This study uses computational methods from corpus linguistics and NLP to analyze how witches are linguistically described in a large corpus of popular English‑language children’s books from the past century. By examining key markers such as adjectives, verbs, and adverbs, it identifies patterns in how witches—often gendered and politicized figures—are characterized, with early texts showing simpler, largely negative descriptors (e.g. “old,” “wretched”) and later works offering more nuanced but still appearance- and age-focused portrayals. The research reflects on how these textual patterns may reinforce or challenge societal notions about (especially older) women and illustrates the value of computational approaches for studying character representation in children’s literature.
Anna Mihlic (Mon,) studied this question.