The development of scientific software relies on the collaboration of various stakeholders for the scientific computing and software engineering activities. Computer languages have an impact on both activities and related concerns, as well as on the engineering principles required to ensure the development of reliable scientific software. The more general-purpose the language is —with low-level, computing-related, system abstractions— the more flexibility it will provide, but also the more rigorous engineering principles and Validation & Verification (V&V) activities it will require from the language user. In this paper, we investigate the different levels of abstraction, linked to the diverse artifacts of the scientific software development process, a software language can propose, and the V&V facilities associated to the corresponding level of abstraction the language can provide to the user. We aim to raise awareness among scientists, engineers and language providers on their shared responsibility in developing reliable scientific software.