Enable global logging of problematic tokens encountered during analysis. * tokens that have no stress information * exclude words that are expected to be missing stress marking * proper nouns * prepositions * out-of-lexicon tokens * etc.?