Fast Mapping in 2-year-old Children’s High-knowledge and Low-knowledge Semantic Categories

• In Event: Poster Session 05
  In Poster Session: PS 05 Topic - Language, Communication

Abstract

Research increasingly points to the importance of semantic knowledge in bootstrapping new word learning. Borovsky et al. (2016) demonstrated better immediate retention of new word mappings acquired through ostensive labeling in higher density category domains for which children produced relatively higher proportions of words. Here we investigated retention of new words after a 5-minute delay as a function of vocabulary and category-domain density. We were particularly interested in the impact of semantic density on retention of words acquired through fast mapping given prior work showing poor retention after this form of word learning compared to ostensive labeling (Horst & Samuelson, 2008; Kucker & Samuelson, 2012).

Children (N=17) ages 23- to 25-months received 4 exposures to each of four-novel labels and referents during training. Mappings drew from 6 common categories in child language (vehicles, animals, fruits, drinks, body parts, clothing). We used the MacArthur-Bates CDI to determine the number of words children produced in each category. and from there calculated a proportion density score for each. We identified each child’s two highest-density and two lowest-density domains before training to tailor the study to each child’s specific knowledge. During a training trial, children viewed a novel object (e.g., a hedgehog) in the presence of a familiar category member (e.g., a dog) and heard “Where’s the vode?” After a 5-minute delay, we measured children’s eye-movements in test trials where they were asked to “Find the ” from an array of two high-knowledge or two low-knowledge novel objects. Children received two test trials for each new mapping to counterbalance the side of the target (right, left). Proportion looking time (children’s looking time to the target divided by their total looking time to the target and distractor) served as our dependent measure. We then separated children into High and Low Vocabulary groups using a median split on total productive vocabulary.

We conducted a 2 (Vocabulary: low, high) x 2 (Domain knowledge: low, high) ANOVA with Vocabulary a between-subjects factor and Domain knowledge a within-subjects factor. We observed a significant interaction of Vocabulary and Domain Knowledge, F(1, 17) = 4.982, p < 0.05. The High Vocabulary Group performed significantly above chance in their high-knowledge domain M = .71, t(4)= 4.16, p < 0.05 and at chance in their low-knowledge domain M = .52, t(5) = .294, p = .78. The Low Vocabulary Group performed at chance in both knowledge domains (See Figure 1).

Overall vocabulary size and semantic category density together impact two-year-olds’ ability to retain novel object mappings after a 5-minute delay. Importantly, children with larger vocabularies did not show universal success in learning, as might be expected. Rather, they retained words only in relatively dense semantic categories. Children with smaller vocabularies show poor short-term retention regardless of their domain knowledge. Work is underway to measure learning after a longer delay and opportunities for consolidation.

Authors

• Cassandra Elisabeth Watters, University of Arizona
  Presenting Author

• Arielle Borovsky, Purdue University, West Lafayette
  Non-Presenting Author

• Lucia Sweeney, University of Arizona
  Non-Presenting Author

• Rebecca Gomez, The University of Arizona
  Non-Presenting Author