English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Can Saint Augustine's Ostensive Definition Teach Language Models Words? Testing 4th-Century Philosophy on DeBERTa

Forum topic · ✨步子哥 · 2026-09-11

Summary

A BabyLM Workshop 2026 paper (arXiv:2609.11870, Lisa Bylinina) operationalizes Saint Augustine's ostensive definition theory by replacing the initial embeddings of concrete words in a DeBERTa masked language model with visual feature vectors derived from image region annotations, then training on a 10M-word BabyLM-scale corpus. Visual seeding leaves a persistent, detectable trace in the embedding space, but most standard BabyLM benchmarks (testing syntax such as subject-verb agreement and binding) show no difference. Only the COMPS benchmark, which tests object-property knowledge, reveals a consistent zero-shot advantage, limited strictly to seeded words. A synthetic seeding control confirms the effect is causal: artificially seeded words gain the same advantage. Notably, seeded function and abstract words show lower masked-prediction loss across all configurations, yet no existing benchmark detects this signal. The paper's core insight is an evaluation problem: language models may retain and exploit internal priors that standard benchmarks systematically miss.

Overview

Paper: arXiv: 2609.11870 | Lisa Bylinina | BabyLM Workshop 2026

In 397 CE, Saint Augustine described in his *Confessions* how he learned language: adults pointed at objects while naming them, and he associated sounds with things — the idea philosophers later called ostensive definition. This post discusses an experiment that directly implements that 1,600-year-old theory in a modern language model.

Experimental Design

  • Model: DeBERTa (masked language model), trained on a 10M-word BabyLM-scale corpus.
  • Before training, embeddings of a subset of concrete words (e.g., "dog", "red", "big") are replaced with visual feature vectors extracted from image region annotations.
  • Function words and abstract words keep random initialization.
  • Training proceeds normally; the "visually seeded" model is compared against a purely random-init baseline across benchmarks.
  • Results

  • The seed leaves a persistent imprint. After training, seeded words' embeddings retain detectable traces of visual features in the embedding space.
  • But most BabyLM benchmarks detect no difference. Abstract syntactic tasks (subject-verb agreement, island constraints, binding theory) show no benefit.
  • The exception: COMPS (Misra et al. 2023), testing object–property knowledge (e.g., "bananas are yellow"). Visual seeding gives a zero-shot advantage in every configuration.
  • A custom Visual-Property Swap benchmark (color, material, size, shape), controlling word frequency and seeding status, confirms: seeding yields a persistent, reproducible advantage only on seeded words.
  • Synthetic Seeding Transfer (Causal Validation)

    The author artificially gave synthetic visual embeddings to originally unseeded words. These "late-seeded" words gained the same advantage — showing the benefit comes from the seeding operation itself, not from any word being inherently easier to learn.

    The Ghost of Function Words

    A deeper finding: function and abstract words also ended up with visual seeds (indirectly) and retained them throughout training. Masked-prediction loss for seeded words was lower in every seed configuration — the training objective itself exploits the seeds.

    > The model "knows" these words are different. But no existing benchmark can detect it. As the paper concludes: *what kind of evaluation could capture this signal remains an open question.*

    Why It Matters

    The paper's value is not that visual seeding makes models stronger — the effect is modest and task-limited. Its insight is an evaluation problem:

    A model can retain visual priors internally, exploit them in its training objective, and still be indistinguishable on standard benchmarks from a randomly initialized model. Our evaluation methods may systematically miss what actually happens inside models.

    Comment

    The author of the post adds: many debates about "what models learn" are really debates about "how we measure learning." Model internals are richer than benchmark scores suggest — we need smarter probes, not just bigger models. Also notable: translating a 1,600-year-old philosophical theory directly into an experimental design shows philosophy can generate experiments, not just papers.

    Links:

  • Paper: https://arxiv.org/abs/2609.11870
  • Code: not yet released
  • Related benchmarks: BabyLM Challenge, COMPS, Visual-Property Swap

Tags

#language-models#ostensive-definition#deberta#babylm#grounding#evaluation-benchmarks#comps#philosophy-of-language

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178634735