2016/12/14 by Rico Sennrich, Sennrich, Rico · 4 citations
Computer Science · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Natural Language Processing Techniques #Text Readability and Simplification #Topic Modeling #cs.CL
paper · pdf · doi:10.48550/arxiv.1612.04629
accepted at EACL 2017 (v3: minor fix to table 6 description)
openalex publication_date 2016/12/14 · arxiv created 2017/02/13 · arxiv updated 2017/02/14 · openalex created_date 2022/10/03 · openalex updated_date 2026/07/28
Analysing translation quality in regards to specific linguistic phenomena has historically been difficult and time-consuming. Neural machine translation has the attractive property that it can produce scores for arbitrary translations, and we propose a novel method to assess how well NMT systems model specific linguistic phenomena such as agreement over long distances, the production of novel words, and the faithful translation of polarity. The core idea is that we measure whether a reference translation is more probable under a NMT model than a contrastive translation which introduces a specific type of error. We present LingEval97, a large-scale data set of 97000 contrastive translation pairs based on the WMT English->German translation task, with errors automatically created with simple rules. We report results for a number of systems, and find that recently introduced character-level NMT systems perform better at transliteration than models with byte-pair encoding (BPE) segmentation, but perform more poorly at morphosyntactic agreement, and translating discontiguous units of meaning.