Children’s acquisition of grammatical variations: evidence from the artificial language learning paradigm
Yiran Chen, Postdoc, UW-Madison Dept of Psychology
There is abundant and systematic variation in language. For example, speakers sometimes pronounce words ending in -ing with -in’ (e.g., pronouns working as workin’) depending on linguistic (e.g.,more -in’ in progressives like ‘working’ than in nouns like ‘morning’) and social (e.g., more -in’ in a casual context than in formal one) factors. To become a native speaker, children need to learn these variations – their distributional patterns and social meaning, as it is an integral part of linguistic knowledge. Emergent evidence suggests that children do start to match these probabilistic patterns in their production of variation early on.
However, many questions on children’s acquisition of grammatical variation remain underexplored, partly because it is hard to tease apart effects of different complicated and intertwining factors in a child’s social and linguistic environment via naturalistic observation. In my talk, I will present my current research that engages with these questions with an artificial language learning paradigm. The first part of the talk focuses on the distributional learning aspect and asks what factors modulate whether learners match or regularize variation in their input. While children do reproduce linguistic variation early on, they are also known to change their language input and make it more regular when faced with probabilistic alternations – a process often referred to as regularization, and is robustly attested in creolization, emergent sign language formation, and learning of lab-created languages. In particular, I will demonstrate how variant similarity as well as shared variability across speakers modulate adult and child learners’ tendency to regularize.
The second part of the talk will deal with how children learn to map linguistic variations onto social dimensions. As adults, we adeptly draw inferences about our conversation partners’ identities, attitudes and intentions based on their speech and use how we speak to signal ours. However, it has been difficult to discern how and when children form these mappings. Leveraging a novel social contrast as well as a mini-artificial language, I will present preliminary data showing that 4-year-olds use speakers’ plural marking patterns to categorize them into social groups. I will also propose future studies that investigate whether social alignment with different speakers changes how children approach to learn variation from them.