1. INTRODUCTION

Studies in applied linguistics have contributed to understanding what areas most in need of development regarding the English language academic skills of Hungarian learners are (e.g., Csizér & Tankó, 2017; Kurtán & Silye, 2006); nevertheless, in academic discourse, reference as a cohesive device has been shown by previous research (e.g., on EFL undergraduates: Chen, 2008; Finnish: Mauranen, 1990; Thai: Indrasuta, 1988) to be particularly problematic for non-native speakers of English. It can be assumed that Hungarian learners also experience difficulties because of differences in the English and Hungarian language systems; however, there are relatively few corpus-based, discourse-level analyses of Hungarian advanced learners’ use of reference as a cohesive device. While many types of cohesive items are easily counted by text analysis software, mechanical word-count will not lead to insights into the complex phenomenon of reference. References weave the text together to make it more cohesive and comprehensible, or, if improperly structured, unintelligible. Consequently, the skillful use of referential cohesion is an essential component of discourse competence, and as such, it plays a major role in the teaching and learning process for both EFL and ESP.

In order to formulate recommendations for the development of relevant aspects of EFL discourse competence within the context of academic writing, this study aims to describe how EFL learners’ use of reference compares to that of expert writers (i.e., experienced writers of English research articles). For this purpose, first, the construction of an analytical tool for the analysis of reference as a cohesive device needed to be devised, as out of the many analytical tools available for the analysis of cohesion, none of them can be applied to reference. The analysis cannot be fully automated; the resolution of the anaphor of a reference item needs human thought processes drawing on relevant experience and background knowledge for the interpretation of the writer’s intentions, and for identifying ambiguities. The development of the analytical tool for reference analysis is achieved by a multi-stage investigation involving both quantitative and qualitative approaches. At the final stage, the newly developed tool is used to carry out comparative analysis of two corpora of texts (research articles (RAs) written by experts and Master’s (MA) theses written by English as a Foreign Language (EFL) students). The comparison demonstrates how reference operates at various textual levels to establish connections between text segments, or when misused, may obstruct text comprehension, or render portions of a text inaccessible.

Three main types of information are obtained using the proposed analytical tool:

To be able to answer this question, first an analytical tool for the study of referential cohesion will be devised that is exempt from the weaknesses of the analytical instruments proposed so far. The second part of the research will then use this instrument to first justify its reliability and validity, and then to describe the use of referential cohesion by Hungarian novice EFL writers in comparison to that of expert writers. The pedagogical implications of the study inform the instruction of EFL and English for Academic Purposes (EAP) writing.

The present study therefore follows a mixed research design consisting of several methods of enquiry: quantitative, qualitative and theoretical. The study is qualitative in its approach in that it “is designed to discover” (Maykut & Morehouse, 1994, 43) textual patterns, and only partially relies on pre-existing taxonomies and methods. Its qualitative nature is also reflected in the fact that some definitions (e.g., reference error) are reached through in-depth data analysis (Fraenkel & Wallen, 1993). Moreover, the study relies partly on time, investigator, and methodology triangulation to ensure its validity and reliability (Cohen et al., 2000). Qualitative methods are used to systematize emerging patterns from the quantitative linguistic data analysis procedure and as a result, to provide input for subsequent stages of research. In addition, the stability of quantifiable corpus data (e.g., number and types of cohesive ties) is measured by comparing findings for each text in the corpus to each other. Those texts that differed markedly (i.e., theoretical papers and case studies) from the other texts in the corpus in some of their characteristics were replaced by more fitting items, so that they would not distort the data. Besides, intra- and inter-rater reliability was used to check the consistency of the results. This empirical procedure will go hand in hand with the reconsideration of the theoretical foundations of the analysis of reference.

First, a more general theoretical background will be established by reviewing approaches to analyzing written discourse. Following the definition of key concepts related to cohesion analysis of academic written discourse, some analytical frameworks will be introduced, and then teaching-related issues will be highlighted. Chapter 3 more narrowly discusses cohesion and cohesive reference in particular, with an initial review of methods of cohesion analysis that the remaining part of the study builds on.

The research design itself consists of five interrelated stages (Chapters 4–9) that together lead to the development of an analytical tool for patterns of referential cohesion in academic discourse. The closing chapters (Chapters 10–11) present overall results, including the pedagogical implications of the study, its limitations and suggested areas for future research.