Scene text analysis stands as a cornerstone of computer vision, with logical layout analysis emerging as a pivotal component for decoding the semantic roles of text regions and advancing scene understanding. While previous research has primarily focused on structured documents, scene text poses unique challenges due to its unstructured and visually diverse nature. In this work, we extend logical layout analysis to signboard images, a domain characterized by complex backgrounds, irregular text placements, diverse viewpoints, unique fonts, varying text sizes, and personalized design styles. These challenges make scene text layout analysis a critical yet underexplored research problem with significant real-world implications. To bridge this gap, we introduce a new benchmark dataset comprising 2,025 manually annotated images from diverse urban environments, containing 44,227 text instances across 9 semantic categories commonly found in signboards. Additionally, we evaluate state-of-the-art logical layout analysis methods on this dataset, providing valuable insights into the challenges and opportunities within this domain. The dataset and code are publicly available (GitHub Repository: https://github.com/Yangchann/LLASignboard .), fostering further advancements in scene text understanding