The developmental English Lexicon Project (d-ELP): Continuous word reading difficulty estimates of the 9,961 most frequently printed English words for US children

Abstract The developmental English Lexicon Project (d-ELP) is a database comprising word reading data for 9,961 of the most frequently printed English words for children, including continuous item response theory (IRT)-based estimates of word reading difficulty, accompanying word characteristics, and raw trial-level data. Data on isolated word reading accuracy were collected from 1,907 children in grades 1–5 across multiple US sites. Each child read a list of words ranging from 250 to 500 items depending on reading skill (less proficient children were often unable to finish the entire list of words), with each list representing the full distribution of word difficulty oversampled for words at the child’s reading level. Each child also attempted 99 linking items representing the full distribution of word difficulty. A total of 631,322 responses were collected, with the average child providing 336. For the final sample of words, the average number of responses per word for the 9,862 non-linking items was 47.14 (range of 10–592), and for the 99 linking items the average was 1,681.42 (range of 1,083–1,907). A Rasch model, with planned missing data design, was fit to children’s responses on the word list and linking item responses. This introduction describes the motivation for this project, the methods used for data collection, the analytic framework used to establish a continuous measure of word reading difficulty across words, descriptive statistics of the relational database and initial behavioral results, and a publicly available search engine providing searchable access to word difficulty ratings and associated word-level statistics.