言据性指说话人对知识来源以及态度或介入程度的说明。英语中存在着丰富的表达言据性的语言资源，它们在建构作者权威性与学术语篇可信度方面发挥着重要作用。因此言据性在近年来受到了学界广泛的关注。此前，大量研究关注的是不同群体习作中据素的使用规律，以及其中言据性发挥的功能，但对此，学界尚未形成统一的、共识性的结论。与此形成鲜明对比的是，目前仅有少数研究讨论了言据性在完整学术论文中词汇语法层面上的实现手段，并就学术新手与专家学者之间的据素使用情况——这一重要的学术写作教学议题——进行了对比研究。 基于修订的Chafe（1986）与杨林秀（2009）的据素模型，本研究将言据性在词汇语法层面上的实现手段分为四个主要范畴：感官据素、信念据素、汇报据素和推断据素。每个范畴都可被进一步细化成几个子范畴。本研究收录的两个语料库分别为十篇硕士毕业论文与十篇国际核心期刊发表论文。基于量性与质性研究方法，本文考察并对比了学术新手与专家作者学术论文中不同据素的使用情况。 研究发现：总体而言，学术新手和专家作者在写作中采用相似的言据性资源（四种主要据素）；相比于新手，专家习作中言据性资源在词汇语法层面上的实现手段更加复杂多样。进一步探究表明，新手在习作中倾向于用感官据素，少用推理据素，体现他们有过度使用主观感知而非逻辑推理的倾向。针对报道据素的分析发现：新手缺乏明确表明自己声音、并在习作中将其一以贯之的自信；他们倾向于中立地汇报以往研究，避免进行主观评价。与新手不同，专家有意识地减少感官据素的使用，坚持作者声音的输出，在客观分析和主观介入两者中找到了平衡。专家通过在习作中更频繁地运用符合科学思维的推断据素、更多地采用各类汇报据素对文献进行评价整合、对命题表现出更高的确定性，从而建立起一个更负责任、更值得信赖的作者形象。 导致两者差异的因素包括语言水平、文化差异和学术能力。该研究将对言据性资源在学术语篇中的运用提供新的理解，对学术英语写作教学也有一定的启示意义。

Evidentiality is one type of linguistic devices for coding information sources and the speaker’s attitudes towards it. English has a rich repertoire of linguistic resources to express evidentiality. It plays a vital role in constructing authority and responsibility of academic writing, and has attracted extensive attention in recent decades. A substantial body of research has explored the patterns of evidentials in different groups’ writing and analysed the function of evidentiality. However, no consensus has been drawn. Relatively few studies focus on the lexicogrammatical realizations of evidentiality in holistic research papers and compare novice writers’ and experts’ use of evidentials, an issue of great pedagogical value in academic writing teaching and learning. Based on a revised version of Chafe’s (1986) and Yang’s (2009) framework of evidentials, the lexicogrammatical realizations of evidentiality are divided into four major categories: sensory evidentials, belief evidentials, reporting evidentials and inferring evidentials. Each category can be further classified into several subcategories. Two corpora are established: one is ten Chinese master’s dissertations, and the other is ten papers published on an international core journal. Combining quantitative and qualitative method, this thesis conducts a comparative study on the deployment of evidentials in novice writers’ and experts’ academic writing. The analysis reveals that (i) novices and experts deploy similar resources to express evidentiality, (ii) the lexicogrammatical realizations of evidentiality in experts’ writing are much more various and complex than those in novice writers’ writing. A further investigation shows that novice writers tend to overuse sensory evidentials and deploy inferring evidentials less frequently in their writing, indicating their dependency on subjective perception rather than logical reasoning. The analysis of reporting evidentials reveals that novice writers are not confident in projecting their voice explicitly and consistently in their writing. Rather, they are inclined to report previous literature in a neutral way without critical evaluation. In contrast, experts find a better balance between objective analysis and subjective involvement by restraining the application of sensory evidentials and projecting their own voices consistently. In addition, experts establish a more responsible and reliable author identity by (i) various inferring evidentials to signal their scientific thinking, (ii) diverse reporting evidentials to evaluate and synthesize literature, and (iii) higher certainty of their propositions. These differences may be attributed to language proficiency, cultural differences and academic capability. This study adds new knowledge to our understanding of evidentiality, and it may be beneficial to academic writing teaching of for novice writers in China.