Constituent morphemes within complex words are identified during visual word processing, and likewise the presence of real morphemes slows down pseudoword rejection. Previous studies suggest that the identification of constituents in compound words is, to a certain degree, position-independent, reflecting the fact that constituents can occur in any position within compounds across the language (e.g., boathouse - houseboat). However, some stems occur more often in the first position of a compound (e.g., doorstep, doorstop, doorknob, doorbell vs. trapdoor), while others occur more often in the second position (e.g., background, playground, underground vs. groundwork), and some stems virtually never appear in a given position (e.g., chinbone, chinrest, but no compound with chin in final position). The question therefore is: Is the cognitive system sensitive to this specific positional information of stems or does it just make a general distinction between position-independent stems and position-bound affixes? We addressed this question with a lexical decision task on pseudocompounds in Italian. We found a small effect for position-specific morpheme interference and a strong effect of interpretability. Interpretability itself, however, was also driven by morpheme position. From these findings we conclude that a stem’s positional distribution throughout language is taken into account as an additional source of information in compound processing and partly enters through the "backdoor" of interpretability