LLS Colloquium: "Variable language, adaptable learners: evidence that language acquisition is robust to noise at two timescales"

On Tuesday 30th November 2021, Charlotte Moore (Concordia University) will be presenting on "Variable language, adaptable learners: evidence that language acquisition is robust to noise at two timescales". The talk focuses on how infants deal with wordform variability in early word learning.  

Variable language, adaptable learners: evidence that language acquisition is robust to noise at two timescales

When learning a language, typically-developing infants face the daunting task of learning both the sounds and the meanings of words. This process is complicated by the existence of wordform variability, where the same concept can be referred to with multiple wordforms (e.g. “dog” and “doggy”). Wordform variability creates a many-to-one relationship between words and their meanings. In this talk, we explore the consequences of wordform variability for infants. We begin by taking a longitudinal look at how much variability infants hear as they learn new words from their linguistic environments. I present findings from a longitudinal corpus of naturalistic infant-directed speech showing that children hear many wordforms for the same concept. Additionally, we examine the relationship between variability and learning by tying infants’ MCDI results to their corpus data. To conclude, we pivot to some experimental data. In this section, we investigate how well toddlers understand familiar words with low and high wordform variability in an eyetracking paradigm. Taken together, these findings suggest that wordform variability is prevalent in speech to infants, and that young learners are well-equipped to accommodate that variability, both when learning new words and when recognizing familiar ones.

The talk will take place at 4pm on Zoom, and there will be an opportunity to ask questions at the end - you can join using this link. 

Event poster: LLS Colloquium: Variable language, adaptable learners: evidence that language acquisition is robust to noise at two timescales