According to Alternative Semantics (Rooth, 1992, see also Krifka, 2008), the function of focus is to signal that alternatives are important for the interpretation of an utterance. Research on online sentence processing has investigated whether listeners, when they hear a contrastively accentend focus element, automatically activate alternatives to this focus element (Braun & Tagliapietra, 2010; Husband & Ferreira, 2016). The activation of alternatives is usually tested by means of a lexical decision task. It is a common assumption in psycholinguistic research that the time to decide whether a letter string is a word in the participant's language is dependent on the pre-activation of this word in the participant's mind. Thus, all else being equal, if contrastive focus accent activates alternatives, a given word should be recognized faster if it is a focus alternative to the contrastively accented word. This effect should go above and beyond semantic priming found in similar studies. Husband and Ferreira (2016) claim that, initially, all words that are related to a prime word in a spoken sentence, will be primed. However, contrastive accent triggers a selection mechanism such that, over time, only contrastive alternatives to the accented word stay primed. By contrast, if the prime word is neutrally accented, no such selection mechanism will be triggered such that all related words will stay active. The critical finding in their study was that, after a delay of 750 ms, for CONTRASTIVELY accented words, only contrastive alternatives were primed, whereas non-contrastively related words were no longer primed. By contrast, for NEUTRALLY accented words, both types of related words were primed, relative to an unrelated control condition. For linguists interested in the online processing of focus, this is such a crucial finding, that we wanted to replicate it in German, with more items, to increase statistical power. To anticipate the results: At 750 ms (the "late" SOA in Husband and Ferreira), both contrastively and non-contrastively related words were primed, for both contrastive and neutral accent. Since German words tend to be both longer and morphologically more complex than English ones, we ran another experiment with a SOA of 3000 ms. Again, in the conventional analyses (see below), there was equal priming for both types of related words in both prosodic conditions. The post-hoc analyses shown on the poster suggested that, if anything, for NEUTRAL accent ("broad focus"), non-contrastively related words are no longer primed at the late delay. This finding emphasizes the need of replication studies. An interesting thing in the present study is that we did not fail to replicate a previous effect such that it was not significant. Instead, comparisons that should not have been significant according to the theory were significant. Post-hoc power analyses with the R-package simr (Green & MacLeod, 2016) showed that statistical power was way above 80% in both experiments. We carried out two different types of analyses: The first replicated Husband and Ferreira: The condition for unrelated probe words with neutral intonation was fixed as reference level and the other five design cells were all compared to this reference level. The second analysis used sum-coding for probetype and focus and looked at main effects and the probetype * focus interaction. An overview of the results is provided under "Files". References: Braun, B. & Tagliapietra, L. (2010). Language and Cognitive Processes, 25, 1024-1043. Green, P. & MacLeod, L. (2016). Methods in Ecology and Evolution, 7, 493-498. Husband, E.M. & Ferreira, F. (2016). Language, Cognition, and Neuroscience, 31, 217-235. Krifka, M. (2008). Acta Linguistica Hungarica, 55, 243-276. Rooth, M. (1992). Natural Language Semantics, 1, 75-116.