Using Interlinear Glosses as Pivot in Low-Resource Multilingual Machine Translation.

We demonstrate a new approach to Neural Machine Translation (NMT) for low-resource languages using a ubiquitous linguistic resource, Interlinear Glossed Text (IGT). IGT represents a non-English sentence as a sequence of English lemmas and morpheme labels. As such, it can serve as a pivot or interlingua for NMT. Our contribution is four-fold. Firstly, we pool IGT for 1,497 languages in ODIN (54,545 glosses) and 70,918 glosses in Arapaho and train a gloss-to-target NMT system from IGT to English, with a BLEU score of 25.94. We introduce a multilingual NMT model that tags all glossed text with gloss-source language tags and train a universal system with shared attention across 1,497 languages. Secondly, we use the IGT gloss-to-target translation as a key step in an English-Turkish MT system trained on only 865 lines from ODIN. Thirdly, we we present five metrics for evaluating extremely low-resource translation when BLEU is no longer sufficient and evaluate the Turkish low-resource system using BLEU and also using accuracy of matching nouns, verbs, agreement, tense, and spurious repetition, showing large improvements.

• Publication year

  2019

• Venue

  arXiv: Computation and Language

• Publication date

  2019-11-07

• Fields of study

  Linguistics, Computer Science

• Identifiers
  arXiv 1911.02709
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Revisiting Low-Resource Neural Machine Translation: A Case Study

  2019cited by this paper
• Morphological Word-Embeddings

  2019cited by this paper
• Character-Aware Decoder for Translation into Morphologically Rich Languages

  2018cited by this paper
• Automatic Glossing in a Low-Resource Setting for Language Documentation

  2018influential reference
• Improving Character-Based Decoding Using Target-Side Morphological Information for Neural Machine Translation

  2018cited by this paper
• Using Morphological Knowledge in Open-Vocabulary Neural Language Models

  2018cited by this paper
• Adapting Word Embeddings to New Languages with Morphological and Phonological Subword Representations

  2018cited by this paper
• Massively Parallel Cross-Lingual Learning in Low-Resource Target Language Translation

  2018influential reference
• Understanding and Improving Morphological Learning in the Neural Machine Translation Decoder

  2017cited by this paper
• Ensembling Factored Neural Machine Translation Models for Automatic Post-Editing and Quality Estimation

  2017cited by this paper
• OpenNMT: Open-Source Toolkit for Neural Machine Translation

  2017cited by this paper
• Word Representations in Factored Neural Machine Translation

  2017cited by this paper
• Six Challenges for Neural Machine Translation

  2017cited by this paper
• Google’s Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation

  2016influential reference
• Very-large Scale Parsing and Normalization of Wiktionary Morphological Paradigms

  2016cited by this paper
• Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation

  2016cited by this paper
• Toward Multilingual Neural Machine Translation with Universal Encoder and Decoder

  2016cited by this paper
• Polyglot Neural Language Models: A Case Study in Cross-Lingual Phonetic Representation Learning

  2016influential reference
• Multi-Way, Multilingual Neural Machine Translation with a Shared Attention Mechanism

  2016influential reference
• Factored Neural Machine Translation

  2016cited by this paper
• The Role of Context in Neural Morphological Disambiguation

  2016cited by this paper
• Multi-Source Neural Translation

  2016cited by this paper
• A Character-level Decoder without Explicit Segmentation for Neural Machine Translation

  2016cited by this paper
• Linguistic Input Features Improve Neural Machine Translation

  2016cited by this paper
• Applying Universal Dependency to the Arapaho Language

  2016cited by this paper
• Multi-Task Learning for Multiple Language Translation

  2015influential reference
• Neural Machine Translation of Rare Words with Subword Units

  2015cited by this paper
• Automatic interlinear glossing as two-level sequence classification

  2015influential reference
• Multilingual Language Processing From Bytes

  2015influential reference
• A Language-Independent Feature Schema for Inflectional Morphology

  2015cited by this paper
• Qun Liu.

  2015cited by this paper
• Enriching, Editing, and Representing Interlinear Glossed Text

  2015cited by this paper
• Character-based Neural Machine Translation

  2015cited by this paper
• Co-learning of Word Representations and Morpheme Representations

  2014cited by this paper
• Compositional Morphology for Word Representations and Language Modelling

  2014cited by this paper
• Enriching ODIN

  2014cited by this paper
• Better Word Representations with Recursive Neural Networks for Morphology

  2013cited by this paper
• A Simple, Fast, and Effective Reparameterization of IBM Model 2

  2013influential reference
• Polyglot: Distributed Word Representations for Multilingual NLP

  2013influential reference
• Character-Based Pivot Translation for Under-Resourced Languages and Domains

  2012cited by this paper
• Combining Morpheme-based Machine Translation with Post-processing Morpheme Prediction

  2011influential reference
• Syntax-to-Morphology Mapping in Factored Phrase-Based Statistical Machine Translation from English to Turkish

  2010cited by this paper
• Developing ODIN: A Multilingual Repository of Annotated Language Data for Hundreds of the World's Languages

  2010cited by this paper
• Morphological pre-processing for Turkish to English statistical machine translation

  2009cited by this paper
• Factored Translation Models

  2007cited by this paper
• Example-Based Machine Translation of the Basque Language

  2006cited by this paper
• Permission is granted to quote short excerpts and to reproduce figures and tables from this report, provided that the source of such material is fully acknowledged. Arabic Preprocessing Schemes for Statistical Machine Translation

  2006cited by this paper
• Factored Neural Language Models

  2006cited by this paper
• Initial Explorations in English to Turkish Statistical Machine Translation

  2006cited by this paper
• Improving Statistical MT through Morphological Analysis

  2005cited by this paper
• Morphological Analysis for Statistical Machine Translation

  2004cited by this paper
• Bleu: a Method for Automatic Evaluation of Machine Translation

  2002cited by this paper
• Machine Translation of Very Close Languages

  2000cited by this paper
• Adding linguistic knowledge to a lexical example-based translation system

  1999cited by this paper
• Two-level Description of Turkish Morphology

  1993cited by this paper
• Using Lexicalized Tags for Machine Translation

  1990cited by this paper
• Typology and Universals

  1990cited by this paper
• DIRECTIONS FOR INTERLINEAR MORPHEMIC TRANSLATIONS

  1982cited by this paper
• Learning

  1965cited by this paper

• Massively Multilingual Joint Segmentation and Glossing

  2026cites this paper
• Low-Resource Interlinear Translation: Morphology-Enhanced Neural Models for Ancient Greek

  2025cites this paper
• A Concept Based Approach for Translation of Medical Dialogues into Pictographs

  2024cites this paper
• GlossLM: A Massively Multilingual Corpus and Pretrained Model for Interlinear Glossed Text

  2024cites this paper
• Can we teach language models to gloss endangered languages?

  2024cites this paper
• Can LLMs Really Learn to Translate a Low-Resource Language from One Grammar Book?

  2024cites this paper
• GrammaMT: Improving Machine Translation with Grammar-Informed In-Context Learning

  2024cites this paper
• Crowdsourcing OLiA Annotation Models the Indirect Way

  2023cites this paper
• Evaluating Pre-training Objectives for Low-Resource Translation into Morphologically Rich Languages

  2022cites this paper
• A Neural Machine Translation Approach to Translate Text to Pictographs in a Medical Speech Translation System - The BabelDr Use Case

  2022cites this paper
• Defining meaningful units. Challenges in sign segmentation and segment-meaning mapping (short paper)

  2021cites this paper
• Integrating Automated Segmentation and Glossing into Documentary and Descriptive Linguistics

  2021cites this paper