array of tokens in the sentence, without any special tokenization (just separated by spaces)
Chunk fc = c.getChunk(src);
Chunk ec = c.getChunk(tgt);
if (fc == null || ec == null) {