Abstract
There exist gaps between evaluative meaning in the source text and that in the target text, even in scientific translation. These gaps result from three types of evaluative shifts: addition (from non-evaluative to evaluative), omission (from evaluative to non-evaluative), and adjustment (from evaluative to evaluative). This paper aims to find out what kind of evaluative expression tends to be involved in which type of evaluative shift. The data include 42 English articles in two scientific magazines and their Japanese and Chinese translation. Evaluative shifts seem random translational phenomena because they depend on the translator’s personal preference. Our results show that there are some tendencies in evaluative shifts in Japanese and Chinese translations. These tendencies may be explained by linguistic, social and translational factors.