The representation of social issues in public discourse and newspaper language: a comment on Gardikiotis, Martin and Hewstone's (2004) study