Split Texts Into Units of Analysis

Cuts each document into the unit every stage of an analysis should share. Character offsets are returned alongside each unit, so any span stays recoverable from the original text.

Usage

split_texts(texts, unit = c("sentence", "paragraph", "document"))

Arguments

texts

Character vector of documents, optionally named. Names become doc_id; positions are used when absent.

unit

"sentence" (default), "paragraph", or "document". Paragraphs split on blank lines; sentences split after ., !, or ?.

Value

A tibble with doc_id, unit_id, unit_text, start, and end. unit_id is the doc_id itself for whole documents, and doc_id.n otherwise. Empty and missing documents contribute no rows.

See also

apply_codes(), which splits with this before coding; uncoded_units() for the units a codebook did not reach.