1248-GRM How Everyday Language Describes Cognitive Abilities: A Folk Vocabulary Resource for the Cattell-Horn-Carroll Taxonomy

Presenter Information

Location

https://campus.kennesaw.edu/colleges-departments/ccse/events/computing-showcase/fa26-cday-program.php

Document Type

Event

Start Date

20-11-2026 2:00 PM

Description

Cognitive abilities are often assessed from everyday language, such as open-ended responses, teacher reports, and self-reports, yet the constructs that interpret that evidence are defined in technical terms, and no resource maps one onto the other. The Cattell-Horn-Carroll (CHC) model organizes cognitive abilities into about 90 narrow abilities under 18 broad abilities, but its labels rarely surface in naturalistic discourse. We ask how far everyday language reaches across the taxonomy, whether it preserves the taxonomy’s structure, and whether it agrees with how experts operationalize the same abilities. We build a folk corpus from Reddit using semantic retrieval, a classifier for ability descriptions, and LLM-based extraction, and assemble per-ability fingerprints covering lexical, conceptual, construal, and evaluative dimensions. Of the 71 non-tentative narrow abilities, 57 have a recoverable folk vocabulary; absence is concentrated among abilities whose individual differences appear only on laboratory tasks. Descriptor geometry recovers the taxonomy locally, though label-blind clustering recovers it only partially. Folk language fails to separate a few ability pairs, several of which CHC distinguishes by a property of what is produced or remembered rather than by the activity people describe. The fingerprints also converge on an independent expert test battery: an ability’s full folk vocabulary matches the battery’s tests to their target abilities better than its technical label or definition, and about as well as hand-written descriptions. The labeled corpus and per-ability fingerprints are released as an open-access resource on Zenodo.

This document is currently not available here.

Share

COinS
 
Nov 20th, 2:00 PM

1248-GRM How Everyday Language Describes Cognitive Abilities: A Folk Vocabulary Resource for the Cattell-Horn-Carroll Taxonomy

https://campus.kennesaw.edu/colleges-departments/ccse/events/computing-showcase/fa26-cday-program.php

Cognitive abilities are often assessed from everyday language, such as open-ended responses, teacher reports, and self-reports, yet the constructs that interpret that evidence are defined in technical terms, and no resource maps one onto the other. The Cattell-Horn-Carroll (CHC) model organizes cognitive abilities into about 90 narrow abilities under 18 broad abilities, but its labels rarely surface in naturalistic discourse. We ask how far everyday language reaches across the taxonomy, whether it preserves the taxonomy’s structure, and whether it agrees with how experts operationalize the same abilities. We build a folk corpus from Reddit using semantic retrieval, a classifier for ability descriptions, and LLM-based extraction, and assemble per-ability fingerprints covering lexical, conceptual, construal, and evaluative dimensions. Of the 71 non-tentative narrow abilities, 57 have a recoverable folk vocabulary; absence is concentrated among abilities whose individual differences appear only on laboratory tasks. Descriptor geometry recovers the taxonomy locally, though label-blind clustering recovers it only partially. Folk language fails to separate a few ability pairs, several of which CHC distinguishes by a property of what is produced or remembered rather than by the activity people describe. The fingerprints also converge on an independent expert test battery: an ability’s full folk vocabulary matches the battery’s tests to their target abilities better than its technical label or definition, and about as well as hand-written descriptions. The labeled corpus and per-ability fingerprints are released as an open-access resource on Zenodo.