how definite are we about the english article system? chinese

HOW DEFINITE ARE WE ABOUT THE ENGLISH ARTICLE SYSTEM? CHINESE

LEARNERS, L1 INTERFERENCE AND THE TEACHING OF ARTICLES IN

ENGLISH FOR ACADEMIC PURPOSES PROGRAMMES.

RICHARD NICKALLS

A thesis submitted to

The University of Birmingham

for the degree of

DOCTOR OF PHILOSOPHY

Department of English Language and Applied

Linguistics

School of English, Drama and American &

Canadian Studies

College of Arts and Law

The University of Birmingham

December 2019

University of Birmingham Research Archive

e-theses repository

This unpublished thesis/dissertation is copyright of the author and/or third

parties. The intellectual property rights of the author or third parties in respect

of this work are as defined by The Copyright Designs and Patents Act 1988 or

as modified by any successor legislation.

Any use made of information contained in this thesis/dissertation must be in

accordance with that legislation and must be properly acknowledged. Further

distribution or reproduction in any format is prohibited without the permission

of the copyright holder.

ABSTRACT

Omission and overspecification of the/a/an/Ø are among the most frequently occurring

grammatical errors made in English academic writing by Chinese first language (L1) university

students (Chuang & Nesi, 2006; Lee & Chen, 2009). However, in the context of competing

demands in the English for academic purposes (EAP) syllabus and conflicting evidence about

the effectiveness of error correction, EAP tutors are often unsure about whether article use

should or could be a focus and whether such errors should be corrected or ignored. With the

aim of informing pedagogy, this study investigates: whether explicit teaching or correction

improves accuracy; which article uses present the most challenges for Chinese students; the

causes of error and whether a focus on article form can be integrated within a modern genre

based/student centred approach in EAP. First, a questionnaire survey investigates how EAP

teachers in higher education explicitly teach or correct English article use. Second, the effect

of explicit teaching and correction on English article accuracy is investigated in a longitudinal

experiment with a control group. Analysis of this study’s post-study measures raise questions

about the sustained benefits of written correction or decontextualised rule-based approaches.

Third, findings are presented from a corpus-based study which includes an inductive and

deductive analysis of the errors made by Chinese students. Finally, in a fourth study hypotheses

are tested using a multiple-choice test (n=455) and the main findings are presented: 1) that

general referential article accuracy is significantly affected by proficiency level, genre and

students’ familiarity with the topic; 2) Chinese students are most challenged by generic and

non-referential contexts of use which may be partly attributable to the lack of positive L1

transfer effects; 3) overspecification of definite articles is a frequent problem that sometimes

gives Chinese B2 level students’ writing an ‘informal tone’; and 4) higher nominal density of

pre-qualified noun phrases in academic writing is significantly associated with higher error

rates. Several practical recommendations are presented which integrate an occasional focus on

article form with whole text teaching, autonomous proofreading skills, register awareness, and

genre-based approaches to EAP pedagogy.

DEDICATION

This thesis is dedicated:

To my loving family, past and present.

ACKNOWLEDGEMENTS

At the University of Birmingham, I am profoundly grateful to Professor Jeannette Littlemore

and Dr. Nicholas Groom for helping me through my Masters and then later supervising me at

Doctoral level. My thesis (perhaps even more so than most) has ended up a long way from

where we started (the proposal they agreed to supervise) and I’m very grateful for their

tolerance and continued interest. I need to express my sincere gratitude to the many students

who have participated in various studies and my many colleagues who have contributed to this

research.

There’s too little space to mention every single academic, student and colleague who has

generously helped me or simply exchanged ideas on this rather dry topic – but they know who

they are. A special thanks to Jane Sjoberg and my nephew Oliver for checking my thesis for

typos and inconsistencies. Thanks to Dr. Oliver Mason for use of his Java-based programme

to extract noun phrases and to Dr. Laurence Anthony of Waseda University for programming

(and regularly updating) the freely available Ant. Conc. Programme. For help in understanding

aspects of the Chinese language I must finally thank Zhijun Wang, Shimeng Liu and Xinyu Li.

i 
 
 
Table of Contents 
INTRODUCTION ......................................................................................................... 1 
1  Brief overview ........................................................................................................ 1 
2  Personal interest ...................................................................................................... 2 
3  Background and justification for the research .......................................................... 4 
4  Previous Research into Chinese L1 learners’ English article accuracy ..................... 8 
5  Summary of research aims .................................................................................... 10 
6  Thesis overview .................................................................................................... 11 
LITERATURE REVIEW ............................................................................................ 15 
1  Key term definitions.............................................................................................. 15 
2  The English article system .................................................................................... 17 
3  Contrastive Analysis ............................................................................................. 26 
4  Empirical research into Chinese L1 learners’ written L2 English errors ................. 32 
5  Other factors influencing English Accuracy .......................................................... 41 
6  English articles within the current EAP pedagogical paradigms............................. 49 
7  Empirical research into the effect of teaching and correcting articles ..................... 52 
8  Research Questions ............................................................................................... 55 

iii 
 
METHODOLOGY ...................................................................................................... 59 
1  Critical methodological issues in Error Analysis ................................................... 59 
2  Error detection ...................................................................................................... 62 
3  Error classification ................................................................................................ 65 
4  Error explanation .................................................................................................. 66 
5  Evaluation of recent methods in English article studies ......................................... 67 
6  The need for a preliminary study ........................................................................... 75 
7  Preliminary study .................................................................................................. 76 
8  Summary of methodological approaches in main studies ....................................... 84 
9  Conclusion ............................................................................................................ 85 
SURVEY OF TEACHER ATTTITUDES AND PRACTICES .................................... 87 
1  Background and recap of Research Question 1 ...................................................... 88 
2  Methodology ......................................................................................................... 91 
3  Results and Discussion .......................................................................................... 96 
4  Conclusion .......................................................................................................... 122 
   

v 
 
 
LONGITUDINAL STUDY OF THE EFFECT OF TEACHING/ CORRECTION ON 24 
(L1 MANDARIN) LEARNERS OF ENGLISH ON A 15-WEEK PRESESSIONAL ENGLISH 
COURSE .......................................................................................................................... 127 
1  Study overview ................................................................................................... 128 
2  Methodology ....................................................................................................... 129 
3  Ethical issues ...................................................................................................... 143 
4  Constraints of the study ....................................................................................... 145 
5  Data analysis ....................................................................................................... 145 
6  Tagging reliability checks and inter-rater reliability checks ................................. 151 
7  Results and discussion......................................................................................... 154 
8  Conclusion .......................................................................................................... 172 
CONTRASTIVE ANALYSIS OF CORPORA .......................................................... 177 
1  Study overview ................................................................................................... 177 
2  Methodology ....................................................................................................... 179 
3  Data analysis ....................................................................................................... 184 
4  Results and discussion......................................................................................... 186 
5  L1 transfer effects ............................................................................................... 211 

