A mixed methods investigation on perceived quality in crowdsourced localisation

With thousands of language communities worldwide and ever growing access to technology and internet communication, the demand for translation of content has been growing steadily. Traditional forms of language translation cannot cope with the future needs and demands of humanity. One of the solutions described and practiced has been localisation crowdsourcing. However, one of the concerns that have been expressed is that crowdsourced localisation might not provide sufficiently high quality. One perspective on quality is that of perceived quality, which can be influenced by extrinsic cues rather than the actual, objectively observable attributes of a product. This research asks the following questions: 1. How is quality in localisation defined and to what extent do different groups agree on which aspects of quality are important? 2. What are prevalent attitudes, 'naive theories' and opinions on crowdsourcing and quality within the localisation field? 3. What is the impact of 'naive theories' on perceived quality in localisation? This research uses a mixed methods approach that includes a survey as well as a set of experiments. It targets a range of groups in localisation, from translators and buyers to students and lecturers of translation studies. As a special focus, it includes volunteers and organisations that collaborate with the non-profit organisation The Rosetta Foundation1 (TRF) to provide voluntary, crowdsourced localisation for not-for-profit projects via the platform Trommons2. Based on the data collected and analysed, this research concluded that accurate representation of meaning was of primary concern for all groups. In addition, specific aspects were raised for each group that were congruent with their main interests and perspectives. Overall, attitudes towards crowdsourced localisation were positive. However, 1 https://www.therosettafoundation.org/ 2 https://trommons.org/ crowdsourcing also introduced quality uncertainty, which was especially prevalent in respondents with lower levels of training, experience and specialisation.