This article proposes a textual analysis of self-citations in humanities articles, foregrounding their dynamic, relational properties. The starting point is the idea that scientific texts are not simply the product of a more or less manipulative rhetoric, nor simply the formated productions of a highly normalized professional activity, but are also socially situated. The empirical study of the presence of self-citations in a corpus of around 100 humanities articles shows that while they present themselves as common-or-garden bibliographical references, they in fact occur in a particular way, crucial to the construction of shared cognitive space. Such space is meaningful in terms of the scientific and intellectual career of the author and of their links to other persons or entities cited in their text. The article thus seeks to analyze scientific publications as textual constructs which use bibliographical references as a way of bringing dynamism to the relations between researchers.