The potential and limits of lay post-editing in an online community

The potential and limits of lay post-editing in an online community Linda Mitchell CNGL/School of Applied Language and Intercultural Studies Dublin City University Dublin, Ireland linda.mitchell7@mail.dcu.ie Abstract This paper aims at exploring the potential of a lay community as post-editors. It fo- cusses on 15 members of an online tech- nology support forum, native speakers of the target language (TL) and some knowl- edge of the source language (SL) trans- lating content that was machine translated from English into German specific to their own domain. It presents the most predom- inant errors remaining in the post-edited output and the impact of these on the qual- ity of the post-edited output as measured by domain specialists evaluating adequacy and fluency. This paper further explores examples of these errors and possible so- lutions to reducing the occurrence of these and maximising the community’s poten- tial. The targeted post-editing quality was “‘good enough”, as determined in the post- editing guidelines. The PE results demon- strate that there is still room for improve- ment in terms of quality. 1 Introduction User-Generated Content (UGC) is constantly growing online. With that growth, the demand for translations with fast turnaround times increases, too. Common solutions to meet this demand are MT or a combination of MT and Post-Editing (PE), the correction of automatically translated text. Previous research in the field of PE has pre- dominantly focussed on professional translators, e.g. de Almeida (2013) who investigates corre- c⃝ 2015 The authors. This article is licensed under a Creative Commons 3.0 licence, no derivative works, attribution, CC- BY-ND. lations between translation experience and post- editing quality, Plitt and Masselot (2010) who compare results of traditional Human Translation to post-editing in an industrial setting using the ex- ample of Autodesk, or Guerberof (2009) who com- pares productivity between translation with Trans- lation Memories (TM) to post-editing of MT con- tent. Participants within these experiments have been found to experience adverse feelings towards post-editing (e.g. de Almeida 2013), which is not correlated to their often excellent performance. There have further been studies with translation students as post-editors, e.g. Koponen (2013), who investigates variation in post-editing prefer- ences or Depraetere (2010), who seeks to establish strategies in post-editing behaviour of translation students. Recently, there have been studies inves- tigating individual lay people or communities as post-editors, such as with subjects who have some knowledge of both the SL and the TL but who are untrained in translation (Aranberri et al. 2014) or domain specialists who are untrained in translation and have no knowledge of the SL (Schwartz 2014). While the studies presented above have mostly been of hypothetical nature here, we are tapping into a new pool of potential post-editors by fo- cussing on an already existent and real online com- munity. This community is a technology support forum, the Norton1 Community, a platform facil- itating discussion on the Norton products among the users of the products in order to solve problems they may be experiencing with additional guidance from Symantec employees. …

Content truncated to 500 words