The impact made by a scientific paper on the work of other academics has many established metrics, including metrics based on citation counts and social media commenting. However, determination of the impact of a scientific paper on the wider society is less well established. For example, is it important for scientific work to be newsworthy? Here we present a new corpus of newspaper articles linked to the scientific papers that they describe. We find that Impact Case studies submitted to the UK Research Excellence Framework (REF) 2014 that refer to scientific papers mentioned in newspaper articles were awarded a higher score in the REF assessment. The papers associated with these case studies also feature prominently in the newspaper articles. We hypothesise that such prominence can be a useful proxy for societal impact. We therefore provide a novel baseline approach for measuring the prominence of scientific papers mentioned within news articles. Our measurement of prominence is based on semantic similarity through a graph-based ranking algorithm. We find that scientific papers with an associated REF case study are more likely to have a stronger prominence score. This supports our hypothesis that linguistic prominence in news can be used to suggest the wider non-academic impact of scientific work.