Translation Revision as Part of Cognitive Translatorial Collaboration: Creativity under pressure

The objective of the research presented in this dissertation is to create an understanding of specialised (non-literary) translation revision as a situated activity that is carried out as part of a collaborative translation process and is dependent on various project-specific circumstances as well as the artefacts (e.g., software tools as well as material and digital objects) available to the reviser. The main theoretical framework is that of socially distributed cognition (SDC). SDC is a branch of situated cognitive theory, also called embodied cognition or 4EA (embodied, embedded, enacted, extended and affective) cognition. Methods and research approaches consistent with the theory have been employed to investigate how cognitive collaboration is established and performed in a system encompassing the translator and reviser, adding to our knowledge of the processes and practices of translation service production.

The four articles included in this dissertation begin with a survey of language service providers’ (LSPs’) revision operations, charting some aspects of the reviser’s task scope and questions of authority. The next step is an investigation of how various background factors contribute to the reviser’s role in collaborative translation production. In the third article, the translator’s and reviser’s joint process of co-creation and the communication channels used by the two participants are described, identifying the revision file as a digital artefact that acts as a communicative device, and spotlighting how cognitive collaboration is built in that artefact. The final article develops methodology for utilising the revision file as cognitive ethnographic data, subjecting revision files to artefact analysis and using systemic functional discourse analysis to investigate linguistic evidence of the distribution of labour.

The research methods employed in the studies that together make up this work were mainly qualitative, using diverse data that provided a rich picture of the phenomena under investigation. Some quantitative results were presented to support the qualitative findings, but as the data was limited and not construed to produce quantitative results, these do not allow generalisations to be drawn. Aiming at triangulation, the data included a survey, interviews with translation professionals, guided tour protocols of simulated revision situations, and authentic textual data. One or more of these datasets was used in each sub-study reported in the included articles.

The main findings presented here relate firstly, to the nature and characteristics of cognitive collaboration between translator and reviser, and secondly, to the processes and circumstances that influence this collaboration. The nature and characteristics of the collaboration can be summarised as flexibility, trust and complementarity. Variation in the production workflows necessitates flexibility of the team composition, the scope and goals of each participant’s tasks, their competence profiles and how they complement each other, and the way in which they exercise agency and negotiate their decision-making spaces. Trust towards the translator plays a major role when the revisers exercise their agency in editing the translation, and broken trust will have consequences. Ultimately, the success or failure of the combined effort may be determined by the complementarity of the translator’s and the reviser’s competence profiles: if one of the participants lacks a required competence or experience, the other must have it. Furthermore, the reviser must be able to identify the translator’s strength and weaknesses correctly in order to be able to correct what needs to be corrected, and to avoid making changes that deteriorate the quality of the translation.

The nature and characteristics of the cognitive collaboration have also been considered in the articles with regard to the creativity of the combined translation process. The distributed process of translation production, carried out by a translator and a reviser/editor, was described as a creative process that included repeated phases of divergent and convergent thinking that the participants engaged in individually and together as they propose translation solutions, evaluate them, and accept or reject them. Creativity was also discussed in connection with different text genres and what they require of the translator and the reviser: texts that can be roughly categorized as fluency-oriented translations often require the consideration and selection of a larger variety of linguistic structures, resulting in relatively frequent translational shifts in the lexicogrammatical structures. Other texts, categorised as precision-oriented translations, could often be translated more directly, resulting in fewer shifts in lexicogrammatical structures.

The processes and circumstances that influence the collaboration have been explored firstly, through an account of project-specific factors that have an impact on the reviser’s task as part of the two-member system of socially distributed cognition, and secondly, through the identification and description of the revision file as a digital communication artefact that steers, limits and enables the revision work and the communication between the participants. The reviser’s task wasfound to take shape as a result of many interconnected and, at times, contradictory factors. Direct causal relationships between a background factor and a specific revision procedure were found to be rare. The most important project-specific influencing factors were the text genre, the translator’s experience and competence profile, and the client’s needs and requirements. The translator’s experience and competence profile was an important factor when choosing whether a translation would be revised or not, while the text genre and the client often influence the scope of revision, primarily with regard to the revision’s level of detail and the revision parameters being emphasised. The impact of text genre in particular can be summarised as the different treatment of fluency-oriented and precisionoriented translations.

The revision file has been identified in this dissertation as the primary communicative artefact that is used for building and shaping collaboration between the translator and reviser, conceptualised here as a system of socially distributed cognition, or a cognitive dyad. When this type of revision file is used in the production process, its characteristics as a digital artefact largely determine the revision procedure and how translators and revisers collaborate. The proposed translation solutions are communicated via tracked changes in the .docx file, and direct communication can be added to the files in the form of comments that are usually displayed in the margin of the file. It seems that the commenting function may not be used extensively, and other channels of communication are not often employed. This leads to a situation where communication between the participants is prevalently task-oriented.

In addition to these main themes, the dissertation also proposes and tests methods that may be useful for studying translation as a situated, distributed action in which the cognitive task is divided between two or more participants. Studying the revision file as a communicative device instead of purely as a tool is an important step forward in the research of collaboration in translation. In addition, the descriptions of the revisers’ task suggest that revisers are not just proofreaders but active participants in the translation effort. They participate in problem-solving in the distributed creative process of translation production, using the revision file as the primary communicative artefact and the vehicle of collaboration. This perspective to the revisers’ work marks a shift from how revisers have previously been seen in translation studies and prepares the ground for further research into their role in changing production systems in the age of large language models and translation tools based on them.