MMTAfrica: Multilingual Machine Translation for African Languages

Proceedings of the Sixth Conference on Machine Translation (WMT), pages 398–411 November 10–11, 2021. ©2021 Association for Computational Linguistics 398 MMTAfrica: Multilingual Machine Translation for African Languages Chris C. Emezue∗ Technical University of Munich Mila Quebec AI Institute† Masakhane Bonaventure F. P. Dossou∗ Jacobs University Bremen Mila Quebec AI Institute† Masakhane Abstract In this paper, we focus on the task of mul- tilingual machine translation for African lan- guages and describe our contribution in the 2021 WMT Shared Task: Large-Scale Mul- tilingual Machine Translation. We introduce MMTAfrica, the first many-to-many multilin- gual translation system for six African lan- guages: Fon (fon), Igbo (ibo), Kinyarwanda (kin), Swahili/Kiswahili (swa), Xhosa (xho), and Yoruba (yor) and two non-African lan- guages: English (eng) and French (fra). For multilingual translation concerning African languages, we introduce a novel backtransla- tion and reconstruction objective, BT&REC, inspired by the random online back trans- lation and T5 modelling framework respec- tively, to effectively leverage monolingual data. Additionally, we report improvements from MMTAfrica over the FLORES 101 benchmarks (spBLEU gains ranging from +0.58 in Swahili to French to +19.46 in French to Xhosa). In this paper, we make use of the following nota- tions: • ⊛ refers to any language in the set {eng, fra, ibo, fon, swa, kin, xho, yor}. • ⋄ refers to any language in the set {eng, fra, ibo, fon}. • AL(s) refers to African language(s). • X−→ Y refers to neural machine translation from language X to language Y. 1 Introduction Despite the progress of multilingual machine trans- lation (MMT) and the many efforts towards improv- ing its performance for low-resource languages, ∗_Authors contributed equally to this work. Cor- respondence to chris.emezue@gmail.com or femipan- crace.dossou@gmail.com †_Independent Research done while interning at Mila Que- bec AI Institute African languages suffer from under-representation. For example, of the 2000 known African lan- guages (Eberhard et al., 2020) only 17 of them are available in the FLORES 101 Large-scale Multilingual Translation Task as at the time of this research. Furthermore, most research that look into transfer learning of multlilingual mod- els from high-resource to low-resource languages rarely work with ALs in the low-resource sce- nario. While the consensus is that the outcome of the research made using the low-resource non- African languages should be scalable to African languages, this cross-lingual generalization is not guaranteed(Orife et al., 2020) and the extent to which it actually works remains largely under- studied. Transfer learning from African languages to African languages sharing the same language sub-class has been shown to give better translation quality than from high-resource Anglo-centric lan- guages (Nyoni and Bassett, 2021) calling for the need to investigate AL←→AL multilingual trans- lation. This low representation of African languages goes beyond machine translation (Martinus and Abbott, 2019; Joshi et al., 2020; ∀ et al., 2020). The analysis conducted by ∀ et al. (2020) revealed that low-resourcedness of African languages can be traced to the poor incorporation of African lan- guages in the NLP research community (Joshi et al., 2020). All these call …

Content truncated to 500 words