Modeling the Developmental Shift in Telicity Acquisition
Summary
This paper introduces a method using GPT-2 token surprisal to label telicity in child language, finding that child models rely on syntactic cues like post-verbal determiners, while adult models use semantic features, supporting syntactic bootstrapping theory.
View Cached Full Text
Cached at: 09/17/26, 09:03 AM
# Modeling the Developmental Shift in Telicity Acquisition Source: [https://arxiv.org/abs/2609.17996](https://arxiv.org/abs/2609.17996) [View PDF](https://arxiv.org/pdf/2609.17996) > 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\. ## Submission history From: Ellie Xia \[[view email](https://arxiv.org/show-email/5806bb7e/2609.17996)\] **\[v1\]**Wed, 16 Sep 2026 01:34:11 UTC \(915 KB\)
Similar Articles
Developmental Trajectories of Situation Modeling and Mentalizing in Transformer Language Models
This paper investigates the emergence of situation modeling and mentalizing abilities in transformer language models across training stages, finding that false belief task performance depends on model size and training volume, emerges late in pretraining, and shows fragility with non-factive verbs.
Language Re-generation: An investigation into information locality effects on reconstruction
This paper investigates how GPT-2 models pre-trained on impossible languages (with disrupted information locality) can recover natural English, showing a bias toward shorter dependency lengths and dissociation between structural and surface recovery.
Causal Interventions Reveal Typologically Organized Syntactic Mechanisms in Multilingual Language Models
This paper uses causal interventions to investigate syntactic mechanisms in multilingual language models, revealing cross-lingual transfer that is graded based on typological similarity.
Collocational bootstrapping: A hypothesis about the learning of subject-verb agreement in humans and neural networks
This paper proposes collocational bootstrapping, a mechanism by which statistical word co-occurrence cues can aid the acquisition of English subject-verb agreement, supported by neural network simulations and analysis of child-directed speech.
Modeling Pathology-Like Behavioral Patterns in Language Models Through Behavioral Fine-Tuning
This paper introduces a behavioral induction framework that fine-tunes language models on structured decision-making tasks to induce stable, context-general shifts in generative distributions, modeling pathology-like behavioral patterns such as depression and paranoia.