BackHome
Refresh: OffView Personal Schedule
Individual Submission Summary
Share...

Direct link:

Morphing Meanings: Using a Digital App to Investigate Vocabulary Learning Beyond Simply Right or Wrong

Fri, March 22, 1:00 to 2:30pm, Hilton Baltimore, Floor: Level 2, Key 2

Integrative Statement

Preschoolers’ vocabulary predicts literacy and reading skills through 4th and 5th grade (Burchinal et. al, 2016; Dickinson & Porche, 2011). However, children’s semantic representations are initially imprecise (a coop is on a farm) before going through a “semantic narrowing” process (Bergelson & Aslin, 2017), towards a more specific definition (a coop is a house where chickens live).

We developed a digital measure probing these varied levels of representation as children learn words. Children selected one of four images to match a word: a target, a phonological foil that sounded like the target (e.g., soup for coop) and two semantic foils including something used with the target (e.g., broom for dustpan); and something in the same category as the target (e.g., marker for pencil; Heuttig & Altmann, 2005). Arguably, selecting a semantic foil reflects deeper knowledge than selecting the phonological foil.

A recent intervention involving book reading and play used this measure at pre- and posttest. The systematic foil types, unique to this new measure, allowed analysis of the selections made when children erred, providing additional information about learning progress. Low-income preschoolers (N = 140) participated in four, 4-week units; each teaching 20 words.

Overall, children showed significant learning, choosing the target more at post-test (M = 42.5%) than pre-test (M = 24.5%), (B = 0.28, p < .001). When children did not choose the target at pre-test, phonological errors were more common (M = 44.8%; Table 1) than would be expected by chance (33%), likely because children did not know the word and selected a similar sounding item. At post-test, phonological errors decreased (B = -0.08, p < .001), while semantic errors increased (B = 0.08, p < .001). Similar patterns did not emerge on control words that children were not taught.

Errors on individual words in relation to children’s overall accuracy also proved interesting: children who answered more items correctly overall at post-test were more likely to choose semantic foils and less likely to choose phonological foils when they erred (B = 0.58, p < .001; Figure 1). Words were also tested using an expressive measure (Hadley et al., 2016). Of children who selected a semantic foil, 17.7% provided at least one relevant detail on the expressive test compared to 11.5% of children who selected a phonological foil, showing that when children make errors on our receptive measure, choosing semantically-related images indicates more knowledge about the word than choosing phonological foils.

While children showed learning for intervention words but not control words these data also suggest that children’s semantic representations deepened even when they were not sure of the word’s precise meaning. When children erred at posttest, they were more likely to choose a semantically related foil, suggesting that children learned something about the new words’ meanings. Our digital measure was sensitive to this incremental learning in a way that other measurements are not and affords the possibilities of a) tracking the refinement of word meanings; and b) discerning why children find some words difficult to learn.

Authors