vii 
 
 
6  Linguistic context of noun phrases ...................................................................... 212 
7  Conclusion .......................................................................................................... 215 
TESTING  OF  HYPOTHESES  USING  ONLINE  MULTIPLE-CHOICE 
GRAMMATICALITY JUDGEMENT TEST .................................................................... 221 
1  Study overview ................................................................................................... 223 
2  Methodology ....................................................................................................... 225 
3  Findings and discussion ...................................................................................... 235 
4  Conclusion .......................................................................................................... 242 
CONCLUSIONS ....................................................................................................... 247 
1  Restatement of aims ............................................................................................ 247 
2  Summary of main findings .................................................................................. 247 
3  Limitations and future research ........................................................................... 258 
4  Recommendations for EAP instructors ................................................................ 259 
5  Final comments ................................................................................................... 262 
REFERENCES ................................................................................................................. 267 
APPENDICES .................................................................................................................. 279 
APPENDIX 1 ................................................................................................................ 279 

viii

ix 
 
 
APPENDIX 2 ................................................................................................................ 281 
APPENDIX 3 ................................................................................................................ 293 
APPENDIX 4 ................................................................................................................ 303 
APPENDIX 5 ................................................................................................................ 311 
APPENDIX 6 ................................................................................................................ 318 
APPENDIX 7 ................................................................................................................ 322 
APPENDIX 8 ................................................................................................................ 324 
APPENDIX 9 ................................................................................................................ 334 
APPENDIX 10 .............................................................................................................. 341 
 

Abbreviations

In this thesis, the following abbreviations are used:

BALEAP

BIA

CEFR

DELTA

EAP

EFL

EGAP

EISU

ELAL

ESOL

ESP

IELTS

OED

SELT

UKVI

British Association of Lecturers of English for Academic Purposes

Birmingham International Academy

Common European Framework of Reference for Languages

Diploma in Teaching English to Speakers of Other Languages

English for Academic Purposes

English as a Foreign Language

English for General Academic Purposes

English for International Students Unit (Former name of BIA)

Department of English Language and Applied Linguistics

English for Speakers of Other Languages

English for Specific Purposes

International English Language Testing System

First Language (once called ‘native language’)

Second Language

Oxford English Dictionary

Secure English Language Test

UK Visas and Immigration

xii

xiii 
 
Table of Figures 
Figure 1: Examples of generic NPs in all forms (Krifka & Gerstner, 1987).......................... 20 
Figure 2: Further examples of non-referential noun phrases from Ekiert (2004)................... 22 
Figure 3: Comparison of means of missed the by category (Liu and Gleason, 2002: 13) ...... 33 
Figure 4: Example of definite article ‘overuse’ from Lee & Chen (2009: 287) ..................... 35 
Figure 5: Reproduced from  Chuang and Nesi (2006: 257): 3 stages of developing a tagging 
system ................................................................................................................................. 75 
Figure 6: Summary of mixed method approach to Error Analysis ........................................ 85 
Figure 7: Survey Question 1 ................................................................................................ 93 
Figure 8: EISU EAP Tutor Responses to Question 1 (n=12) ................................................ 97 
Figure 9: BALEAP mailing list responses to Question 1 (n=127) ........................................ 97 
Figure 10: ELAL MA Distance programme responses to Question 1 (n=53) ....................... 98 
Figure 11: EISU EAP Tutors’ responses to Question 2 (n=12)............................................. 99 
Figure 12: BALEAP list EAP Tutors’ responses to Question 2 (n=127)............................... 99 
Figure 13: ELAL MA Distance programme responses to Question 2 (n=53) ..................... 100 
Figure 14: Question 3 prompt ............................................................................................ 101 
Figure 15: EISU EAP Tutors Question 3 (in class) n=12 ................................................... 102 
Figure 16: BALEAP EAP Tutors Question 3 (in class) n=120 ........................................... 102 

xiv

xv 
 
 
Figure 17: ELAL MA Distance programme Question 3 (in class correction) n=52 ............ 102 
Figure 18: EISU  Tutors Question 4 (weekly homework) n=12 ......................................... 103 
Figure 19: BALEAP EAP Tutors Question 3 (weekly homework) n=127 .......................... 103 
Figure 20: ELAL MA Distance Programme Tutors Q.3 (weekly homework) n=53 ............ 104 
Figure 21: EISU  Tutors Question 3 (in first draft)  n=12................................................... 104 
Figure 22: BALEAP EAP Tutors Question 3 (in first draft) n=127 .................................... 105 
Figure 23: MA list tutors  Question 3 (in  first draft) n=52 ................................................. 105 
Figure 24: BIA EAP Tutors  Question 3 (conference paper) n=12 ..................................... 105 
Figure 25: BALEAP EAP Tutors Question 3 (conference paper) n=127 ............................ 106 
Figure 26: MA List Tutors Question 3 (conference paper) n=52 ........................................ 106 
Figure 27: Transcribed extract Tutor #1 ............................................................................ 108 
Figure 28: Transcribed extract Tutor #2 ............................................................................ 110 
Figure 29: Transcribed extract Tutor #3 ............................................................................ 112 
Figure 30: Transcribed extract Tutor #4 ............................................................................ 114 
Figure 31: Transcribed extract Tutor #5 ............................................................................ 115 
Figure 32: Tutor #5 continued ........................................................................................... 116 
   

xvi

xvii 
 
Figure 33: BALEAP list pedagogical recommendations (number of mentions, n=101) ...... 117 
Figure 34: Transcribed extract Tutor #6 ............................................................................ 118 
Figure 35: Reasons given for not teaching articles explicitly (n=101) ................................ 121 
Figure 36: Three group experimental design ...................................................................... 129 
Figure 37: Policy of tagging for correctness ...................................................................... 148 
Figure 38:  Example 1: Identification of error problems .................................................... 149 
Figure 39: Example 2: Overt errors ................................................................................... 150 
Figure 40: Example 2 from annotator instruction materials ................................................ 152 
Figure 41: Oversuppiance issue disagreements .................................................................. 153 
Figure 42: Appropriacy for genre disagreements ............................................................... 153 
Figure 43: Disagreements about whether Type 3 or Type 4 ............................................... 154 
Figure 44: Box plot showing distributions of median TLU scores  (n=24) ......................... 154 
Figure 45: Target-Like Use of definite article (Type 1, 2 and  5 combined) ....................... 157 
Figure 46: Target-Like Use of Ø (Types 1, 3, 4 and 5 combined) ...................................... 158 
Figure 47: Target-Like Use of a/an (Types 1, 3 and 5 combined) ...................................... 158 
Figure 48: TLU of 2DA definite article (n=24, all participants by group) .......................... 159 
Figure 49: Progress by three groups (Mdn TLU of Ø) n=24 .............................................. 164 
   

