The poetry of the First World War bears poignant testimony to the unprecedented tragedies of human destructiveness. It has been the subject of numerous studies that looked at it from diverse perspectives, drawing from various contemporary psychological, historical, and literary theories. The present study can be regarded as yet another contribution to the existing picture of this rich and enduring field where literature and history are explored alongside. Informed by the emerging field of digital literary studies, this article aims to revisit the war experience through a computational analysis of the poetry of the Great War and explore how soldier-poets conveyed their emotions in the face of traumatic war experience. We adopted a top-down corpus-driven approach that moved from key semantic domains to lexical items and then to their collocates. First, we detected the key semantic domains of the genre through a corpus tool, Wmatrix. 'Anatomy and physiology', 'religion and the supernatural', and 'senses' were identified as the key semantic domains in our corpus. The unusual abundance of binaries led us to group and treat these lexical items as a separate category: 'binary oppositions'. In the second phase of the research, we analysed these key semantic domains to foreground the frequent lexical items under each domain and explored the surrounding context of those items through their collocational networks exploiting another corpus tool, Lancsbox. Finally, we further investigated the frequent items and corresponding concordance lines to identify the common stylistic features in the Great War Poetry.