Languages

You pass the language code.

The blank import registers that language. Import only the languages you need. The bundle is that language's part-of-speech model and rarity table, gzipped.

Tagging is what keeps verbs out of phrases. Each language has its own model. A word stays labelled as a verb when the tagger marks it as a verb in at least half of its occurrences in the document. One wrong label does not override that. Italian is trained on less data than the others, and it is the weakest of the five. See How it works.