xviii

xix 
 
Figure 50: Bar chart – omission and oversuppliance of definite articles in Task 1 .............. 166 
Figure 51: Bar chart – omission and oversuppliance of definite articles in Task 2 .............. 166 
Figure 52: Participant responses to Survey Question 3 (n=23) ........................................... 169 
Figure 53: Participant responses to Survey Question 4 (n=23) ........................................... 170 
Figure 54: Tutor examples................................................................................................. 189 
Figure 55: Student examples ............................................................................................. 189 
Figure 56: Median TLU scores in the three article types (3 prompts, n =24). ..................... 191 
Figure 57: 2GADA errors compared to 2DA correct use ................................................... 197 
Figure 58: 4GAZA errors compared to 4ZA correct use .................................................... 200 
Figure 59: 4GAIA errors compared to 4IA correct use ...................................................... 202 
Figure 60: 3GAIA errors compared to 3IA correct use ...................................................... 204 
Figure 61: 5GADA errors compared to 5DA correct use ................................................... 206 
Figure 62: 1GADA errors compared to 1DA correct use ................................................... 207 
Figure 63: 3GAZA errors compared to 3ZA correct use .................................................... 208 
Figure 64: 1GAIA errors compared to 1IA correct use ...................................................... 210 
Figure 65: 5GAIA errors compared to 5IA correct use ...................................................... 211 
Figure 66: Omissions of article with nouns pre-qualified with proper nouns ...................... 213 
   

xxi 
 
Figure 67: Typical omission and oversuppliance errors  (with pre-modified head nouns) ... 214 
Figure 68: Example of question development (Question 10) .............................................. 227 
Figure 69: Example of multiple dropdown question .......................................................... 228 
Figure 70: Paired noun phrase examples ............................................................................ 229 
Figure 71: Histogram showing overall IELTs scores ......................................................... 233 
Figure 72: Histogram showing IELTS writing test scores .................................................. 234 
Figure 73: Histogram of test scores ................................................................................... 235 
Figure 74: Oversuppliance (ten questions) box-plot (n=455) ............................................. 237 
Figure 75: Boxplot of omission rates (grouped by proficiency) in complex noun phrases .. 240 
Figure 76: The relationship between general writing ability and overall test scores (n=455, 100 
question items) .................................................................................................................. 242 
Figure 77: Mapping specific-general functions of lexis to research project structure .......... 261 
Figure 78: Deduced uses by Chinese L1 Engineering student [anonymised] ...................... 262 
Figure 79: Student 1, conversation  #1 ............................................................................... 324 
Figure 80: Student 1, conversation #2................................................................................ 325 
Figure 81: Student 1, conversation #3................................................................................ 325 
Figure 82: Student 1,  conversation #4 ............................................................................... 326 
Figure 83: Student 1, conversation #5................................................................................ 326 

xxii

xxiii 
 
 
Figure 84: Student 1, conversation #6................................................................................ 327 
Figure 85: Student 2, conversation #1................................................................................ 328 
Figure 86: student 2,  conversation 7 ................................................................................. 328 
Figure 87: Student 2, conversation 8 ................................................................................. 329 
Figure 88: Student 2, conversation  9................................................................................. 329 
Figure 89: Student 2,  conversation 10 ............................................................................... 330 
Figure 90: Student 3, conversation 11................................................................................ 331 
Figure 91: Student 3, conversation 12................................................................................ 332 
Figure 92: Student 3 conversation 13................................................................................. 332 
Figure 93: Distribution of written placement test scores  (Chapter 5) ................................. 341 
Figure 94: Distribution of student scores in the placement test........................................... 341 
Figure 95: Test of normality (Q scores) ............................................................................. 342 
Figure 96: Tests of Normality (Detrended Q scores) .......................................................... 343 
 
   

xxiv

xxv 
 
Table of Tables 
Table 1: Bickerton’s (1981) semantic and pragmatic Framework ........................................ 18 
Table 2: A comparison of important corpus findings ........................................................... 41 
Table 3: Error Analysis Steps/common methodological issues ............................................ 60 
Table 4: Review of influential article acquisition/corpus-based studies ................................ 68 
Table 5: Five learner corpus studies compared .................................................................... 72 
Table 6: IELTS score descriptions (from IELTS, 2012) ....................................................... 82 
Table 7: Response rates ....................................................................................................... 95 
Table 8: Participant information ........................................................................................ 133 
Table 9: Essay prompts ..................................................................................................... 136 
Table 10: BME Teaching materials ................................................................................... 139 
Table 11: Five extra English article sessions...................................................................... 141 
Table 12: Tagging system for correct uses (Díez-Bedmar & Papp, 2008) .......................... 146 
Table 13: Tagging system for incorrect uses (Díez-Bedmar & Papp, 2008) ....................... 146 
Table 14: TLU ratings for all groups summarised (n=24) with no adjustment .................... 155 
Table 15:TLU ratings all groups (n=24) with adjustments* for zero counts ....................... 156 
Table 16: Wilcoxon Signed Rank Test  *=statistically significant at 0.05 .......................... 160 
   

xxvi

xxvii 
 
Table 17: Omission and Oversuppliance of definite articles (all types, Mdn coefficients*) 165 
Table 18: Misuse of article errors ...................................................................................... 167 
Table 19: Participant information (including borderline B1/B2 students). .......................... 181 
Table 20: Four Task prompts ............................................................................................. 183 
Table 21: Framework adapted from Bickerton’s (1981) semantic and pragmatic uses ........ 185 
Table 22: Proportion of errors by context .......................................................................... 187 
Table 23: L1 writers (n=5)  article uses compared to Chinese L1 essays (n=24) ................ 188 
Table 24: Overall TLU  (all essays) in all contexts (n=24) ................................................. 192 
Table 25: TLU is essay  prompts compared to case study prompt ...................................... 194 
Table 26: The effect of pre-modification ........................................................................... 199 
Table 27: Density of noun phrases (in 610 article errors) ................................................... 212 
Table 28: Analysis of 9186 correct noun phrases ............................................................... 212 
Table 29: Results of ten questions investigating oversuppliance (n=455) ........................... 236 
Table 30: Articles omitted in twenty nominal groups (with and without pre-modification) 239 
   

xxviii

1 INTRODUCTION

1.1 Brief overview

This PhD research stems from an interest in one of the most confusing and frequent

grammatical choices that learners of English must make: whether to use the, a/an, or no article

(Ø) in written production. The is by far the most commonly occurring word in the English

language (Sinclair, 1991) and, since the most frequent choice is not to use any determiner, the

Ø article has been argued to be the most frequently occurring free morpheme (Master, 1997).

