from nltk.corpus import ptb

verbforms = set(['break', 'breaks', 'broke', 'breaking', 'broken'])

def has_break(list):
    for wordtag in list:
        if wordtag[0] in verbforms and wordtag[1].startswith('VB'):
            return True
    return False

break_trees = [t for t in ptb.parsed_sents() if has_break(t.pos())]