Numerous techniques have been developed for text entry by gaze, and similarly, a number of evaluations have been carried out to determine the efficiency of the solutions. However, the results of the published experiments are inconclusive, and it is unclear what causes the difference in their findings. Here we look particularly at the effect of the language used in the experiment. A study where participants entered text both in English and in Finnish does not show an effect of language structure: the entry rates were reasonably close to each other. The role of other explaining factors, such as calibration accuracy and experimental procedure, are discussed.