Natural sentences that contain *any*

We scraped the Gutenberg Project and a subset of English Wikipedia to obtain the list of sentences that contain any. Next, using a combination of heuristics, we filtered the result with regular expressions to produce two sets of sentences (the second set underwent additional manual filtration): * 3844 sentences with sentential negation and a plural object with any to the right to the verb; * 330 sentences with nobody / no one as subject and a plural object with any to the right.

The first set was modified to substitute the negated verb by its non-negated version, so we contrast 3844 sentences with negation and 3844 affirmative ones (neg vs. aff). In the second dataset, we substituted nobody for somebody and no one for someone, to check the some vs. no contrast.

You can use our script to find sentences with negation and any in any given English corpus:

python dataset_preparation/select_sentences_from_real_text.py <corpus.txt>