# Frequency vs. Association for Constraint Selection in Usage-Based Construction Grammar

## TL;DR

This paper compares frequency-based and association-based models for constraint generalization in Usage-Based Construction Grammar across eight languages, finding that association-based models outperform frequency-based ones.

## Contribution

It provides the first large-scale cross-linguistic evaluation of models for constraint generalization in Construction Grammar, demonstrating the superiority of association-based models.

## Key findings

- Association-based models outperform frequency-based models in all tested languages.
- Association models provide more accurate generalizations.
- The study uses a novel metric based on the Minimum Description Length paradigm.

## Abstract

A usage-based Construction Grammar (CxG) posits that slot-constraints generalize from common exemplar constructions. But what is the best model of constraint generalization? This paper evaluates competing frequency-based and association-based models across eight languages using a metric derived from the Minimum Description Length paradigm. The experiments show that association-based models produce better generalizations across all languages by a significant margin.