On the assumption that accuracy is considered important, a focus on these small words in

English for Academic Purpose (EAP) classes appears necessary, as it has long been pointed

out that the/a/an together account for more than one in every ten words in academic writing

(Berry, 1991).

While the complexities of the English article system are likely to challenge all international

students whose first language is not English, the literature suggests that students from a first

language (L1) background that does not grammaticalise definiteness/specificity may face

particular challenges in applying certain aspects of the English article system (Master, 1997;

Lu, 2001; Ekiert, 2004; Díez-Bedmar & Papp, 2008; Snape, Mayo & Gürel, 2013; Crosthwaite,

2012, 2016a). The extent to which Chinese Mandarin grammaticalises these notions will be

discussed in Chapter 2, but some researchers have claimed that Chinese learners’ first language

affects their use of the/a/an/Ø in academic writing at university at Upper-Intermediate levels

of English proficiency (Milton, 2001; Chuang & Nesi, 2006; Díez-Bedmar & Papp, 2008; Lee

& Chen 2009). Indeed, it has been suggested that article errors (omission and overspecification

of the definite article) are the most commonly occurring grammatical errors made in English

academic writing by Chinese L1 undergraduate university students (Chuang & Nesi, 2006).

The increase in the numbers of Chinese L1 students in UK Higher Education over the past 15

years (see Section 1.3.1) has further heightened the need to understand these students’

challenges with English articles in academic writing.

Having briefly outlined the topic, this chapter now has four aims: to explain my personal and

professional interest in this research, to justify this interest in academic terms, to highlight why

further research is necessary and to outline the structure of this thesis.

1.2 Personal interest

My curiosity regarding EAP students’ article usage can first be explained anecdotally. In my

observations of over 30 EAP teachers in my role coordinating an English for Academic

Purposes Presessional at a UK university (some of whom were new entrants to EAP from the

fields of English as a Foreign Language and English as a Second Language), it has been my

personal experience that the English article system can be taught effectively by EAP tutors. As

will be illustrated in Chapter 4, which presents a survey of English for Academic Purposes

(EAP) tutors’ beliefs and practices regarding the teaching of English articles, many tutors

manage to focus students on accurate article use in authentic academic texts in ways which

motivate students and promote learner autonomy. However, my experience over 14 years of

teaching EAP suggests that some tutors are unsure about whether they can and how they should

improve students’ accuracy in English articles. Can and should they teach article use? If so,

which aspects of the article system challenge Chinese L1 learners most? Should they correct

article errors in academic writing feedback and does this help learners or simply discourage

them?

Unfortunately, some EAP tutors have not stopped to even ask the questions above. In a

depressingly similar fashion to my own experiences as a learner of second languages, I have

also sometimes noticed students being overcorrected and discouraged, taught rules they already

learned in middle school, made to complete decontextualised grammar gap-fill exercises that

may have little effect on written production, told oversimplified rules and even provided with

incorrect advice and feedback.

On the opposite end of the spectrum, some tutors totally avoid focussing on the English article

system in EAP. As discussed in Chapter 4, possible causes can include a tutor’s lack of

confidence of their own understanding of the article system, some tutors’ belief that students

can naturally improve article accuracy without intervention, a principled objection to focussing

on form within EAP, or a perception that article errors are trivial surface errors. Irrespective

of the tutors’ beliefs, many also perceive that they have insufficient time to devote to this minor

error in the typical EAP curriculum in which the need for accuracy of articles has to compete

with many more important priorities. Moreover, given that genre analysis, register awareness,

and many competing lexico-grammatical areas are simpler to teach and have proven efficacy,

many tutors need convincing that a focus on article accuracy could and should merit occasional

attention in the EAP classroom.

Although the personal anecdotal evidence presented above is not proposed to support the

claims made in this thesis, these experiences underpin my personal and professional motivation

for addressing the topic. The following section will present the more academic background to

this issue before summarising the aims of this thesis: to give EAP tutors greater clarity on

whether they can and how they can facilitate improved English article accuracy among Chinese

learners.

1.3 Background and justification for the research

1.3.1 Justification for the focus on Chinese L1 Postgraduate level students

The focus on the English article use of Chinese L1 international students at postgraduate level

is justified due to the massive expansion of Chinese L1 students studying in many UK HE

institutions over the last 15 years (Universities UK, 2019). According to this source, 19.6% of

the total UK student population in 2017/18 were international students making a net worth

contribution to the UK of £20.3bn. Of these 458,490 international students, 106,530 came

from China and this figure has been rising steadily since 2013. Over 70% of students in the

summer Presessional English for Academic Purposes Programme at the author’s own

institution are Chinese L1 students who progress onto postgraduate programmes in which they

often outnumber home students, notably in the university’s Business and Engineering Schools.

This reflects broader national trends, for which current data indicate that the majority of

postgraduate Chinese speaking students enrol onto Business studies, Economics, Accounting

and Finance, and Engineering programmes at UK universities (The Higher Education Statistics

Agency, 2019). It thus seemed appropriate to focus on Chinese L1 Business School students in

the studies presented in Chapter 5 and 6. These official UK statistics may already be out of date

according to a recent Guardian Newspaper report (Weale, 2019) which suggests that

applications for study in the UK 2019/20 have increased by 30% due to China-US tensions. It

is therefore more relevant than ever to understand any L1 effects upon language inaccuracy.

The question of whether Chinese students could be assisted to improve their English article

accuracy is also related to the instruction provided within the university setting and the

following section therefore provides the definition and background of English for Academic

Purposes (EAP).

