During the transformation of programs written in a source language to programs written in a target language structural mismatches need to be taken into account. A structural mismatch between two languages occurs when the transformation of a source AST node cannot substitute the source node by one node of the target AST. Tackling these structural mismatches with current day transformation systems is complicated and breaks the modularity of the transformation rules because of the implicit composition of the results of the transformation rules. Currently, the composition solely depends on the grammar of the source language and the structures created by other transformation. More explicit composition mechanisms are required to facilitate the implementation by means of separate and cross-language mechanisms which can restore the modularity of the transformation rules.