Upper and lower limits of pronominal reference in discourse: topichood, anaphora and anadeixis

There is arguably a fundamental difference between the (canonical) anaphoric and ‘anadeictic’ indexical referring procedures in discourse. Whereas canonical anaphora serves to maintain the continuity of a previously-introduced discourse entity, integrating the two discourse units concerned into a composite whole, ‘anadeixis’ does not. Rather, its essential function is to ‘re-reference’ a previously-evoked entity whose topicality has faded, or which has not (yet) achieved topicality. The paper exploits this functional distinction in characterizing two polar configurations: first, at the ‘lower’ end, where a referent has just been evoked within a backgrounded clause or phrase. Here, an immediate reprise via an anaphoric expression (null anaphor, 3rd person pronoun) results in a minor hiatus; however, an anadeictic expression is more natural. By contrast, where a macro-discourse topic is evoked at the start of a text, but where a variety of more local topics have intervened subsequently, resuming this entity via an anaphoric rather than anadeictic marker is nonetheless felicitous. These two ‘polar’ scope spans have in common the property of topicality: in the former situation, the referent at issue is not (yet) a topic, hence retrieving it via an anaphoric marker is somewhat problematic; but in the latter case, the highly topical referent is resumable via anaphora perfectly naturally. The overall aim is to determine the extent and limits of anaphoric pronominal reference in non ‘pro-drop’ languages (here English and French), as a function of topichood. The examples presented give further evidence for the view that topics have a gradient status.