1.3.2 English for Academic Purposes (EAP)

Rather than being defined simply as English lessons in a university setting, English for

Academic Purposes (EAP) is generally argued to be a separate paradigm of instruction (Swales,

1990; Flowerdew & Peacock, 2001; Hyland & Hamp-Lyons, 2002; Alexander, Argent &

Spencer, 2008). Linked to the greater use of English across the world by non-native speakers

of English (Flowerdew & Peacock, 2001), the proliferation and expansion of EAP courses in

the UK is also related to the internationalisation strategies of many UK universities. In the UK

context, the term EAP covers a broad range of activities including highly specific courses

embedded within disciplines; presessional courses designed for students who have not reached

the English proficiency entry requirements and need to show some progress before starting

their intended academic programme; insessional classes covering a ‘core skills’ approach

offered to all students during their programme, and new forms of integrated academic English

skills and content instruction such as Foundation programmes. The core defining feature of

any EAP course is that the objectives are needs-focussed with the aim to help international

students achieve success in their intended academic community (Hyland & Hamp-Lyons,

2002).

Any discussion about whether English article accuracy can be a focus in the EAP curriculum

needs to recognise the competing demands for any EAP tutor’s time in the classroom. As will

be developed in Chapter 2, a focus on language forms such as the/a/an/Ø needs to be integrated

within a curriculum which has an arguably greater priority of preparing students for the

functions and processes of academic writing. While most EAP Presessional courses offer a

mixture of general and academic English, EAP is often defined by its more dominant focus

on the socialisation of students within their discourse communities which involves special

attention to the genres of writing that students will be producing in these communities (Swales,

1990; Swales, 2004; Swales & Feak, 2004) and the many variances in genre across disciplines

(Hyland, 2002a, 2007b; Nesi & Gardner, 2012). In contrast to the general English syllabus

which prepares students for English language demands in their entirety (an integrated syllabus

which tends to recycle grammatical accuracy issues over shorter texts of various topics), the

‘EAP paradigm’ involves more time restricted programmes on academic English more

focussed on writing and reading (Alexander, Argent, & Spencer, 2008). This generally entails

deeper exploitation of a smaller number of denser academic texts (ibid.). With regards to

grammatical features such as English articles, an ‘EAP paradigm’ could thus be expected to

include such a focus more incidentally than general English, as and when needed, if at all

(reference will be made to the researcher’s programme in Chapter 5).

While acknowledging that some researchers continue to see accuracy in articles as acquired

naturalistically without instruction (Alexander, Argent, & Spencer, 2008), this thesis makes the

assumption that a focus on form in language generally accelerates accuracy. In EAP’s related

sister field of Second Language Acquisition (SLA) the consensus view for the past 35 years

has rejected theories of naturalistic L2 language learning of the written form of L2 (Long, 1983,

1990; Norris & Ortega, 2000; Ellis, 2006). The generally accepted view in SLA research is that

students need both content-driven ‘comprehensible input’ (Dulay, Burt & Krashen, 1982;

Nunan, 1991) in addition to a focus on form that helps learners notice patterns more

consciously (Norris and Ortega, 2000; Ellis, 2006). However, the assumption that a focus on

form has theoretical benefits should not be confused with a traditional decontextualised

approach to grammar teaching. This thesis in no way seeks to argue for a return to the

behaviourist and prescriptively ill-informed approaches of writing composition classes in the

1960s and 1970s (Paltridge, 2001). More realistically, the investigations into the effects of

different methods of teaching articles were carried out on the theoretical assumption that form-

focussed instruction can sometimes contribute to students’ noticing of written accuracy. This

assumption would appear solidly grounded on theory since the total rejection of any focus on

grammar is less common today in the EAP field and many EAP researchers conclude that an

incidental focus on form integrated within an EAP genre-based approach can be beneficial

(Hyland & Milton, 1997; Hinkel, 2003; Fang, Schleppegrell & Cox, 2006; Hyland, 2008).

One unresolved polemic within the EAP literature, more directly affecting the topic of this

research, relates to written error correction. The review of the literature in Chapter 2 will more

critically juxtapose the research that supports focussed written corrective feedback (Ellis, 1994;

Polio, Fleck & Leder, 1998; Ferris, 1999; Chandler, 2003; Bitchener, Young & Cameron, 2005;

Dale, Anisimoff & Narroway, 2012; Ferris & Kurzer, 2019) with research that questions the

value of the labour-intensive process of written corrective feedback (Truscott, 1996, 2004;

Crosthwaite, 2016b). Within EAP, Alexander, Argent, & Spencer (2008: 210) certainly

discourage the correcting of ‘every misuse of articles’. Thus, the practices I have anecdotally

witnessed during summer presessionals over 11 years of teaching and coordinating EAP

courses does not always appear to sit so harmoniously within the recommended practice of

EAP. It was therefore of interest to know why EAP teachers were so frequently seen to be

correcting article use and whether such correction served a purpose.

The justification of this thesis thus follows the practical objectives of past error analyses

(Corder, 1967; Master, 2002; Chuang & Nesi, 2006): to improve pedagogical approaches

through understanding. Through knowing the extent to which different contexts of article use

cause challenges for Chinese L1 learners and the reasons for which mastering some contexts

may prove more difficult, EAP pedagogy and materials development could be refined to better

suit these learners who are so highly represented in student numbers. However, as shall be

reviewed in the next section and the Literature Review, there are gaps in knowledge and

disagreement about the extent to which different article use contexts cause challenges for

Chinese L1 learners and even whether they continue to make frequent errors at Upper-

Intermediate levels of English proficiency.

1.4 Previous Research into Chinese L1 learners’ English article accuracy

The question of whether Chinese L1 university students make the most errors through definite

article omission (Chuang & Nesi, 2006; Díez-Bedmar & Papp, 2008) or through omission of

Ø/oversuppliance of the definite article (Lee & Chen, 2009) will be discussed further in the

next chapter and then investigated using both corpus methods and a testing approach. As the

Literature Review will show, previous studies have arrived at very different conclusions

regarding this question. While some researchers (e.g. Crosthwaite 2016a) have found that most

article contexts pose few challenges for Chinese students by the time they have reached Upper-

Intermediate levels, other researchers (e.g. Lee & Chen 2009) report frequent and meaningful

effects of article inaccuracy among Chinese majors in English and Applied Linguistics with

high levels of English.

