Structural and functional differences between bundles of different lengths: A corpus-driven study

This pilot study aims to investigate the differences between varying lengths of bundles in structure and function by comparing the 100 most frequent three-, four-, and five-word bundles in a self-built corpus of dissertations which contains about 3.5 million words. The findings reveal considerable variances between bundles of different lengths in terms of both structure and function. In general, the variances between three-and four-word bundles are greater than those between four-and five-word bundles, and three-and five-word bundles. Structurally, three-and four-word bundles differ significantly in all six main categories. Four-and five-word bundles vary in five categories, while three-and five-word bundles are only different in four categories. Functionally, noticeable variances were observed in research-, text- and participant-oriented bundles between three-and four-word bundles, and three-and five-word bundles. However, four-and five-word bundles only differ significantly in text- and participant-oriented bundles. Interestingly, bundles of varying lengths also vary in patterns that are used to perform the same functions. The results of this study might inform researchers that they need to take bundle lengths into consideration when making generalizations of their findings or comparing bundles between various studies.