Word Span Task

The task follows the word span of Kane and colleagues (2004), as Unsworth and Engle (2006) describe it: short nouns shown one per second, three lists of each length from 2 to 7 words in a random order, 18 lists and 81 words in all, and recall in order right after each list, with a word counted only in the place it was shown. Here recall is a click on the words, as the automated complex span tasks recall letters (Unsworth et al. 2005), so it needs no typing or spelling in any language. Every list is drawn from the same twelve words, as letter span draws from a fixed set of letters. Settings give one or two lists of each length, or a longest list of 5 to 8 words. Two practice lists, of 2 and 3 words, show the words shown, the answer and how many were in place.

The twelve words are everyday nouns written for this test, the same twelve things in each of the eight languages: things of twelve different kinds, short in every language, with no two starting alike or rhyming in a language and none used by another word test of the library. Every list is drawn from the session's seed, with no word twice in a list.

The main score is the share of words recalled in the place they were shown, the proportion correct score Unsworth and Engle (2006, 2007) report for span tasks; they prefer it to counting only lists recalled perfectly for its better psychometric properties. The partial span score (the same as a count), the absolute span score (the summed lengths of lists recalled perfectly), the partial-credit unit score (Conway et al. 2005), the span (the longest length at which at least two of the three lists were recalled perfectly), lists recalled perfectly, words recalled in any place and recall at each list length are also given. Scores support research and do not diagnose.