This exploratory research project aims to investigate metadiscourse features in English essays written by upper secondary pupils attending schools in Norway, Sweden and the UK. Metadiscourse refers to the linguistic features that authors use to interact with their readers. This project recognises two main types of metadiscourse: signposting and stance. Signposts are words and phrases that authors use to guide their readers through the unfolding text. Stance markers are used to offer evaluations, navigate knowledge claims, and anticipate reader reactions. A large body of research has investigated the use of metadiscourse in professional and tertiary-level educational settings. However, comparatively few studies have investigated metadiscourse features in pre-tertiary essay writing. This research project contributes to this currently limited pool of research by analysing metadiscourse in final- year upper secondary pupils’ English essays in both L1 and L2 educational contexts. Furthermore, by incorporating interview methods, this research also aims to investigate English teachers’ general views towards metadiscourse and to what extent their instruction affects their pupils’ compositional decisions. The project involved collecting a corpus of non-fiction essays and holding interviews with teachers in upper secondary schools situated in Norway, Sweden and the UK. The essays were written for assignments set by teachers and grouped in five genres: political essays, literary essays, opinion pieces, linguistic investigations and commentaries. A metadiscourse taxonomy was adapted based on previous studies and a close reading of a sub-sample of 50 essays. The resulting taxonomy, which comprises 26 sub-categories and accounts for over 1,000 metadiscourse types, was utilised in four steps. Firstly, the types were used to electronically scan the corpus using a concordancing program. Secondly, the concordance lines were manually read to filter out non- metadiscoursal results. Thirdly, the number of each metadiscourse sub-category per 1,000 words in each essay was calculated. Finally, the descriptive statistics and concordance lines were used to identify trends regarding the use of each sub-category in the corpus. Additionally, semi- structured interviews were held with 19 teachers to gain insight into the metadiscourse-related advice they offered their pupils. The interview data were used to supplement the interpretation of the results from the textual analysis. The findings are reported in four articles that each focus on separate aspects of metadiscourse and different stages of the research process. Article 1 reports results from a preliminary study using a sub-set of 56 essays collected from the Norwegian and UK schools. This preliminary analysis was conducted in order to devise the adapted taxonomy, as well as to gain insight into the pragmatic usage of metadiscourse features in the upper secondary essays. Article 2 reports the results from an analysis of signposts in a corpus of 115 essays from the Norwegian, Swedish and UK schools, supplemented by data from the teacher interviews. Whereas the pupils frequently used a wide range of linguistic features to explicitly signal sentential relations, their use of markers that signal structural relations was somewhat sporadic, probably due to the short length of the essays. Although signposts were used similarly across the three educational contexts, their usage seemed to reflect the purposes of the target genres. While the UK teachers tended not to address the use of these features, the teachers in Norway and Sweden tended to provide pupils with decontextualised lists of signposts, which raises questions about whether upper secondary teachers in these L1 and L2 contexts should offer more explicit instruction in the pragmatic use of organisational features. Article 3 reports results from an analysis of epistemic stance and engagement features in the same corpus, alongside data from the teacher interviews. The pupils used a wide range of features to navigate knowledge claims, draw on extra-textual material, and anticipate reader reactions. These features seemed to be used in ways that reflected the communicative purpose of the target genre. The findings also indicated that the pupils sometimes used boosters inappropriately, which suggests pupils at this level may benefit from explicit instruction in the appropriate use of these features. The interviews revealed that the teachers offered advice regarding epistemic stance and engagement features, but this was sometimes inconsequential, categorical, or outdated. Article 4 reports results from an analysis of attitude markers in 135 essays collected from the Norwegian, Swedish and British schools. For this study, 218 attitude markers belonging to four sub-categories were used to scan the corpus. The results revealed the wide range of types that the pupils used to offer their affective evaluations of the material in question and how these varied across the educational contexts and genres. While many other metadiscourse features seemed to be used similarly across the educational contexts, attitude markers were more varied and frequent among pupils in the UK. This may be explained by several factors, such as the UK pupils having a broader lexical vocabulary, or that the UK genres required pupils to more frequently offer their affective reactions. Overall, these articles offer insight into the wide range of linguistic features that pupils rely on to signal textual relations, negotiate knowledge claims, engage readers, and express attitudes. On the one hand, many of these features seemed to be used at relatively similar frequencies across the three educational contexts. This might demonstrate the seemingly high written proficiency of L2 learners of English in Norway and Sweden. Alternatively, this may partly be due to the linguistic similarities of Norwegian, Swedish and English, enabling the Scandinavian upper secondary pupils to directly transfer many metadiscourse features from their L1s to English with relative success. On the other hand, metadiscourse usage seemed to reflect the communicative purposes of the target genres. Other factors may also have influenced the pupils’ metadiscourse usage, such as teacher advice, essay writing prompts, and individual preferences. The interview data suggest that the teachers tended to offer advice that was somewhat disconnected from professional writing practices, which consequently requires further investigation. This project contributes to the field by offering insight into the types, frequencies and usage of metadiscourse features among this under- researched group. The analysis required compiling an adapted taxonomy that accounts for the idiosyncratic types and sub-categories that characterised the corpus, which provides a comprehensive starting point for future studies that aim to investigate how metadiscourse features are used in pre-tertiary educational contexts. A further major contribution of this project is the use of interview methods to investigate teacher views regarding metadiscourse-related instruction. The findings have implications for teachers who aim to develop their pupils’ pragmatic knowledge of how signposts and stance markers vary across different genres. Engaging pupils in writing essays of varying lengths across a range of genres can contribute to preparing them for the written demands they are likely to face in higher education and among professional discourse communities.
