A character vector of verb lemma that indicate speech (e.g., say, state). A default list is included in verb_lemma('quote'), but certain lemma might be more accurate/appropriate depending on the corpus.