Dalarna University's logo and link to the university's website

Appropriate use of linking adverbials (LAs) is a key feature of successful academic writing because these devices(e.g., furthermore, however, thus) enhance meaning and establish textual cohesion explicitly (Shaw 2009).Previous research has shown that LAs appear prominently in academic prose. In fact, these studies have revealedthat academic writing includes more LAs than other registers including conversation, fiction, and news (Biber etal. 1999; Liu 2008). Despite their importance in academic writing, second language (L2) writers of English havebeen reported to struggle to use LAs appropriately. Over the past few decades, considerable research has comparedthe use of LAs between first-language (L1) English writers and various L2 English groups including L1 Chinese(e.g., Gao 2016), L1 Korean (e.g., Ha 2016), and L1 Spanish writers (e.g., Carrió-Pastor 2013), as well as amongspecific L1 groups (e.g., Appel & Szeib 2018). These studies have shown that L2 English writers frequentlyoveruse, underuse, and/or misuse these devices. While these studies have been important in understanding L2writers’ challenges with LAs, surprisingly little attention has been given to whether L2 students’ use of LAs intheir writing changes over time or the degree to which their behaviors change with experience. Using corpusbased methods, this study reports findings of an analysis of the developmental trajectory of English-as-a-secondlanguage (ESL) university students’ use of LAs in their academic writing. The study was guided by the followingresearch question: To what extent does L2 university students’ use of linking adverbials in their writing changeover time? Through this analysis, this study aims to provide a greater understanding of the relationship betweeneducational experience and L2 writing development.Data consist of a specialized corpus of 126 high-rated source-based argumentative essays written by 63ESL undergraduate students in US-based first-year writing (FYW) courses at two different points in time. Thefirst subcorpus (ESL-1) includes 63 argumentative essays (66,424 words) written by these students in the first oftwo FYW courses, while the second subcorpus (ESL-2) consists of 63 argumentative papers (87,638 words)written by the same student writers in the second FYW course. To analyze LAs in the student essays, Liu’s (2008)taxonomy of LAs was used because his list is considered to be one of the most comprehensive (Gao 2016), witha total of 110 lexical items. The framework consists of four broad semantic categories: additive (e.g., additionally,similarly), adversative (e.g., however, in contrast), causal (e.g., as a result, hence), and sequential (e.g., first, inconclusion). Each category in this framework is classified further into subcategories. Using the concordance toolAntconc (Anthony 2018), every LA item in Liu’s (2008) list was searched in both subcorpora, and then wemanually examined each example in its textual context to ensure every item functioned as an LA. Item frequencieswere counted per text and normalized per 1,000 words. To determine whether the differences were statisticallysignificant, paired samples t-tests, with Bonferroni correction, were performed, with the alpha set at .05 (twotailed).Analysis reveals statistically significant changes in the overall frequency of LAs, with the ESL-2subcorpus consisting of fewer LAs than the ESL-1 subcorpus. Upon closer analysis, the results show that the useof additive and causal LAs decreased over time, while adversative and sequential LAs increased. However, astatistically significant difference was only found for the additive category. Analysis of the proportionaldistributions of the categories shows that with experience ESL student writers rely less on additive and more onadversative, yet the distributions of causal and sequential do not seem to change. With a few exceptions, the mostfrequently used words/phrases for all the categories are strikingly similar in both subcorpora, though thefrequencies at which they are used changes. Thus, the preliminary findings suggest that the distribution of LAsappears to change and matches more closely with published academic prose (cf. Liu 2008) as ESL students gainmore experience with academic writing; however, the specific linguistic LA devices used do not seem to markedlychange. The paper begins by reporting and discussing the results, followed by implications for L2 writing researchand pedagogy.