In interpretation, the converted version should be able to cause the same reaction from the audience as the speaker intended with his original speech. To that end, transforming the sense of a language into the corresponding linguistic units of another language must be done at the deep structure level, not at a surface level. Most interpreters are well aware of this fact, in principle, but they are not free from occasional word for word translation. To explore the cause of this phenomenon, this study divided college students into two groups and showed English movie clip to both groups. One of the group’s clip did not have subtitles while the other group’s clip had Korean subtitles. Comprehension was accessed by a twenty-item fill-in-the-blank test. The results showed that students in the experimental group with Korean subtitles retrieved English words more easily than students in the non-subtitle group. In the experimental group itself, literally translated Korean subtitles brought higher scores than Korean subtitles transformed at the deep structure. This indirectly demonstrates that converting words at surface level requires less information processing capacity than the processing at deep structure level. This explains, in part, why interpreters sometimes end up with literal translation in their simultaneous interpretation. The extreme multi-processing under time pressure might be responsible for this.
