The degeneracy distribution of the genetic code has been addressed by a theorical model based on non-power representation of integer numbers. The mathematical properties of this model led to the definition of dichotomic classes, i.e. non-linear functions of the information of two adjacent bases, that proved to be intimately linked with the chemical properties of the codon bases, as well as with the duality of Rumer's codon classes. Although theoretically derived, dichotomic classes permitted to build statistical classifiers able to retrieve the correct translational frame of coding sequences, pointing at a tentative biological significance. Here, we will discuss possible experimental approaches to verify the biological relevance of dichotomic classes in translation. Finally, the construction of a reporter system will be presented, capable to measure the translational efficiency of di-codons nesting different dichotomic class combinations.