EAP vocabulary in native and learner writing : from extraction to analysis : a phraseology-oriented approach

Paquot, Magali [UCL]

This thesis deals with the phraseology of English for Academic Purposes (EAP) vocabulary in learner writing. The seven chapters of the thesis are organised as follows: Chapter One aims to characterize English for Academic Purposes. It gives a general account of its distinctive linguistic features before focusing on vocabulary needs in academic settings. A distinction is made between receptive and productive vocabulary and the term ‘academic vocabulary' is defined in the light of its particular nature and role in academic discourse. It then offers a review of corpus-based studies of vocabulary in professional academic discourse, learner writing and native student writing. Chapter Two deals with the fuzzy boundaries of the phraseological spectrum. It successively addresses the questions of the categorization of phrasemes and their defining criteria before presenting the typology and definitions adopted in this thesis. It then sheds light on the dual nature of the term "collocation" and argues that "to continue to thrive", the phraseological approach and the distributional approach to collocation will need to agree on a common terminology. Finally, it focuses on the relationship between discourse and phraseology as phrasemes with specific rhetorical functions have been reported to be typical of academic discourse. Chapter Three reviews major findings about the influence of the first language on single words and phrasemes in the foreign language. It then focuses on some methodological flaws of transfer studies which cast doubt on the validity of several of the findings discussed in the first part of the chapter. Chapter Four is devoted to a detailed description of the corpora, methods and software programs used in this thesis. We make use of corpora of professional academic prose, learner writing and native student writing and compare them within the framework of the Integrated Contrastive Model. We adopt a quantitative approach to phraseology and describe the extraction procedure used after discussing a number of parameters that can influence the results of a co-occurrence analysis. Chapter Five is dedicated to the selection of words that are typical of academic texts and which will provide a basis for the comparison of native speaker and EFL learner academic writing. It proposes a new methodology based on the criteria of keyness, range and evenness of distribution. Chapter Six aims to test our working hypothesis that upper-intermediate to advanced EFL learners, irrespective of their mother tongue backgrounds, share a number of linguistic features that characterize their academic writing. It focuses on words and phrasemes that expert writers and EFL learners use to serve typical organizational or rhetorical functions. The function of exemplification is presented in detail so as to serve as an illustration of the type of data and results obtained when examining the whole range of lexical strategies available to EFL learners, as opposed to expert writers, when they want to establish cohesive links in their essays. Other functions were similarly described but the numerous analyses conducted are not presented in detail in this thesis as they would become tedious for the reader. Instead, the focus is placed on the general interlanguage features that emerge from these analyses, which fall into five categories: aspects of overuse and underuse, register-awareness, phraseological patterns, semantic misuse and sentence position. Important pedagogical implications related to teaching practices and the role of corpora in materials design are then examined. Chapter Seven is largely methodological in nature. It first seeks to demonstrate how learner corpus data can be used to select interlanguage features worthy of investigation in transfer studies. It then tries to implement Jarvis's (2000) unified framework to investigate L1 influence on the basis of corpus data and investigates their potential to uncover new types of evidence of transfer. It proposes to employ statistical techniques based on comparisons of means - ANOVA tests and the Dunnett post-hoc test - to operationalize the first two effects of L1 influence described in the framework. The major advantages and limitations of the approach are highlighted. The thesis ends with a general conclusion, which briefly summarizes the major findings and offers several avenues for future research.