Introduction: Plenary speeches provide politicians with a crucial means to demonstrate leadership and seek endorsement, defined as the voluntary recognition by others that a focal actor is granted influence. A common rhetorical device is the use of "we"-language, often treated as a uniform signal of alignment between speaker and audience. Yet, communication research distinguishes between inclusive "we" (we-as-us) and exclusive "we" (we-versus-them), suggesting they may evoke distinct responses.


Purpose: We disentangle whether politicians' inclusive and exclusive uses of "we"-language differentially predict their immediate and distal endorsement.


Method: We conducted an analysis of 3,606 members of parliament and 119,344 plenary speeches delivered in the German Parliament. Moving beyond traditional word-counting methods, we used new computational text analysis tools (contextual word embeddings with vector projection analysis) to identify inclusive and exclusive "we"-language in plenary speeches. We linked them to two indicators of endorsement: (1) immediate approval through applause and interruptions by fellow members of parliament, and (2) longer-term electoral outcomes.


Results: Inclusive and exclusive uses of "we" were only moderately correlated at both the speech and the speaker level, suggesting that politicians typically emphasized one form over the other - both within individual speeches and across their broader political profile. Inclusive and exclusive "we"-language showed distinct associations with audience reactions: While both were linked to applause, inclusive "we"-language was associated with fewer and exclusive "we" language with more interruptions. Combining the two tended to weaken these effects, indicating that each works best on its own. Both inclusive and exclusive "we"-language was associated with higher reelection likelihood. Combining them did not alter this association, indicating neither benefit nor harm.


Conclusions: By disentangling inclusive and exclusive forms of "we"-language with new computational text analysis tools, this study provides a more nuanced understanding of how collective pronouns shape leader endorsement in the political context.