LegitimNarrate: The first Dataset for Assessing New Venture Legitimacy through Narrative Analysis

New ventures face challenges due to their liability of newness and need to gain legitimacy to attract critical resources for growth and survival. The area of organizational legitimacy has captured considerable interest among management researchers. Scholars have deeply explored this concept, focusing on its definition, sources, dimensions, and attainment strategies. However, accurately measuring this construct continues to be a challenging endeavor. The advent of deep learning and natural language processing offers new opportunities for analyzing the legitimacy of new ventures through their narratives. This study analyzes crowdfunding narratives to identify legitimacy mechanisms that enhance entrepreneurs’ abilities to realize their projects. This paper introduces a dataset for text classification, specifically created for the analysis of new venture legitimacy. The dataset, gathered from Kickstarter plateform, contains 97 descriptions of French crowdfunding campaigns, segmented into 4955 sentences. The sentences were meticulously annotated by four experts in management, categorized by the presence and type of legitimacy mechanisms employed. Furthermore, we utilize widely-used pre-trained language models (PLMs) as baselines for this analytical task