LEX-ana: a lexical database from Nguni-medium children's stories

LEX-ana is a lexical database built from children's literature in four Nguni languages of South Africa: isiXhosa, isiZulu, siSwati and isiNdebele. It contains 130,376 unique word forms, drawn from a corpus of 501,324 words across 2,077 stories from seven digital sources. For each word form, LEX-ana provides frequency (raw and Zipf), contextual diversity, word length, orthographic neighbourhood measures (Coltheart's N, OLD20 and neighbour frequency) and sublexical characteristics such as consonant clusters. Each language was cleaned and analysed separately. Files: LEX-ana database (one row per unique word per language with annotations)- LEX-ana_ND_28092026.xlsx, LEX-ana_SW_28092026.xlsx, LEX-ana_XH_28092026.xlsx, LEX-ana_ZU_28092026.xlsx- LEX-ana_all_langs_28092026.xlsx: all four languages in one file- LEX-ana_README.txt: definitions of all 33 variables Metadata files (cleaned corpus, one row per token with source metadata)- Metadata_ND_28092026.xlsx, Metadata_SW_28092026.xlsx, Metadata_XH_28092026.xlsx, Metadata_ZU_28092026.xlsx- Metadata_all_langs_28092026.xlsx: all four languages in one file Grapheme inventory- grapheme_frequencies_final.xlsx: the graphemes of each language (one tab per language) with their frequencies in LEX-ana Processing pipeline- LEX-ana_pipeline.zip: the scripts that clean the tokenised corpus and build LEX-ana (R and Python), the mother-tongue speaker check workbooks, the removed tokens of each language (every token removed during cleaning, with the reason), and README_pipeline.txt, which describes every step. Unzip and run from the unzipped folder. The raw, uncleaned tokens are not included due to copyright issues. The pipeline is shared so that others can adapt the same processing steps to their own corpora. Orthographic neighbourhood measures were calculated with Jiwar (Alzahrani, 2025) and LexiCAL (Chee et al., 2021), which are not included here. See README_pipeline.txt for where to get them. HOW TO CITECox, P., Berghoff, R., Bowles, T., & Shibeshi, Z. (2026). LEX-ana: a lexical database from Nguni-medium children's stories (Version 1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.23137656 How the name LEX-ana came about: LEX-ana joins LEX-, from lexicon, with -ana (pronounced AH-nah), the diminutive suffix shared by all four Nguni languages in the database. The four languages share it because they inherited it: it goes back to the Proto-Bantu word meaning “child”. The name therefore means “little lexicon”, which suits a lexicon built from children’s literature.