As shall be presented in the next chapter, this thesis adopts a weak form of the Contrastive

Analysis Hypothesis. The assumption is made that learners with an L1 containing an article

system [+Article] could have a different ‘natural order of article acquisition’ to learners whose

language has no article system, but that this effect should not be oversimplified (as many other

factors affect acquisition and written accuracy). Master (1997: 216) found high levels of

inaccuracy among higher levels of article-less L1 background EAP students including Chinese

L1 learners and concluded that transfer effects were a large contributor. However, many EAP

tutors are likely to hold an overly simplified working hypothesis about the effect of L1 on their

Chinese students’ accuracy in academic English based on less evidence-based research. Over

many years, one hugely influential book has been Smith and Swan’s (2001) Learner English:

A teacher’s guide to interference and other problems, within which Chang’s (2001: 318)

chapter creates a somewhat negative impression of Chinese L1 writing with a strong

assumption of ubiquitous error through L1 transfer effects in L2 English:

There are no articles in Chinese. Students find it hard to use them consistently correctly. They

may omit necessary articles: *Let’s make fire *I can play piano. Or insert unnecessary ones:

*He finished the school last year. *He was in a pain. Or confuse the use of definite or

indefinite articles: *Xiao Ying is a tallest girl in the class. *He smashed the vase in the rage.

However, Chang provides little evidence to support such claims. In fact, as shall be discussed

in Chapter 2, more empirical recent research suggests that the L1 of Chinese students has a far

more complex and nuanced effect on their accuracy in different contexts of article use (Díez-

Bedmar & Papp, 2008; Crosthwaite, 2016a). These latter empirical findings are extremely

useful, particularly in the context of the competing demands for attention in the EAP syllabus

and the obvious benefit of focussing learners on specific article uses effectively, rather than on

the article system in its entirety. After reviewing this empirical research in Chapter 2, this

study will further investigate these nuanced effects and thereby contribute to the insights such

evidence can give to materials and syllabus designers in EAP.

The issue of whether article errors are minor surface errors that cause only annoyance or can

cause meaningful problems in communication is fundamental to the question of whether tutors

should attempt to teach accurate use of the/a/an/Ø articles in EAP. The issue of error gravity

(Corder, 1967) will thus need to be addressed in several chapters in this thesis. In addition to

discussing the effects of Chinese learners’ article morpheme errors in terms of sentence-level

accuracy, this research will also seek to explore the effect of article inaccuracy on the overall

writing errors that Chinese learners make (including articles, pronouns and adverbs) and the

effects these errors have on the reader’s perceived academic register of writing. A number of

studies have noted informal characteristics in the English academic writing of Chinese L1

university students (Mayor, 2006; Gilquin & Paquot, 2007; Lee & Chen, 2009; Chen, 2014).

Such claims are often refuted (see Leedham, 2015), but there is a possibility that this impression

is related to the observation made in the research that [-Article] L1 background students go

through a ‘flooding’ stage of definite article overuse in their acquisition process (Huebner,

1983; Master, 1997;Thomas, 1989; Zdorenko & Paradis, 2008), which has also been claimed

to be seen in Chinese L1 university students’ writing (Lu, 2001; Lee & Chen, 2009). It is

certainly possible that such effects are often overestimated in a similar way to how many EAP

tutors assume all article errors are due to L1 transfer effects, but this interference is an

interesting focus for research.

1.5 Summary of research aims

The intention is to further the EAP field’s understanding of the types of article use that cause

Chinese learners the most frequent challenges, in order to inform the development of more

targeted pedagogical resources relevant for Chinese learners in EAP. To develop this

pedagogy, a greater understanding is required not only of the errors that Chinese learners make,

but the causes of such errors (including L1 effects) and EAP tutors’ current approaches to

teaching article use. Another aim is therefore to understand the current attitudes and practices

of EAP tutors with regard to teaching and correcting article use. The findings of this research

could then help tutors to decide which article uses to teach or ignore and whether or not they

should correct errors in student writing. With a greater understanding of these article errors,

after briefly focusing learners on article forms, tutors could potentially focus students upon

self-correction and peer-marking which are more consistent with current student-centred

approaches that encourage learner autonomy. By investigating students’ accuracy and

teachers’ attitudes, the thesis inevitably touches on some of the ongoing debates in EAP.

Although not the main objective of this research, this thesis will contribute to the discussion

about whether an occasional focus on accuracy (morpheme choice) is fully compatible with

current approaches to teaching EAP recommended in the literature.

1.6 Thesis overview

Chapter 2

After reviewing the English article system and the effect of article errors in writing, a

contrastive analysis of Chinese Mandarin will be presented. The review of the research

literature will suggest some limited support for the L1 effect hypothesis, but also show a highly

complex range of factors affecting development and accuracy with English articles. Having

identified several gaps in the literature, the chapter summarises four Research Questions.

Chapter 3

While the procedural methodologies of the individual studies are presented in their individual

sections (Chapters 4 – 7), a methodological overview is presented in Chapter 3 of some of the

key issues that impact upon the methodological choices made in the course of the research.

The chapter also describes the ethical and methodological challenges that were identified by a

pilot study, and how these issues were resolved.

Chapter 4

Chapter 4 presents a small survey that was conducted to contextualise the research by

investigating the extent to which EAP teachers in Higher Education programmes in the UK

explicitly teach/correct English article use.

Chapter 5

The first main corpus-based study presented is a longitudinal study of the effect of explicit

teaching on 25 (L1 Mandarin Chinese) English learners during a 15-week Presessional

programme. The relative effects of correcting English article errors, teaching the article system

or not making any mention of articles will be compared and contrasted.

Chapter 6

Combining the data collected for the preceding study with some essays from lower proficiency

learners and case study responses, a corpus of 50,319 words was analysed for comparison with

a smaller corpus of L1 essays on the same topic. A comparison is then made between the essays

and a mini-corpus of case-studies in order to investigate genre effects on accuracy. In addition

to deductive analysis using the analytical framework developed by Bickerton (1981), some

qualitative methods and a freer/simpler corpus-driven inductive analysis of the same data are

conducted to investigate further the possible causes of errors.

Chapter 7

The final study uses a grammatical judgement multiple-choice methodology (455 participants)

to investigate several hypotheses formed as to why Chinese students oversupply or omit articles

and this chapter further investigates the link between the students’ general proficiency

development and their typical article use errors.

Chapter 8 (Conclusion)

After summarising and critically evaluating some of the findings in the four studies presented,

the final chapter provides some practical suggestions for EAP teachers wishing to focus on this

area of language. In addition to many concrete recommendations, it is hoped that several of

the main findings can help inform teachers’ ‘working hypotheses’ as to why students make

errors in article use and how students can be helped to improve their article accuracy.

