R0021/2026-03-25/Q008/SRC02

Research	R0021 — Prompt engineering definitions
Run	2026-03-25
Query	Q008
Search	S01
Result	S01-R02
Source	SRC02

Computational Linguistics — Polysemy Research

Source

Field	Value
Title	Polysemy — Evidence from Linguistics, Behavioral Science, and Contextualized Language Models
Publisher	MIT Press (Computational Linguistics)
Author(s)	Various
Date	2024
URL	https://direct.mit.edu/coli/article/50/1/351/118497/
Type	Peer-reviewed academic paper

Summary

Dimension	Rating
Reliability	High
Relevance	High
Bias: Missing data	Low risk
Bias: Measurement	Low risk
Bias: Selective reporting	Low risk
Bias: Randomization	N/A — not an RCT
Bias: Protocol deviation	N/A — not an RCT
Bias: COI/Funding	Low risk

Rationale

Dimension	Rationale
Reliability	Peer-reviewed MIT Press journal.
Relevance	Directly addresses the nature and prevalence of polysemy.
Bias flags	None.

Evidence Extracts

Evidence ID	Summary
SRC02-E01	Polysemy is pervasive; most content words are polysemous; more frequent words more polysemous