---
title: "Typological Feature Induction as a Benchmark for Inductive Metalinguistic Reasoning"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.04421v1"
permalink: "https://feed.astralor.com/items/itm_ce3d0f0dc5cefeb6e33ed460"
published: "2026-11-03T07:06:51.000Z"
score: 57
channel: "tech"
tags: ["LLM","推理效率","Typological Feature Induction","Benchmark","Inductive Metalinguistic Reasoning","Existing","TFI","Given"]
---

# Typological Feature Induction as a Benchmark for Inductive Metalinguistic Reasoning

## At a glance

Existing metalinguistic reasoning benchmarks require both inference of rules and their application, making failures hard to diagnose. We introduce typological feature induction (TFI) which targets the former step. Given…

## Key points

- Existing metalinguistic reasoning benchmarks require both inference of rules and their application, making failures hard to diagnose.
- We introduce typological feature induction (TFI) which targets the former step.
- Given sentences in an unnamed language paired with translations in a known pivot language, models must predict a typological feature of the unnamed language.

## Full text

Existing metalinguistic reasoning benchmarks require both inference of rules and their application, making failures hard to diagnose. We introduce typological feature induction (TFI) which targets the former step. Given sentences in an unnamed language paired with translations in a known pivot language, models must predict a typological feature of the unnamed language. From Grambank and multi-parallel corpora, we automatically build evaluation questions over 195 languages. We find that LLMs identify 59--80% of these languages from only 16 sentence pairs; and misaligning the translations barely affects their accuracy, suggesting reliance on prior knowledge of the language. Enciphering text to prevent recognition increases the corpus' utility, but mainly for features where target and pivot languages agree. On languages which models cannot identify, neither open model outperforms a majority-value baseline. Since induction from raw text aligns with field linguists' use-case, this work indicates the models tested are not yet ready for this role. TFI offers a scalable path to track progress in this direction.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_ce3d0f0dc5cefeb6e33ed460