Mean bigram probability (type-based, non-positional), a phonotactic-probability proxy

For each word, the mean over its adjacent letter bigrams of the corpus bigram probability (count divided by the total bigram count). Computed from integer counts and rounded, so it is identical in the R and Python engines.

Usage

add_bigram_frequency(df, reference = NULL)

Arguments

df

A data frame with a word column.

reference

A character vector of reference words (defaults to df$word).

Value

df with an added numeric bigram_freq column.