ABSTRACT

Research article (RA) abstracts succinctly and skilfully epitomize the core information of the full text and have thus attracted the attention of a number of scholars. While previous studies mainly focused on the rhetorical structures, meta-discursive features and lexico-grammatical features, few have made explorations from the perspective of information theory. To bridge this gap, the present study conducted an entropy-based analysis to explore the distribution pattern of information content across moves and the variations across disciplines. 318 RA abstracts across the natural sciences, social sciences and humanities (106 abstracts per discipline) were selected and three indices, i.e. the 1-/ 2-/ 3-gram entropies, were used to examine whether different indices yielded different features. The results show that in an RA abstract, the information content is unevenly distributed across moves; different entropy indices may reflect different linguistic properties; and both similarities and variations exist in information content across disciplines. These phenomena can be attributed to the functions of moves, the linguistic meanings of indices and disciplinary features. This study has implications for RA abstract writing instruction and practice, as well as for broadening the applications of quantitative linguistic methods into less touched fields.

摘要

研究文章（Research article, RA）摘要简洁、巧妙地概括了全文的核心信息，因而引起了众多学者的关注。以往的研究主要集中在修辞结构、元话语特征和词汇语法特征上，很少有人从信息论的角度进行探索。为了弥合这一差距，本研究进行了基于熵的分析，以探索信息内容在移动中的分布模式和跨学科的变化。选择了自然科学、社会科学和人文学科的 318 篇 RA 摘要（每个学科 106 篇摘要），并使用三个指标，即 1-/ 2-/ 3-gram 熵来检查不同的指标是否产生不同的特征。结果表明，在 RA 摘要中，信息内容在移动中分布不均；不同的熵指数可能反映不同的语言特性；跨学科的信息内容既有相同点也有不同点。这些现象可以归因于动作的功能、指标的语言意义和学科特征。这项研究对 RA 摘要写作指导和实践具有影响，以及将定量语言学方法的应用扩展到较少接触的领域。