The Phonology of Second Occurrence Focus 1 Caroline Féry and Shinichiro Ishihara University of Potsdam This paper investigates the question of whether and how ‘Second Occurrence Focus’ (SOF) is realized phonetically in German. The apparent lack of phonetic marking on SOF has raised much discussion on the semantic theory of focus (Partee, 1991; Rooth, 1992). Some researchers have reported the existence of phonetic marking of SOF (Rooth, 1996; Beaver et al., 2007), claiming that SOF is not marked by pitch, but by duration, although it is very subtle at best. In these studies for English, however, SOF always appears at a postnuclear position, where no pitch accent can appear for an independent reason. That is, the lack of pitch accent on SOF is not due to the lack of F- marking, but rather to the phonological condition where SOF appears. In our experimental study with German sentences, we examined both sentences with prenuclear SOF and with postnuclear SOF, comparing with their first occurrence focus (FOF) and non-focus counterparts. The results show that the phonetic prominence of focus (higher pitch/longer duration) is realized differently according to the type of focus as well as according to the position of the target expression. We account for these differences by considering several phonetic effects, those that are information structure-related and those that are phonologically motivated. Keywords: Second Occurrence Focus (SOF), German, prosody 1 Introduction The phonetic realization of SOF is related to the question of the interface between LF and PF in grammar. In this paper, we investigate the question whether and how ‘Second Occurrence Focus’ (SOF) is realized phonetically in 1 Some parts of this paper were presented at P&P2 in Tübingen, SFB632 Internal Workshop, NELS36, and the 23rd ELSJ Conference in Fukuoka. We would also like to thank two anonymous reviewers, as well as the following people for their comments and discussion: David Beaver, Daniel Büring, Gisbert Fanselow, Ingo Feldhausen, Manfred Krifka, Frank Kügler, Mats Rooth, Shravan Vasishth and Ede Zimmermann. Many thanks also go to Felix Engelmann, Elisabeth Fleischhauer, Pawel Logatschew, Susanne Gensel, Kristin Irsig and Esther Sommerfeld for their help with the experiments and data analysis. This study is part of the SFB632 “Information Structure” funded by the DFG (project A1).