The Triangular AHP fuzzy scale assigns a fuzzy set of values to each term. These are on a triangular scale (peak =the most important value, base = the least important value).
When you have inconsistent pairwise comparisons, I think it is better to use the reciprocal, because instead of assigning fuzzy sets to the original values of the pairwise comparison matrix, it assigns fuzzy sets to the reciprocal values of the matrix. So, in general I think their difference is that the reciprocal scale's emphasis is on the middle value, while the regular scale distributes the weights more evenly.