BHARATI: Morphology-Aware Tokenizers for Classical Indian Languages with Subword Fertility Analysis

• URL: http://arxiv.org/abs/2607.23319v1
• Date: Sat, 25 Jul 2026 18:23:06 GMT
• Title: BHARATI: Morphology-Aware Tokenizers for Classical Indian Languages with Subword Fertility Analysis
• Abstract summary: This paper presents BHARATI, a set of SentencePiece BPE token trained on a balanced 781 MB corpus spanning seven languages.<n>We describe three successive tokenizer versions: v1 (English and Sanskrit only), v2 (four-language support with byte-fallback for Tamil), and v3 (full seven-language native subword coverage)<n>Subword fertility analysis demonstrates that v3 averages 2.6 tokens per Indian Knowledge System (IKS) technical term, compared to 5.25 tokens per term with GPT-2's tokenizer and 3.75 tokens with the multilingual SentencePiece baseline.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Standard subword tokenization algorithms such as Byte-Pair Encoding (BPE) and SentencePiece are trained predominantly on modern language corpora and produce inefficient segmentations when applied to classical Indian languages. Sanskrit, Tamil, and other classical Indic languages exhibit agglutinative morphology, productive sandhi (phonological fusion at word boundaries), and domain-specific vocabularies absent from general-purpose training data. This paper presents BHARATI, a set of SentencePiece BPE tokenizers trained on a balanced 781 MB corpus spanning seven languages (English, Hindi, Sanskrit, Tamil, Telugu, Kannada, and Malayalam) with native script support for all languages. We describe three successive tokenizer versions: v1 (English and Sanskrit only, with broken byte-fallback for Tamil), v2 (four-language support with byte-level fallback for southern languages), and v3 (full seven-language native subword coverage). Subword fertility analysis demonstrates that v3 averages 2.6 tokens per Indian Knowledge System (IKS) technical term, compared to 5.25 tokens per term with GPT-2's tokenizer and 3.75 tokens with the multilingual SentencePiece baseline, with the largest gains on a set of reserved IKS terms that are represented as single tokens by construction. On a held-out test set of 490 IKS-domain sentences (70 per language across seven languages, released with the measurement script), v3 reduces sequence length by roughly 90% relative to GPT-2 and byte-level encoding (which lack native Indic subwords) and by approximately 25% relative to the mBART-50 multilingual baseline, averaged across the six Indic languages, directly translating to increased effective context length for downstream language models. The tokenizer models (32,000 vocabulary), training scripts, and evaluation benchmarks are released under open licenses.

