Effect of cognitive load on working memory updating and prose recall skills of interpreters

Aims and objectives: The aim was to study working memory updating (WMU) and prose recall of two spoken-language interpreter groups, consecutive and simultaneous. The main question was whether cognitive load affects performance and whether specific experience develops specific working memory (WM) processes, manifested as between-group differences under increasing cognitive load. Design: Two groups of expert interpreters, 17 simultaneous (SIs) and 18 consecutive (CIs), participated in two experiments. For WMU, the effect of group and level of cognitive load was studied with a category experiment. For prose recall, speech sequences of 13–44 words were used to study cognitive load. Data and analyses: WMU data include cognitive performance measures: number of correctly updated words (serial order, number of categories) and number and types of errors vis-à-vis cognitive load. Prose recall data were analysed with idea units, plot point scores and reproduction of metatextual categories. Findings: Results showed that cognitive load affects performance: updating was impaired in both groups in condition ’simultaneous translating and repeating’. SIs performed better than CIs, which was confirmed further by error analysis. In prose recall, the effect of cognitive load was obvious: recall of extra-long speech sequences was impaired in both groups. Results also confirmed that with speech sequences exceeding 30 words, informational density and cognitive load increase. Originality: This is the first study comparing WMU skills of SIs and CIs under varying cognitive load. It introduces a modified category WMU task with simultaneous translation, and new analysis methods for prose recall: plot point scores and reproduction of metatextual categories. Significance: A new contribution is that cognitive load affects WMU performance, less with SIs than with CIs. The experiments show that a well-targeted design can reveal differences between groups using different interpreting modes (SI vs. CI) that probably also exist in authentic interpreting situations.