What is frequency per million (also called relative frequency)?

    The frequency per million is a normalized frequency describing how many times a particular word/lemma/tag/phrase (depending on your query) would occur in 1 million tokens. It can be used for comparing occurrences across corpora with different sizes in contrast to the total number of occurrences which is almost always based on the size of the corpus. 

    This frequency is counted as follows:

    number of hits / corpus size in millions of tokens = frequency per million
    More information can be found on our website. 

    see also:
    Published