NusaWrites: Constructing High-Quality Corpora for Underrepresented and Extremely Low-Resource Languages
This work addresses the lack of high-quality, culturally relevant datasets for underrepresented and extremely low-resource languages, which is crucial for democratizing NLP technology, though it is incremental as it builds on existing corpus construction methods.
The study tackled the problem of low-quality corpora for underrepresented and low-resource languages by comparing methods like online scraping and native speaker writing, finding that paragraph writing by native speakers produced datasets with superior lexical diversity and cultural relevance. It introduced the NusaWrites benchmark covering 12 languages and concluded that existing multilingual models need extension to more underrepresented languages.
Democratizing access to natural language processing (NLP) technology is crucial, especially for underrepresented and extremely low-resource languages. Previous research has focused on developing labeled and unlabeled corpora for these languages through online scraping and document translation. While these methods have proven effective and cost-efficient, we have identified limitations in the resulting corpora, including a lack of lexical diversity and cultural relevance to local communities. To address this gap, we conduct a case study on Indonesian local languages. We compare the effectiveness of online scraping, human translation, and paragraph writing by native speakers in constructing datasets. Our findings demonstrate that datasets generated through paragraph writing by native speakers exhibit superior quality in terms of lexical diversity and cultural content. In addition, we present the \datasetname{} benchmark, encompassing 12 underrepresented and extremely low-resource languages spoken by millions of individuals in Indonesia. Our empirical experiment results using existing multilingual large language models conclude the need to extend these models to more underrepresented languages. We release the NusaWrites dataset at https://github.com/IndoNLP/nusa-writes.