A pedagogical platform for spoken post-editing (PE): the integration of speech input into COPECO
Jeevanthi Liyana Pathirana, Pierrette Bouillon, Jonathan Mutal · 2023
We present the integration of speech input into COPECO, an online teaching and learning platform developed to collect learner translation/postediting corpora [22].Speech offers potential to increase translators' productivity and wellbeing by reducing typing time and effort.A multimodal, speechenabled COPECO will complement other researchers' and developers' efforts to integrate speech dictation and post-editing into translation tools, training and practice (e.g.MateCat allows post-editors to activate a speech-to-text component or TradDICT Learn [28], an online-learning platform to develop sighttranslation and dictation skills).COPECO helps trainers to compare different translation modalities, prepare courses and share teaching resources.To our knowledge, this is the first online platform to allow comparison between written and spoken post-editing and gathering data/statistics on speech-based postediting behaviour (e.g respeaking whole segments vs. correcting minor errors, etc).This helps translation trainers understand frequent errors made by learners when using speech which can then be used to improve their course content.It will also help to build a "speech post-editing corpus" which will be beneficial in the long run for research and analytical purposes.Unlike proprietary speech recognition add-ons with defined services, we develop speech commands for post-editing based on translators' behaviour and needs, optimising COPECO based on requirements with minimal costs.