This research aims to propose alternative visual design of audio railway disruption announcements in order at reducing handicap situations encountered by travelers who can't hear or understand oral messages (deaf, hearing-impaired or foreign people). Five announcements were translated into two visual modes: in French sign language (with a sign language avatar named Jade) and in animated graphical designs. Different combinations were tested in order to find the better compromise between spatial and temporal organization for the both visual messages. For spatial parameter, messages were either juxtaposed (the avatar and the graphical message were side by side) or integrated (the avatar and the graphical message are mixed); for temporal parameter, messages were either totally alternately presented or with a synchronized start or with a partial synchronization (for messages parts having the same meaning). Forty two deaf and 24 hearing people who used sign language (mean age 33.9, sd=7.3 and 31.2, sd=8.6 respectively) were included in the experiment. They were told to look at the messages carefully on a screen in order to understand them and were asked to give their meanings. Each message was presented two times. The observer's eyes movements was registered while he watched the screen (with an eye tracking system) in order to know exactly where and when he was looking. First results showed significant higher comprehensive score for deaf than for hearing participants. Integrated display appeared to be better than juxtaposed presentation. No significant difference was recorded between temporal displays (alternative vs. synchronized). Eye tracking data are under analyzed. Currently inclusion of participants is still ongoing. The prospect of this work will allow designing an innovative system which will be able to display a translation of the sound messages in train stations.