An Exploration of Typically Developing and Late-Talker Vocabulary Trajectories: Comparing Multiple Approaches

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

Abstract

Purpose: Longitudinal datasets have been invaluable in understanding the underlying mechanisms of how typical development proceeds and where it diverges (e.g. Thomas et al, 2009; Adolph et al, 2008). However, interpreting developmental trajectories can be challenging. In this work we look at individual trajectories of vocabulary growth in typically-developing toddlers and late talkers to illustrate the importance of three factors when analyzing longitudinal data: grouping participants into different subpopulations, using different estimation curves, and dealing with ceiling effects.
Method: We examined vocabulary growth in 84 monolingual 17.67-month-olds (SD=0.96) and continuing, monthly, for 12 months using theMacArthur-Bates Communicative Development Inventory (CDI) (Fenson, 2007). We grouped participants in two ways: 1) typically developing vs late talkers (<20th % at Visit 1; Figure 1A, 1B) and 2) further subdividing late talkers into persistent (< 20th% at Visit 12) and late bloomers (> 20th% at Visit 12; as in Fernald & Marchman, 2012; Figure 1C, 1D). Table 1 depicts group composition at the first and last visits. Using hierarchical models, we examine both linear and quadratic growth trends to understand differences between these groups. Further, to address typically developing children reaching ceiling, we do these analyses using both the number of words produced on the CDI (Figure 1A, 1C) and using an estimated vocabulary size as proposed by Mayor and Plunkett (2011) (Figure 1B,1D).
Results: Figure 1A shows that linearly, late talkers and their typically talking peers do not differ. Quadratically, late talkers show convex trajectories, or growth suggesting catch-up toward the latter half of the study, while typical talkers show concave growth, suggesting ceiling effects. We then use estimated vocabulary size to combat these ceiling effects. Figure 1B shows that without ceiling effects, typically developing children show steep linear growth, but late talkers still show convex trajectories, again suggesting many begin to catch-up to their typical peers.
Next, we again look at CDI vocabulary size, but now further delineate our late talking group into persistent and late bloomers. Figure 1C shows that both late talking groups individually show convex trajectories, though late bloomers show more growth than persistent late talkers. A clearer picture comes from using estimated vocabulary size as in Figure 1D, where only late bloomers show a convex trajectory, with persistent late talkers gaining vocabulary at a far slower rate than their typical and late blooming peers. These results suggest there is a late talking group of children that are on a slower linear trajectory, falling farther and farther behind their typical and late-blooming peers, suggesting they would be candidates for early intervention.
Conclusion: Three main conclusions were reached: 1) Using estimates of true vocabulary size based off the CDI could combat ceiling effects sometimes seen in longitudinal work; 2) Comparing late bloomers to those persistent late talkers may help contribute to identification and intervention strategies; 3) Identifying and understanding the underlying mechanisms contributing to typical versus delayed language acquisition will be best understood through regular, longitudinal assessment. Implications for use in future research and intervention are discussed.

Authors

• Jennifer Weber, University of Colorado Boulder
  Non-Presenting Author

• Eliana Colunga, University of Colorado Boulder
  Presenting Author