2 LITERATURE REVIEW

The many variables which can affect Chinese L1 learners’ accuracy in any grammatical choice

are as wide ranging and complex as the English article system itself. The first section of this

chapter shows how the grammaticalisation of definiteness in the English language has led to

an extremely complex system that does not have a simple one-to-one mapping of form and

meaning. A contrastive analysis of Chinese is then presented in order to build hypotheses about

the predicted errors in later chapters. In addition to a greater understanding of the many

variables affecting article accuracy, the review of the literature presented here highlights many

areas of uncertainty that will be addressed in the Research Questions summarised in the final

section. More generally, the last part of this chapter aims to evaluate the arguments introduced

as to whether accuracy of English articles in academic writing should and could be an

appropriate focus of attention in the teaching of English for Academic Purposes.

2.1 Key term definitions

Ø article

Throughout this thesis, reference will be made to the, a/an, and Ø (zero) articles. In the

terminology employed in some of the published literature (Langendoen, 1970; Chesterman,

1991), the ‘null’ article (sometimes termed Ø2) refers to the zero article found with singular

nouns/proper nouns) as opposed to Ø used for plural/common nouns. However, to avoid

confusion, this thesis will define all contexts in which a noun phrase can be used alone (bare)

without an article or other determiner as a Ø article context. For the sake of simplicity the

thesis therefore does not distinguish between Ø article usage with proper nouns, singular, plural

or mass nouns.

Omission/overspecification

Based on Dulay, Burt & Krashen’s surface taxonomy (1982), most studies use the term

‘omission’ to denote a missing article in an obligatory context (Chuang & Nesi, 2006; Díez-

Bedmar & Papp, 2008; Lee & Chen 2009; Crosthwaite, 2016a, 2017). However the terms

‘overinclusion’ (Chuang & Nesi, 2006), ‘redundancy’ of article use (ibid.), ‘overspecification’

(Lee & Chen, 2009), ‘oversuppliance’ (ibid.) and ‘oversupply’ (Crosthwaite, 2016a) are among

the many different terms used in the literature to describe article use when such use is

ungrammatical/inappropriate. This study will use the nouns ‘overspecification’ and

‘oversuppliance’ and the verb forms ‘overspecify’ and ‘oversupply’ to denote inappropriate

use (when considered an error). When use of article is unnecessary but does not affect

grammatical accuracy or appropriacy, the word ‘redundancy’ is used to denote a more broad

connotation of marked use. To illustrate, the definite article in the following example taken

from a conclusion in the corpus of essays analysed in Chapter 6 shows a noun phrase (the

managers) which was agreed as correct, if marked, by three L1 English tutors. The three tutors

agreed that Ø was more likely, but that the was permissible in context of the parallel structure:

Motivation is not the only way to incentivise people in practice, since the managers just

search for the best ways in every situation [M_09552-4-Wk1]

Underuse/overuse

In contrast with omission (and its connotation of grammatical accuracy), ‘underuse’ is used in

this thesis to denote a lower than expected frequency used when compared to L1 discourse on

the same prompt or genre. Overuse is the parallel term used for quantitatively more frequent

use in a text as a whole, whether or not such use is grammatically correct.

2.2 The English article system

Although many Indo-European languages have some form of article system, their ancestors

(such as classical Latin and Sanskrit) did not. Today, many other world languages (Japanese,

Hindi, Chinese) do not have an article system, even if they have other markers of definiteness.

It is generally accepted that while all languages can refer to the specificity, definiteness and

number of a noun referent in some fashion, some languages have articles [+Article] while

others are article less [-Article]. The Oxford English Dictionary (OED Online, 2010), in its 23

main entries for the use of the definite article, reported that the in its current form was first seen

in the East Midlands dialect in c 1150, firmly establishing itself in English by the time of

Chaucer. Since then it can be seen from the OED that the use of the definite article has been in

a constant state of flux. In the 17

century all dates were preceded by the (e.g. the 1685) and

languages were always prefixed by the (i.e. the Latin), indeed the OED cites Webster in 1934

stating that ‘The modern descendants of the Latin are called the Romance languages’.

According to most sources (Bybee, Perkins & Pagliuca, 1998) it is agreed that the definite

article originated from the Old English demonstrative that (Old English did not have an article

system) while the indefinite article a/an came from the numeral one.

2.2.1 Bickerton’s four ‘Types’ of article use

Whether or not the English article system is today used uniformly by its different users, any

framework that explains its functions and forms needs to incorporate both its pragmatic and

grammatical functions. Grammatically, the form of article will in some cases depend on the

noun’s countability and number. Pragmatically, in its referential uses, articles reflect both the

context (whether the reader can infer its definiteness or indefiniteness) and the construal of

context (whether the reader can be expected to know about the referent and whether it is to be

construed in a specific or general way). Bickerton’s Framework, as shown in Table 1 below,

acknowledges that article choice is determined by the discourse features of the referents,

namely whether construed by the user as a specific referent [± SR] and whether known [± HK]

to the hearer. This framework provides four ‘Types’ of English article use and has the

advantage of explaining both definite, indefinite and generic uses of articles (but not idiomatic

or proper noun uses). The next section outlines each of these in turn.

Table 1: Bickerton’s (1981) semantic and pragmatic Framework

Type

Features

Context

Form of

Article

Examples

[-SR, +HK]

Generic: The reader knows of

its existence, but not a specific

reference.

a/an

The

An elephant never forgets.

Elephants never forget.

The elephant never forgets.

[+SR, +HK]

Writer and reader both know of

the specific referent.

The

Remember to feed the

elephant!

That’s the biggest elephant I’ve

ever seen.

[+SR, -HK]

The writer knows of a specific

referent the reader does not

know.

a/an

The local zoo has an elephant.

The local zoo has elephants.

[-SR, -HK]

Neither writer nor reader

believe the noun refers to a

specific thing

a/an

The zoo does not have an

elephant.

The zoo does not have

elephants.

Examples taken from Langendoen (1970), Cziko (1986), Heubner (1983), and Butler (2002).

Type 1 [-SR, +HK] Generic

The Type 1 (Generic) function of the English article system is used to indicate the more

general nature of the noun as a class. This generic function is often equally served by plurals,

a/an or the, as an example from Langendoen (1970) illustrates:

i. An elephant never forgets

ii. The elephant never forgets

iii. Elephants never forget

Although generic article use is one of the rarer forms of article use generally, it is far more

common in written academic English than other forms of discourse due to its use in defining

terms and topics. Indeed, Master posited some evidence (1987: 184) that generic the is most

common in essay paragraph topic statements, supporting the assumption that paragraphs

normally start with a core generalisation since, ‘topic sentences are by definition more

generalised than other sentences in the paragraph’. After analysing a corpus of articles from

the journal Scientific American, Master (1987) reported that the Ø article was the most frequent

generic article form (54%) of all generic uses, followed by generic the (38%) and finally the

generic a/an article (8%).

