What is it about?
Visually prominent entities affect how speakers end up describing scenes. One possibility is that they catch the speaker's attention and are then processed first, therefore taking pride of place in a sentence. These experiments find no evidence that this is the case. Instead, it's proposed that what visually salient entities do is favour specific interpretations of scenes. In sum, visual salience is affecting linguistic choices indirectly rather than directly.
Why is it important?
This research is important because, in language production, many researchers are using images to elicit sentences and tools such as eye movements to make inferences about the underlying processes, so we need to know what fixations on different parts of a visual field really mean and what effect they have on the preparation of the linguistic output.