Scholars in the humanities often have to account exhaustively for the structure of large masses of data. Tree-diagrams implemented by means of suitable computer programs can be of considerable assistance in achieving a cohesive representation of the data. This paper discusses the respective merits of the two main approaches to tree representation and introduces a new method based on the use of unrooted trees. After a detailed examination of the topological properties of such trees, two algorithms are described. The second part of the paper consists in practical applications of the method of tree representation to a corpus of contemporary English poetry. Several sets of data made up of both lexical and grammatical items (adjectives, modals, auxiliaries and personal pronouns) have been submitted to the method. The findings are assessed in terms of their heuristic value in the light of modern linguistic theory and compared with the results obtained by means of more traditional statistical procedures.