Most speech translation technologies, which mimic human interpreters, may eventually replace human interpreters. In contrast, our long-term goal is to construct an augmented intelligent system that can support interpreters when they struggle. As initial states, this paper explores the challenges encountered by professional interpreters based on subjective ratings and eye-tracking measurements (pupil diameter). We investigate different impacts of minute fluctuations in pupil diameter from different levels of professional interpreters in response to a stimulus. We analyze different levels of professional interpreters while they simultaneously interpret long utterances in two ways (Japanese-to-English and English-to-Japanese) to determine whether different impacts are experienced when the interpreters are native languages of the source language or the target language.