Poster #106 - Does bilingual language experience bolster dual-label cross-situational word learning?

• In Event: 3-106 - Poster Session 11
  In Poster Session: PS 11 Section - Language, Communication

Integrative Statement

Infants learn many more words for objects than those that are explicitly labelled. Cross-situational learning is a possible mechanism for this impressive learning. It entails the process of learning by tracking patterns of labels and objects across multiple events, without overt naming (Smith & Yu, 2008). For the bilingual infant, the task of mapping a label to a referent is even more complex than for monolinguals, as she must learn two labels, one in each language. Despite this added challenge, bilinguals reach language milestones around the same age as monolinguals (Pearson, Fernandez, & Oller, 1993) and have vocabulary sizes equal to monolinguals when both languages are considered (Core, Hoff, Rumiche, & Señor, 2013). Recent work with monolinguals has shown that 18-month-olds can pair two labels to a novel referent using cross-situational statistics (Verde, Antovich, & Graf Estes, 2018). The present research examines whether bilinguals can achieve the same feat, given their natural prior experience mapping multiple labels to the same referent.

We tested bilingual 18-month-olds’ ability to map two labels to the same novel referent using cross-situational learning. In training, infants were shown two objects simultaneously while hearing one label for each object; two distinct labels for each object were presented across trials. During test, infants saw two objects but heard repetitions of one target object’s label. Infants’ looking to the screen was recorded and then coded frame-by-frame (Fernald, Zangl, Portillo, & Marchman, 2008). We examined whether infants looked reliably longer to the target object in response to hearing the label, a measure of successful label-referent mapping.

Our preliminary results show bilinguals did not look reliably longer to the target objects in the first block of 8 test trials (N=15, t(14) = -0.99, p = .34), nor the second block of trials (N= 12, t(11) = -.049, p = .64). This contrasts with prior results where monolinguals successfully learned dual labels by the second testing block. This is surprising given that bilinguals, but not monolinguals, have natural, everyday experience assigning dual labels to a single referent. If existing dual-labeling experience supports cross-situational dual-mapping, we would anticipate bilinguals would outperform monolinguals on this task.

This is the first study to explore cross-situational dual-word-learning in bilingual infants but we caution these results are preliminary. Bilinguals might fail to map dual labels in this paradigm for several reasons. For example, the dual labels here shared the same phonology; bilinguals may need more distinct phonological cues that mirror their natural language exposure. Additionally, research suggests bilingual children apply mutual exclusivity less stringently than monolinguals (Davidson & Tell, 2005). This difference may make dual-mappings more ambiguous for the bilingual, possibly necessitating more trials to disambiguate. We are currently collecting more data and will analyze looking during training in order to explore the role of mutual exclusivity in bilingual infants’ dual-label learning.

Authors

• Erica Verde, University of California-Davis
  Presenting Author

• Dylan Antovich, UC-Davis
  Non-Presenting Author

• Katharine Graf Estes, University of California, Davis
  Non-Presenting Author