Positional analyses in political science target the identification of (dis) similarities between parties based on their stance on a given policy or political demand (claim)–which are, unsurprisingly, a well explored source of information for text-as-data approaches to this task. Political actors, however, do not only make claims regarding a given policy: they often provide justifications (frames) for their claims. Frames are likely to provide novel perspectives on positional analysis: they have, however, been largely neglected so far. Our work fills this gap: In a first experiment, we show how including political frames in the analysis leads to a more accurate picture using categorical (manually labelled) data from two datasets: i) the Manifesto dataset from the 2021 German federal election, and ii) a more structured dataset extracted from the German voting advice application “Wahl-O-Mat”. In a second experiment, we investigate whether transformer-based language model representations are able to infer party relations from the textual information, independent of the annotation. Our approach is a) able to identify relevant differences between claims and frames and b) represent the political spectrum when applied to the more structured dataset.