earticle

영어

This study presents an analysis of coherence in interpreting output using rhetorical structure theory(RST). An English speech (ST) was interpreted into Korean by ten interpreting students, which were then transcribed for RST analysis. Each text was segmented into elementary discourse units, and relations were assigned to text spans based on the four fields of RST relation definitions. RST annotation tools were employed to create RST trees for each text analyzed. Three features of RST trees were selected and used in describing coherence in interpreting outputs. First, the depth of RST trees of TTs were observed. As the depth of each RST tree indicated the degree of hierarchical organization of ideas, a TT with a deeper tree was deemed to represent more advanced interpreting skills. Second, occurrence of a non-sequitur was discovered in a TT and its impact on the overall texture was explained by using the RST tree which starkly visualized non-sequiturs as unconnected root nodes. Finally, RST relations contained in the ST were compared with those in TTs in order to describe interpreting accuracy on the level of coherence. Comparison of top-level schemas in the ST and the TTs provided a simple and clear way of identifying coherence shifts between the original and the interpreting outputs.