Individual Submission Summary
Share...

Direct link:

Poster #12 - Timing of naming during children’s cross-situational word learning

Fri, March 22, 2:30 to 3:45pm, Baltimore Convention Center, Floor: Level 1, Exhibit Hall B

Integrative Statement

The ability to map words to referents across situations involves registering and remembering previous naming encounters (Yu & Smith, 2007; Smith & Yu, 2008; Smith & Yu, 2013). The timing between naming instances may therefore matter for children’s ability to track word-referent statistics. Previous research has shown that young infants benefit from naming instances being closer together in time when faced with cross-situational word learning (Vlach & Johnson, 2013). Here, we examine if the timing between naming instances is linked to preschooler’s ability to track word-referent statistics.

We presented 4- to 7-year-old children (N=29, Mean age = 5.5 years, 19 females) with a cross-situational word learning task in which children had to learn 8 word-object pairings (Suanda, Mugwanya, & Namy, 2015). During training, children saw two words and two objects at a time; word-referent pairings were ambiguous within a trial, but across trials, each word occurred 3 times and most often with its referent object. Critically, we randomized training trials such that objects varied on the number of trials (where other objects were named) between the first and second naming instance (1-6 trials) and the second and third naming instance (1-6 trials). After training, children were tested on their learning of the word-referent pairings. On each test trial, a word was presented with its target object and a distractor object (an object that had occurred once during training with the target word). Children were asked to select the object that was being named. Each object was tested once for a total of 8 test trials.

Mean accuracy scores for the full sample showed that children’s accuracy scores at test were not different than chance performance [M = 0.56, SD = 0.23; t(28)=1.32, p=0.20]. Additionally, children’s mean accuracy scores did not vary by age [t(27)=0.46, p=0.65]. Children’s accuracy scores did vary, however, by the number of trials presented between the first and second naming instance [logistic mixed effects model on trial-by-trial data: b=-0.26, STE=0.10, z=-2.58, p=0.01]. Follow up analyses confirmed that children successfully learned the names of objects with 1-2 training trials between the first and second naming instance [t(28)=3.69, p=0.02]. Object-labels with 3 or more trials between the first and second naming instance were not learned [t(28)=-0.26, p=0.8]. This effect was specific to the second naming encounter, as the number of training trials between the second and third naming instance were not linked to accuracy scores [b=0.20, STE=0.12, z=1.67, p=0.10].

Currently, we are working on 1) replicating the findings, and 2) examining if the third naming instance matters to children’s word-referent learning in this task. These results do suggest, however, that, similar to young infants, preschoolers’ statistical word-referent learning is affected by the temporal dynamics of naming instances. Early naming instances that are closer together in time support children’s ability to track word-object co-occurrences.

Authors