One of the reasons why I believe the consensual tradition of Christian orthodoxy deserves so much deference is that its theological language has been time-tested. It has been tested in the laboratory of Christian history and Christian experience. It has passed through the crucible of ecclesiastical conflict and has been vindicated by lay Christian consensus across time and space. The challenges of translation and contextualization still remain, but the semantic categories passed down to us have survived for a reason.
Calvin once mused about a scenario in which no extra-biblical language would be needed in order to communicate what Scripture clearly teaches about the nature and works of the Triune God. But, as Calvin rightly conceded, Christian theologians must employ extra-biblical terminology, not because the language of Scripture is insufficient, but because it is so often distorted.
This is also why I am so wary of newer categories that have little to no precedent in the Christian tradition. The historical categories of Trinitarian orthodoxy–hypostasis and ousia, procession and mission, inseparable operations and appropriation–have been tested and tried–biblically, theologically, philosophically, and pastorally. They have, as a result, a kind of sturdiness and reliability that can’t be found in the newer categories of so many recent evangelical treatments of the Trinity–like the granite walls of Yosemite compared to loose shale. The newer terms–relationship, role, functional subordination, eternal relations of authority and submission–are, at best provisional, and must undergo a significant probationary period in order to test their biblical and theological utility. In some cases, their incommensurabilty and inconsistency with the traditional ways Christians have interpreted Scripture and the Triune mystery at its heart are more than apparent.
So for my part, its better to tread the old paths of orthodox terminology, with all of their careful and intricate beauty and rationality, than to begin afresh with newer and less tried alternatives.