# Broken Words, Broken Performance: Effect of Tokenization on Performance of LLMs

arXiv: https://arxiv.org/abs/2512.21933 · HTML: https://mlsanity.com/paper/2512.21933/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 9 of its 25 references that are in the map.
- First posted on arXiv 26 Dec 2025.

## Builds on (most cited first, 9 of 9)

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding](https://mlsanity.com/paper/1804.07461/) (2018)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [Neural Machine Translation of Rare Words with Subword Units](https://mlsanity.com/paper/1508.07909/) (2015)
- [BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions](https://mlsanity.com/paper/1905.10044/) (2019)
- [Structured Prediction as Translation between Augmented Natural Languages](https://mlsanity.com/paper/2101.05779/) (2021)
- [Tokenizer Choice For LLM Training: Negligible or Crucial?](https://mlsanity.com/paper/2310.08754/) (2023)
- [BPE Gets Picky: Efficient Vocabulary Refinement During Tokenizer Training](https://mlsanity.com/paper/2409.04599/) (2024)
- [Scaffold-BPE: Enhancing Byte Pair Encoding for Large Language Models with Simple and Effective Scaffold Token Removal](https://mlsanity.com/paper/2404.17808/) (2024)

## Built on by (most cited first, 2 of 2)

- [TokEval: A Tokenizer Evaluation Suite](https://mlsanity.com/paper/2608.18062/) (2026)
- [The Text Uncanny Valley: Non-Monotonic Performance Degradation in LLM Information Retrieval](https://mlsanity.com/paper/2605.07186/) (2026)

## Abstract (arXiv)

Tokenization is the first step in training any Large Language Model (LLM), where the text is split into a sequence of tokens as per the model's fixed vocabulary. This tokenization in LLMs is different from the traditional tokenization in NLP where the text is split into a sequence of "natural" words. In LLMs, a natural word may also be broken into multiple tokens due to limited vocabulary size of the LLMs (e.g., Mistral's tokenizer splits "martial" into "mart" and "ial"). In this paper, we hypothesize that such breaking of natural words negatively impacts LLM performance on various NLP tasks. To quantify this effect, we propose a set of penalty functions that compute a tokenization penalty for a given text for a specific LLM, indicating how "bad" the tokenization is. We establish statistical significance of our hypothesis on multiple NLP tasks for a set of different LLMs.