This thesis tests the hypothesis that interpretation based on meaning is likely to be more cognitively demanding than interpretation based on the words and syntax of the original (surface structure). This is because in the former case surface structure must be processed and then meaning identified, while the second step is absent from the latter approach. Four student interpreters performed consecutive interpretation with notes and the time lag between hearing the input and the start of notation (ear­-pen span) was taken as a proxy for cognitive load. All interpreters had a lower ear­-pen span when noting words than when encoding meaning in symbols, suggesting that cognitive load was lower for words. The ear-pen span for abbreviations was longer than for words, perhaps because selection or creation of abbreviations increased cognitive load. Where symbolic notation was used, the syntactic structure of the interpretation mirrored the original in around half of cases, suggesting that the surface structure is regularly used in rendition even where notes are based on meaning.