This paper conducts a comprehensive exploration of methodology in historical linguistics, focusing on language subgrouping. Employing Tangut, a severely eroded medieval language, as a case study, it scrutinizes previous linguistic analyses that depart from the rigorous Neogrammarian method, specifically referencing Beaudouin (2023). These non-compliant analyses have impeded recent progress in understanding the genetic relationships within Burmo-Qiangic, a field marked by prolonged debates and with gradual advancement recently. In a subsequent step, adhering to Neogrammarian principles, namely, Ausnahmslosigkeit der Lautgesetze and positive shared innovations in language subgrouping, the paper discusses the plausibility of, as well as the good practice to argue for, a “Tangut-Horpa clade” within the Gyalrongic branch of Burmo-Qiangic. By advocating for the universality of these Neogrammarian principles, the paper aims to improve the accuracy and reliability of subgrouping languages characterized by significant typological diversity. This, in turn, contributes to a deeper comprehension of rigorous methodology within the context of the Sino-Tibetan language family.