Sentiment classification transfers to English under the unchanged protocol, with a frozen multilingual encoder the strongest arm at wF1 0.73 and LLMs matching it only with few-shot prompting¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? On the balanced English test split, the e5 encoder reaches weighted F1 0.73; LLMs reach parity only with few-shot prompting (0.72-0.73) while zero-shot sits near 0.67. → Esteban U. Vega Barajas 2026

Evidence¶

Esteban U. Vega Barajas 2026¶

Esteban U. Vega Barajas. (2026). A Durability and Cross-Language Transfer Benchmark for a Validated Teaching-Feedback Classification Protocol. https://arxiv.org/abs/2607.11873 doi:10.48550/arxiv.2607.11873

q2 · i? · design · r2

English sentiment transfer experiment on the balanced 45,000-comment corpus; classical and encoder arms use the full 6,750-comment test split while LLM arms use a 1,500-comment cost-capped subset, so cross-rule values are not directly comparable. The article reports the encoder "strongest arm, at wF1 0.73".

"On the full test split the modern encoder is the strongest arm, at wF1 0.73. Both LLMs match it only with few-shot prompting: zero-shot sits near 0.67 for Haiku and Opus alike, and few-shot reaches 0.72 to 0.73."

Discussion¶

Related Claims¶

• The frozen BETO-2019 design and the modern e5-base encoder are numerically close on Spanish, with no descriptive separation on either task — related
• Under the fixed protocol, hosted LLMs produce the largest gain on Spanish thematic classification, with thematic macro F1 rising from 0.46 (TF-IDF) to roughly 0.82-0.86 for the two hosted LLMs — related
• On Spanish sentiment, the 2026 frontier model shows no advantage over the cheap model, with Haiku zero-shot at 0.923 versus Opus zero-shot at 0.884 and the head-to-head McNemar favoring Haiku — related
• The star-to-sentiment labeling rule agrees with human review-level labels at Cohen's kappa 0.66 with 82% accuracy on 3,000 RateMyProfessor teacher reviews