Modeling the frequency of phrasal verbs with search engines

There are well over a thousand phrasal verbs in English. For non-native speakers they are notoriously difficult to remember and use in the right context. We tried to construct a ranking of phrasal verbs according to their estimated occurrence frequency, based on quantitative information available from the public indexable Web. Technically, we used major Web search engines to acquire phrase-occurrence statistics, measured consistency between the rankings implied by their results and confirmed that a rough set of ‘classes’ of phrasal verbs can be distinguished. While this technique relies on inaccurate and possibly biased estimation functions, we show that the overall distribution of ranks seems to be consistent among all the queried search engines operated by different vendors.