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

Direct link:

Fast Mapping Nouns on a Tablet-Based Language Screener: Two-Year-Olds Remember Only Their First Guess

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

Integrative Statement

A central and intractable question in the field of language development concerns the problem of referential ambiguity when children are exposed to new words. Two models have been proposed to characterize the mechanism underlying children’s word learning in situations of referential uncertainty. Children may track associations between words and multiple potential referents across many situations to ultimately converge on correct mappings (i.e., cross-situational models; Yu & Smith, 2007). Alternatively, children may form a single hypothesis about each novel word’s meaning, maintaining this hypothesis to be verified or falsified upon subsequent occurrences (Trueswell et al., 2013). This “one-track” learning mechanism, termed “propose-but-verify,” received support from a test of fast mapping ability (Carey & Bartlett, 1978) with children ages 3-5 on the Quick Interactive Language Screener (QUILS; Aravind et al., 2018), a touchscreen assessment of language products and process (Golinkoff et al., 2017).

This paper reports on a test of fast mapping ability from a downward extension of the tablet-based QUILS with 2-year-olds, addressing the question of word learning mechanism with this younger population, because less-experienced word learners could be more likely to use an associative mechanism. We ask what is retained from prior exposures to a novel word and whether what is retained changes from 24 to 36 months. Six items testing “fast mapping nouns” each included two trials: Trial 1 presented a novel noun with 2 known objects and a nonce object (target) on the screen, and Trial 2, which immediately followed Trial 1, tested children’s ability to extend the word to a different member of the same category (Figure 1). If children tracked associations between the word and multiple possible referents on Trial 1, then children could succeed at Trial 2 even after failing on Trial 1. If, however, they used the propose-but-verify mechanism, then success on Trial 2 would be contingent on Trial 1 success.

A sample of 100 2-year-olds (M age=30.32 months; SD=3.49) completed all 6 fast-mapping trials. When children were successful on Trial 1, they were successful on Trial 2, but when inaccurate on Trial 1, performance hovered around chance (33%) on Trial 2 (Figure 2). Logistic mixed regressions confirmed that Trial 2 performance was greater than chance when children succeeded on Trial 1 (95% CIs of [56.5%, 79.7%]), but was at chance when children failed on Trial 1 (95% CIs of [25.8%, 44.4%]). A mixed-effects logistic regression was fit predicting Trial 2 accuracy from Trial 1 accuracy, including age as an interacting fixed factor. Results indicate a significant interaction of Trial 1 and age (β=0.0490, SE=0.001, z=33.018, p<0.001). When children succeeded on Trial 1, retention of the single hypothesized meaning improved from 24 to 36 months. However, when children failed on Trial 1, there was no improvement with age on Trial 2.

In sum, 2-year-olds, like 3- through 5-year-olds, did not retain associations of word-to-world other than hypothesized mappings. Additionally, retention of hypothesized mappings, but not of other non-mapped associations, improved with age. These results provide further support for the “propose-but-verify” mechanism of word learning.

Authors