Poster #57 - Environmental Context and Referential Ambiguity: a Human Simulation Paradigm Study

• In Event: Poster Session 06

Abstract

Background
Word learning theories differ in their depictions of how environmental context shapes learning. While some stress its invaluable role (Rowe & Snow, 2020), others stress the ambiguity and challenge the environment poses for learning (Gleitman & Trueswell, 2020). Consistent with the latter view are studies employing the Human Simulation Paradigm (HSP; Medina et al., 2011), where adults watch muted vignettes of parent-toddler interactions and guess “mystery words” uttered in those vignettes. HSP research shows that despite access to social and environmental cues, adults struggle to identify the mystery words, suggesting a level of ambiguity insurmountable via environmental cues alone (e.g., Medina et al., 2011). Interestingly in most HSP studies, mystery words are unrelated to one another. In toddlers’ everyday experiences however, words are contextually structured, with words (e.g., “ball”) occurring in some contexts (e.g., play) more so than others (e.g., feeding; see Tamis-LeMonda et al., 2019). In the current study, we adapt the HSP to examine whether organizing vignettes by context leads to better identification of word meaning.
Methods
A picture-book version of the HSP was developed where participants see textless scenes and guess a target word present in the original text. Based on an initial norming study, we identified six referentially-ambiguous scenes (see Medina et al., 2011). 69 participants then participated in the primary study where these referentially-ambiguous scenes were preceded by scenes (“cue trials”) thematically related to the target word in the ambiguous scene (thematic condition) or by scenes that were not thematically related to the target word (control condition; see Figure 1).
Results
Participants in the thematic condition were indeed more likely to identify target words in referentially ambiguous scenes (M = .29; SD = .15) than participants in the control condition (M = .19; SD = .18), as well as the norming study (M = .20), smallest p =.012. However, when comparisons were made for individual scenes, the effect of thematic context was driven primarily by the less ambiguous scenes (see Figures 2B-C), suggesting limits to whether thematic context can overcome in-the-moment referential ambiguity. In a final exploratory analysis, participants’ errors were analyzed to investigate whether participants in the thematic context condition were more likely to produce errors semantically related to the target word (see Figure 2D). These error analyses revealed that this was not the case, suggesting that thematic context did not even help narrow the target word’s semantic neighborhood.
Discussion
Word learning theories differ in their depictions of the relationship between environmental context and learning. These results underscore why such differing views exist – they are both consistent with the empirical record (Custode et al., 2020; Trueswell et al., 2016). The current results demonstrate that although contextual cues may help reduce ambiguity to a certain degree and in certain situations, contextual cues also have their limits. Future research that furthers our understanding of the nature and limits of environmental context on learning will go a long way in unifying different theories of word learning.

Authors

• Georgia Grace Capobianco, University of Connecticut
  Presenting Author

• Sumarga H. Suanda, University of Connecticut