Related papers

• Separate Before You Compress: The WWHO Tokenization Architecture [0.0]
  Current Large Language Models (LLMs) mostly use BPE (Byte Pair Linguist) based tokenizers.<n>We propose a new three-layer architecture, WWHO (Where-What-How), and an algorithm named SGPE (Syllable Grapheme Pair Linguist)<n>Using Sinhala and Devanagari (Hindi/Sanskrit) as highly complex Abugida scripts, we trained WWHO on a cleaned 30-million-sentence dataset and evaluated on a 1,499,950-sentence test set.
  arXiv  Detail & Related papers  (2026-03-26T10:56:48Z)
• Omnilingual SONAR: Cross-Lingual and Cross-Modal Sentence Embeddings Bridging Massively Multilingual Text and Speech [61.759910921200834]
  Cross-lingual sentence encoders typically cover only a few hundred languages.<n>We introduce OmniSONAR, a new family of omnilingual, cross-lingual and cross-language sentence embedding models.
  arXiv  Detail & Related papers  (2026-03-17T14:47:35Z)
• VerChol -- Grammar-First Tokenization for Agglutinative Languages [0.0]
  Tokenization is the foundational step in all large language model (LLM) pipelines.<n>For agglutinative languages, a single word may encode root, tense, aspect, person, number, gender agreement, case, and postpositions into one orthographic unit.<n>Statistic tokenizers fragment these words into byte pair chunks that sever morpheme boundaries and token counts.
  arXiv  Detail & Related papers  (2026-03-06T04:07:15Z)
• Evaluating Subword Tokenization Techniques for Bengali: A Benchmark Study with BengaliBPE [0.0]
  BengaliBPE is a language-aware subword tokenizer for the Bengali script.<n>It applies Unicode normalization and morphology-aware merge rules to maintain linguistic consistency and preserve subword integrity.<n>It provides the most detailed segmentation and the best morphological interpretability, albeit with slightly higher computational cost.
  arXiv  Detail & Related papers  (2025-11-07T15:23:32Z)
• Tokenization Matters: Improving Zero-Shot NER for Indic Languages [2.964265227875254]
  Tokenization is a critical component of Natural Language Processing (NLP)<n>This work systematically compares BPE, SentencePiece, and Character Level tokenization strategies using Indic languages.<n>Results show that SentencePiece is a consistently better performing approach than BPE for NER in low resource Indic languages.
  arXiv  Detail & Related papers  (2025-04-23T17:28:38Z)
• SuperBPE: Space Travel for Language Models [103.09169510391972]
  We introduce a "superword" tokenizer, SuperBPE, which incorporates a simple pretokenization curriculum into the byte-pair encoding (BPE) algorithm.<n>SuperBPE encodes a fixed piece of text with up to 33% fewer tokens than BPE on average.<n>Our model trained with SuperBPE achieves an average +4.0% absolute improvement over the BPE baseline across 30 downstream tasks.
  arXiv  Detail & Related papers  (2025-03-17T17:53:23Z)
• CoSTA: Code-Switched Speech Translation using Aligned Speech-Text Interleaving [61.73180469072787]
  We focus on the problem of spoken translation (ST) of code-switched speech in Indian languages to English text. We present a new end-to-end model architecture COSTA that scaffolds on pretrained automatic speech recognition (ASR) and machine translation (MT) modules. COSTA significantly outperforms many competitive cascaded and end-to-end multimodal baselines by up to 3.5 BLEU points.
  arXiv  Detail & Related papers  (2024-06-16T16:10:51Z)
• Parrot: Multilingual Visual Instruction Tuning [66.65963606552839]
  Existing methods typically align vision encoders with Multimodal Large Language Models (MLLMs) via supervised fine-tuning (SFT)<n>We propose PARROT, a novel approach that leverages textual guidance for visual token alignment at the language level.<n>We introduce the Massive Multilingual Multimodal Benchmark (MMMB), a new benchmark comprising 6 languages, 15 categories, and 12,000 questions.
  arXiv  Detail & Related papers  (2024-06-04T17:56:28Z)
• MYTE: Morphology-Driven Byte Encoding for Better and Fairer Multilingual Language Modeling [70.34758460372629]
  We introduce a new paradigm that encodes the same information with segments of consistent size across diverse languages. MYTE produces shorter encodings for all 99 analyzed languages. This, in turn, improves multilingual LM performance and diminishes the perplexity gap throughout diverse languages.
  arXiv  Detail & Related papers  (2024-03-15T21:21:11Z)
• Machine Translation by Projecting Text into the Same Phonetic-Orthographic Space Using a Common Encoding [3.0422770070015295]
  We propose an approach based on common multilingual Latin-based encodings (WX notation) that take advantage of language similarity. We verify the proposed approach by demonstrating experiments on similar language pairs. We also get up to 1 BLEU points improvement on distant and zero-shot language pairs.
  arXiv  Detail & Related papers  (2023-05-21T06:46:33Z)
• CLSE: Corpus of Linguistically Significant Entities [58.29901964387952]
  We release a Corpus of Linguistically Significant Entities (CLSE) annotated by experts. CLSE covers 74 different semantic types to support various applications from airline ticketing to video games. We create a linguistically representative NLG evaluation benchmark in three languages: French, Marathi, and Russian.
  arXiv  Detail & Related papers  (2022-11-04T12:56:12Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.