chunked/nltk_chunking/_add_new_pipeline/chunk_4.txt

Try to keep these inputs as pure Python as possible
as it makes compatibility easier (even through other languages via JSON).