PoKi is a corpus of 61,330 poems written by children from grades 1 to 12. PoKi is especially useful in studying child language because it comes with information about the age of the child authors (their grade).

Poems were submitted online by children in school grades 1-12. Permissions were obtained from domain host prior to scraping. Poems were webscraped on June, 2018 from http://teacher.scholastic.com/writewit/poetry/jack_readall.asp.

This repository includes three versions of PoKi and R analytic scripts for reproducing the results.

Details of the data are in this paper:
PoKi: A Large Dataset of Poems by Children. Will E. Hipson, Saif M. Mohammad. arXiv preprint arXiv:2004.06188 . April 2020.

BibTex:
@article{hipson2020poki,
title={PoKi: A Large Dataset of Poems by Children},
author={Hipson, W. E and Mohammad, S M.},
journal={arXiv preprint arXiv:2004.06188},
year={2020}
}

PoKi is made freely available for research with the condition that the research be used for the benefit of children. Anyone who uses PoKi must cite this paper https://arxiv.org/abs/2004.06188 (see BibTex below for full citation). Should you intend to use PoKi in your research, please reach out to us ([email protected]); we'd love to hear how you're using it!

The table below provides a brief glimpse of how the poems in PoKi are distributed by age (grade).

Description: Dataset containing the raw web-scraped poems from http://teacher.scholastic.com/writewit/poetry/jack_readall.asp. No text pre-processing at this point. Note that there may be a few extra poems beyond the number reported in the manuscript as some poems contain very little text and are eventually removed during the preprocessing steps.

Variables:
id = unique identifier for each poem
title = title of poem (submitted by poem author)
author = first name of author
grade = school grade (1-12)
text = raw poem text
char = number of characters

Description: Lemmatized version of PoKi. We used lemmatization tools in the Natural Language Toolkit (http://www.nltk.org/) to lemmatize poem text (e.g., running, ran, runs -> run | writing, wrote, writes -> write). This dataset contains the same columns as in poki.csv, where 'text' now contains lemmatized text.

Description: Data used for word-level analysis reported in Hipson and Mohammad (2020) PoKi: A Large Dataset of Poems by Children. We use the NRC VAD lexicon and the NRC Emotion Intensity lexicon.

Variables:
id = unique identifier for each poem
grade = school grade (1-12)
author = first name of author
_sd = within poem standard deviation for emotion tokens
valence = mean valence
arousal = mean arousal
dominance = mean dominance
anger = mean anger
fear = mean fear
sadness = mean sadness
joy = mean joy
total_words = total number of non stop words within poem
gender = probable gender based on author name