Children use simple cues before complex verb meanings to learn event boundaries
Modeling the Developmental Shift in Telicity Acquisition
Computation and Language
Summary
Learning how we talk about events that end or go on for a while, like eating one apple versus eating apples in general, is tricky for kids. The authors used a computer method to study how children and adults use clues in language to understand this. They found that kids rely mostly on simple clues like certain words after verbs, while adults use more complex information about the verbs themselves. This suggests that kids first learn language using easy cues and only later understand more detailed meanings.
What this means in practice
- •For natural language processing engineers: Improve AI language models to better distinguish event boundaries by incorporating child-like reliance on syntactic cues for telicity interpretation.
- •For speech therapy clinicians: Design assessment tools that recognize key syntactic cues children use to understand event completion, aiding diagnosis of language development delays.
Authors
Ellie Xia, Parisa Kordjamshidi, Alan Hezao Ke
Abstract
Acquiring telicity, which is the distinction between bounded (e.g., ate an apple) and unbounded (e.g., ate apples) events, requires first language (L1) learners to map surface-level and semantic cues to abstract event structures, but the computational trajectory of this mapping is not well understood. We introduce a Difference in Surprisal method that uses GPT2 token surprisal over paired temporal adverbial diagnostics (in an hour versus for an hour) to automatically label telicity across English CHILDES corpora, validated against expert linguist judgments. Using these labels, we train diagnostic logistic regression classifiers on 12 syntactic and lexical semantic features to compare how child speech and child-directed speech encode telicity. The two models diverge: the child model reaches near perfect accuracy through a single deterministic cue, the presence of a post-verbal determiner, while the adult model relies more heavily on verb class and other lexical semantic features, with the determiner cue neutralized. This trajectory supports Syntactic Bootstrapping: learners first exploit high-frequency structural cues as a scaffold to bootstrap, before developing fully compositional, verb-based event structures.