Make a stemmer from a set of (term, stem) pairs.

new_stemmer(term, stem, default = NULL, duplicates = "first",
            vectorize = TRUE)



character vector of terms to stem.


character vector the same length as term with entries giving the corresponding stems.


if non-NULL, a default value to use for terms that do not have a stem; NULL specifies that such terms should be left unchanged.


action to take for duplicates in the term list. See ‘Details’


whether to produce a vectorized stemmer that accepts and returns vector arguments.


Giving a list of terms and a corresponding list of stems, this produces a function that maps terms to their corresponding entry. If default = NULL, then values absent from the term argument get left as-is; otherwise, they get replaced by the default value.

The duplicates argument indicates the action to take if there are duplicate entries in the term argument:

  • duplicates = "first" take the first matching entry in the stem list.

  • duplicates = "last" take the last matching entry in the stem list.

  • duplicates = "omit" use the default value for duplicated terms.

  • duplicates = "fail" raise an error if there are duplicated terms.


By default, with vectorize = TRUE, the resulting stemmer accepts a character vector as input and returns a character vector of the same length with entries giving the stems of the corresponding input entries.

Setting vectorize = FALSE gives a function that accepts a single input and returns a single output. This can be more efficient when used as part of a text_filter.

See also

stem_snowball, text_filter, text_tokens.


# map uppercase to lowercase, leave others unchanged stemmer <- new_stemmer(LETTERS, letters) stemmer(c("A", "E", "I", "O", "U", "1", "2", "3"))
#> [1] "a" "e" "i" "o" "u" "1" "2" "3"
# map uppercase to lowercase, drop others stemmer <- new_stemmer(LETTERS, letters, default = NA) stemmer(c("A", "E", "I", "O", "U", "1", "2", "3"))
#> [1] "a" "e" "i" "o" "u" NA NA NA