RECORD
Prompting with Phonemes: Enhancing LLMs' Multilinguality for Non-Latin Script Languages
Computation and Language · Submitted 4 Nov 2024
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim Our study demonstrates that integrating phonemic signals improves performance across both non-Latin and Latin script languages, with a particularly significant impact on closing the performance gap between the two.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
Although multilingual LLMs have achieved remarkable performance across benchmarks, we find they continue to underperform on non-Latin script languages across contemporary LLM families. This discrepancy arises from the fact that LLMs are pretrained with orthographic scripts, which are dominated by Latin characters that obscure their shared phonology with non-Latin scripts. We propose leveraging phonemic transcriptions as complementary signals to induce script-invariant representations. Our study demonstrates that integrating phonemic signals improves performance across both non-Latin and Latin script languages, with a particularly significant impact on closing the performance gap between the two. Through detailed experiments, we show that phonemic and orthographic scripts retrieve distinct examples for in-context learning (ICL). This motivates our proposed Mixed-ICL retrieval strategy, where further aggregation from both leads to our significant performance improvements for both Latin script languages (up to 12.6%) and non-Latin script languages (up to 15.1%) compared to randomized ICL retrieval.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Although multilingual LLMs have achieved remarkable performance across benchmarks, we find they continue to underperform on non-Latin script languages across contemporary LLM families. | background 0.65 | 0.88 | 0.03 | asserted 1.00 |
| s1 | This discrepancy arises from the fact that LLMs are pretrained with orthographic scripts, which are dominated by Latin characters that obscure their shared phonology with non-Latin scripts. | interpretation 0.49 | 0.18 | 0.03 | asserted 1.00 |
| s2 | We propose leveraging phonemic transcriptions as complementary signals to induce script-invariant representations. | contribution 0.52 | 0.31 | 0.09 | asserted 0.58 |
| s3 | Our study demonstrates that integrating phonemic signals improves performance across both non-Latin and Latin script languages, with a particularly significant impact on closing the performance gap between the two. | result 1.00 | 0.95 | 0.76 | asserted 1.00 |
| s4 | Through detailed experiments, we show that phonemic and orthographic scripts retrieve distinct examples for in-context learning (ICL). | result 0.95 | 0.94 | 0.05 | asserted 1.00 |
| s5 | This motivates our proposed Mixed-ICL retrieval strategy, where further aggregation from both leads to our significant performance improvements for both Latin script languages (up to 12.6%) and non-Latin script languages (up to 15.1%) compared to randomized ICL retrieval. | result 0.93 | 0.95 | 0.43 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v3
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
Although multilingual LLMs have achieved remarkable performance across benchmarks, we find they continue to underperform on non-Latin script languages across contemporary LLM families. This discrepancy arises from the fact that LLMs are pretrained with orthographic scripts, which are dominated by Latin characters that obscure their shared phonology with non-Latin scripts. We propose leveraging phonemic transcriptions as complementary signals to induce script-invariant representations. Our study demonstrates that integrating phonemic signals improves performance across both non-Latin and Latin script languages, with a particularly significant impact on closing the performance gap between the two. Through detailed experiments, we show that phonemic and orthographic scripts retrieve distinct examples for in-context learning (ICL). This motivates our proposed Mixed-ICL retrieval strategy, where further aggregation from both leads to our significant performance improvements for both Latin script languages (up to 12.6%) and non-Latin script languages (up to 15.1%) compared to randomized ICL retrieval.
Human review
How the literature uses it
1 arXiv paper cite this in 2 sentences
Papers citing it · 1 sentence without a confident label
Cites it as background 1 paper
-
[ref] propose phonemic prompting as a way to enhance the multilingual capabilities of LLMs.
Phonemes to the Rescue: Multilingual Tokenization Based on International Phonetic Alphabet · Background and Related Work · 0.87
What it cites · 6 sentences without a confident label
Cites as background 2 papers
-
Non-phonemic approaches to bridging multilingual gaps via finetuning have involved modified training, synthetic data via translation, and contrastive presentations [ref] .
Enhancing Cross-lingual Transfer by Manifold Mixup · Background and Related Work · 0.94
-
The increase in model parameters and rise of instruction datasets have led to the emergent capability of LLMs to perform diverse tasks via few- to zero-shot demonstrations [ref] through in-context learning (ICL) during inference [ref] .
CG-BERT: Conditional Text Generation with BERT for Generalized Few-shot Intent Detection · Introduction · 0.87
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.