Poster #5 - Infants’ Perspective on Referential Uncertainty: Parents’ Object Labeling During Infant Visual Attention Moments

• In Event: 3-159 - Poster Session 12
  In Poster Session: PS 12 Section - Attention, Learning, Memory

Integrative Statement

An oft-cited difficulty infants are posited to face during early word learning is referential uncertainty: given the infinite number of possible referents for a heard word, how do infants discover correct object-label mappings (Quine, 1960)? Much research aims to uncover the mechanisms by which infants overcome this theorized challenge (e.g., Smith & Yu, 2008), however little empirical work has examined the underlying premise that the data for word learning in infants’ everyday environments is messy. An important source of word-learning data is parental language input during everyday interactions such as toy play. Infants actively shape their environment during these interactions, moving their eyes and bodies to explore objects. Studies of infants’ novel word learning suggest that infants create good opportunities for word learning, reducing or even eliminating referential uncertainty by holding and looking to objects, providing clear visual object information to associate with a heard object label (Pereira, Smith, Yu, 2014; Yu, Suanda, & Smith, in press). We examine the quality of infants’ word-learning data in naturalistic toy play contexts by asking: Do parents take advantage of moments of infant visual attention to objects by naming the attended object?

Method and Results. 242 infants aged 9-27 months and their parents played with three novel objects at a time for 6 minutes (Figure 1A). Parents were taught the names of each object and infants wore head-mounted eye-trackers from which looks to objects were coded frame-by-frame. Each look was classified into one of 3 categories based on whether the parent named the attended object (a “Hit”), a different object (a “Mislead”), or no object (a “Miss”) during that look (Figure 1B). On average, infants generated 36.9 (SD=11.3) looks to an object per minute and parents named an object 14.9 (SD=11.3) times per minute. Moreover, when naming occurred, it was during an infant look to the named object in 49% (SD=13%) of naming events (Figure 2A). Moreover, parents missed many opportunities for word learning by naming the attended object during only 18% (SD=9%) of object looks (Figure 2B), and provided misleading information to infants by naming a different object than the one being attended in 13% (SD=9%) of object looks.

Discussion. Although infants created good moments for word learning, parents did not take full advantage of these opportunities, often failing to provide the attended object’s name and sometimes providing misleading information by naming a non-attended object. Parents’ relatively low “Hit” rate is even more striking given that only three objects were played with at a time during this study; everyday word-learning environments likely involve many more potential referents. These findings provide a first step toward understanding infants’ word-learning data in everyday contexts. Future analyses will examine the looking and object manipulation behaviors infants engage in during the several seconds prior to “Hits”, to characterize the cues parents may use to determine the object of infants’ attention. These data will inform theories of infant word learning as well as provide actionable information for promoting parents’ ability to take advantage of opportunities for infant word-learning.

Authors

• Lauren K Slone, Indiana University
  Presenting Author

• Linda B Smith, Indiana University
  Non-Presenting Author

• Chen Yu, Indiana University
  Non-Presenting Author