The Smooth Signal Redundancy Hypothesis (SSRH) states that there is an inverse relationship between language redundancy and acoustic saliency. Less redundant items, e.g. infrequent or unpredictable ones, become more salient, and vice versa. The SSRH further assumes that prosodic structure, i.e. prosodic boundaries and prominence, mediates the relationship between language redundancy and acoustic saliency. In this paper, we tested whether semantic priming, one of the measures of language redundancy, affects prosodic structure at word boundaries. In a production experiment we presented German sentence pairs with identical target words. These target words were presented either in a context where they were primed by semantically related words or in a context where they were not primed. Results showed an effect of semantic priming on prosodic structure in that primed targets were significantly shorter than non-primed ones. This effect was increased when measures of lexical frequency were taken into account as well.