Martin (1992: 123):
Endophoric reference typically involves presuming information from the preceding co-text (anaphora) or from the following co-text. Where presumed information follows, it usually appears quickly, either within the structure of the same nominal group, or in the same or an adjacent clause complex — otherwise participant identification would break down. As Halliday & Hasan (1976: 72-3) point out, the always points forward structurally, to information in modifying elements within the same nominal group, typically the Qualifier. This forward reference within the same nominal group will be referred to as esphora, adapting slightly a term from Ellis (1971).
Martin's rebranding of structural cataphora as esphora is poor theorising on several counts.
Firstly, although structural cataphora is a type of endophora, the term 'esphora' is equivalent to its opposite, 'exophora' (the prefix es- derives from Latin ex-).
Secondly, the term 'cataphora' means forward pointing, and so is the general term that subsumes the more delicate term 'structural cataphora'. Martin's rebranding fails to recognise this relation of delicacy.
Thirdly, the rebranding is made from the perspective of the view 'from below', that is: in terms of how each is realised in form — ± within nominal group — instead of from the functional perspective, 'from above', that is: in terms of the function (cataphoric reference) being realised.
Finally, even in the absence of the above-mentioned inconsistencies, the rebranding provides no explanatory advantage, and so is unmotivated theoretically. It merely puts Martin's stamp on a distinction already devised in the work of Halliday & Hasan.