chunked/nltk_chunking/main_classes_tokenizer/chunk_10.txt

): adding them, assigning them to attributes in the
  tokenizer for easy access and making sure they are not split during tokenization.