Argumentation theory combines philosophical concepts and computational models to deliver a practical approach to reasoning that handles uncertain information and possibly conflicting viewpoints. This paper focuses on the structured approach to argumentation that incorporates domain specific knowledge and argumentation schemes. There is a lack of implementations and implementation methods for most structured models. This paper shows how taking a principled approach, using the programming language Haskell, helps addressing this problem. We construct a framework for developing structured argumentation models and translations between models (given intertranslatability of models). We furthermore provide a methodology to quickly test and formally prove desirable properties of such implementations using a theorem prover. We demonstrate our approach on the Carneades argumentation model and Dung's abstract argumentation frameworks, implementing both the models and a translation from Carneades into AFs. We then provide implementations of correspondence properties and an initial formalisation of Dung's AFs into a theorem prover. The final result is a verified pipeline from the structured model Carneades into existing efficient SAT-based implementations of Dung's AFs.