Preschool Aged Children Learning Novel Words From Storybooks in the Presence of Background Noise

• In Event: Poster Session 09
  In Poster Session: PS 09 Topic - Language, Communication

Abstract

The current studies examine how different types of background noise interfere with word learning from a picture book. Noise can theoretically impact word learning in multiple ways: it reduces the quality of the input signal (reducing children’s information about the word form), but it can also be distracting, interfering with their ability to process and retain information. We compare different types of noises to help us distinguish these different impacts on word learning.

In the first study, 18 children aged 3.5-5.5 years participated. Children followed along with a storybook while listening to a prerecorded female speaker reading the story. Two nonwords were taught in each story, with a total of four stories. After each story, children were asked three questions about the words they learned in the story. These words were taught implicitly, similar to typical storybooks. There were a total of eight consonant-vowel-consonant (CVC) nonwords and two types of maskers presented at a +10 SNR: single talker noise and broadband noise. Word learning was tested by selecting the picture that matched the referent in a 3AFC and producing the referent. We found a main effect of noise in the first measure and an effect of age in the production measure. Participants had higher accuracy in the broadband condition than in the single talker condition. The production measure did not show any significant difference between noise conditions, but older children were more likely to produce the correct referent than younger children. These findings are consistent with previous word recognition work suggesting that children (unlike adults) show poorer performance with a single talker than with broadband noise. By the time of the annual meeting, we anticipate having completed two additional studies run online as a result of COVID19. Using the aforementioned method, we will first test children with single-talker noise and speech-shaped noise that is modulated by the temporal envelope of the single talker. This noise has the temporal envelope fluctuations of speech without the spectral properties of speech. Thus comparing this with the single talker allows us to examine the role of spectral detail. We hypothesize that children will not show a difference between these two forms of noise; this would suggest that the temporal envelope fluctuations present in a single talker’s voice are a major reason why this type of noise causes more distraction.

The third study will again have two maskers: the same single-talker noise and a simple harmonic complex noise. This masker has the same spectral properties of speech but without the temporal envelope fluctuations. We hypothesize that there will be a main effect of noise: that children will learn better in the harmonic complex noise because it does not contain the amplitude variations of the single talker masker. This would again suggest that amplitude modulations are a major contributor to the distractibility children experience with a single talker. This work is important to tease apart the components of speech to understand which features (temporal envelope, spectral detail) may interfere more with word learning success.

Authors

• Madison Buntrock, University of Maryland - College Park
  Presenting Author

• Emily Shroads, University of Maryland - College Park
  Non-Presenting Author

• Rochelle Newman, University of Maryland - College Park
  Non-Presenting Author