The role of working class mothers’ talk in explaining their child’s ‘impoverished’ language development and the resulting ‘word gap’ between social classes is hotly debated. Academic research in this area spans decades, crosses continents, and gathers up a wide range of disciplines and positions, ranging from research seeking to intervene in and optimise mothers’ talk, to research that vigorously criticises any attempt to do so. Through an extended analysis of Jacques Donzelot’s seminal study The Policing of Families, and Michel Foucault’s concept of a ‘regime of truth’, we explore how motherhood is constructed by academic debate as something to be endlessly optimised. Academic debate functions by (1) reducing expectations concerning the role and remit of experts so as to place the onus on mothers to implement findings which the former will facilitate, (2) complicating the contributing factors to language delay, thus avoiding apportioning blame too directly whilst giving endless cause to do further research, and (3) committing mothers to a permanent labour in which they are expected to better themselves as measured by the manifest language development of their children. Its strongest critics remain within the constraints of this regime of truth to the extent that they argue for humility of expertise, for recognition of broader sociocultural factors, and for the importance of privileging the expertise and agency of mothers. This article considers how all parties are, in effect, obliged to declare the truth of motherhood and will find themselves implicated in its governance.