The importance of generic reference in academic writing is thus clear. However, an added

complication in generic noun phrases for L2 learners of English is that, as shown in Krifka and

Gerstner’s (1987) examples in Figure 1 below, generic contexts can allow various morphemes

(including Ø). That is to say that most generic phrases can be formulated in a variety of ways

with an article or Ø with plural count nouns (sentences 1, 2, 3 and 5), often with a choice of

bare plural or singular definite with little or no change in meaning.

1. The lion is a ferocious beast. (singular definite generic NP)

2. A lion is a ferocious beast. (singular indefinite generic NP)

3. Lions are ferocious beasts. (generic bare plurals)

4. Gold is a precious metal. (generic bare singular)

5. One cat, namely the lion, is a ferocious beast. Some cats, namely the lion and the tiger,

are ferocious beasts. (taxonomic NPs)

6. Rice was introduced in East Africa some centuries ago. (generic bare mass noun)

Figure 1: Examples of generic NPs in all forms (Krifka & Gerstner, 1987)

Mass nouns generally only allow the limited Ø article choice (sentence 6) but, apart from this

simple observation, the bewildering choice and the complicated rules of whether bare plurals

(Lions are dangerous), definite singulars (The lion is dangerous), or indefinite singulars (A

lion is dangerous) are preferred or restricted are highly complex issues (Langendoen, 1970;

Huebner, 1983; Cziko, 1986; Butler, 2002; Yang & Ionin, 2008; Ionin et al., 2011). To some

extent this complexity will clearly prove a challenge to all L2 learners of English, but section

2.3.1 will discuss claims that Chinese L1 learners may have particular issues with this type of

use.

Type 2 [+SR, +HK] Definite Articles

The definite article is the one type of article that can (and must) take only one morpheme as its

article: the. To be definite the referent must not only be specific, but must be construed as

‘known’ to the hearer. Being ‘known’ means it could be previously known or implied in an

inference to be made known through an utterance, or as Verspoor (2008) explains, ‘those

entities that are not necessarily identifiable to both the speaker and hearer, but the hearer can

infer that the speaker refers to a unique one in his mind as in “Be aware of the dog”, “I went

to the park”, or “I took the bus”’. As Berry points out (1991: 254), in such cases the definite

article ‘notifies the reader of the future significance of the referent’.

The definite article is also important in academic English because, although much of academic

writing at university relates to general ideas (and generic noun phrases), many of these ideas

will be supported by reference to specific examples. Moreover, the language of an academic

paper will be quasi-specific due to intra and inter-textual references; anaphoric (second

mention) reference; reference to the literature in the field; and cataphoric reference among other

functions. Quirk & Crystal (1985) identify eight functions of the definite article in writing:

as a marker of specific reference: the immediate situation (the roses are beautiful); unique

reference (the sun/the moon); anaphoric reference (second mention); cataphoric reference

(post-modified noun phrases and ‘of’phrases); sporadic reference (my sister goes to the theatre

every month); logical use with adjectives (the same, the only, superlatives) and reference to

body parts (the mind). Six of these functions of the definite article, setting sporadic reference

and body parts aside, seem most important for academic writing. Of these six written functions,

evidence suggests moreover that it is cataphoric reference which learners will most need in

academic English since, according to Biber et al. (1999), 40% of definite articles used in

academic writing have this function.

Type 3 [+SR, -HK] Indefinite articles

Where the reader cannot be assumed to have knowledge of the reference (for instance first

mention), a specific referent will be determined by a/an if a singular count noun, or Ø if

referring to a mass noun or plural count noun. Again, their use in academic English will be

related to specific examples used to support the more general claims in academic writing.

Type 4 [-SR, -HK] Non-referential noun phrases

When neither writer nor reader believe the noun refers to a specific thing, there is again the

same choice of morpheme: a/an for singular count nouns and Ø for non-count or plural. Further

examples from Ekiert (2004) are listed in Figure 2 overleaf. Such non-referential noun phrases

are likely to dominate much of academic writing, particularly when the genre requires general

claims, an evaluation of different (unreal) scenarios, conditional clauses and possible unproved

hypotheses.

Alice is an accountant.

I guess I should buy a new car.

Ø Foreigners would come up with a better solution.

Figure 2: Further examples of non-referential noun phrases from Ekiert (2004)

2.2.2 The conventional use of articles

Following Liu & Gleason (2002) and Ekiert (2004), this thesis will classify all uses of articles

that fall outside the Bickerton Framework, as ‘conventional use’, in that they are not a matter

of choice and do not follow immediately evident contextual considerations. This includes

idiomatic uses and proper nouns that are often taught to students as ‘things you just need to

learn’. This is not to say that such uses were not historically motivated. In fact, many uses of

proper nouns can be seen as logical (the definite article is normally only used when the proper

noun combines with a common noun). However, many uses would be difficult to explain

without in-depth research of the historical roots of the linguistic motivation.

Many discourse markers in academic English (e.g. the first point, on the other hand) could be

argued to fit into what Quirk & Crystal (1985) term ‘logical uses’ of definite article. However,

it might be more difficult to explain the majority of discourse idioms (e.g. on the whole, on the

rise, in the main). Even greater confusion is moreover caused by proper nouns, which normally

take Ø article (e.g. France, Mont Blanc, Peugeot, Tower Bridge) but sometimes take a definite

article (e.g. the UK, the Alps, the Seine, the Tower of London). This confusion may affect

[+Article] students as well as [-Article] learners because these conventions are so different (for

example the French would say ‘la France’ in addition to ‘le Royaume Uni’, adding a definite

article regardless of whether a country is singular, plural, a group of islands or a kingdom).

In summary, the grammaticalisation of the article system in English has led to an extremely

complex range of choices for the English L2 learner that may impact upon their writing style

and quality in academic English. Regardless of L1 background, the main problem facing all

learners is first and foremost that the article forms do not have simple form to function patterns

of use (Butler, 2002) and various forms of article have duplicated uses in different functions.

Furthermore, the learner needs to know the characteristics of the head noun before making their

choice. For Type 1 and Type 3 contexts, the choice of article often requires a knowledge of

the head noun’s singular/plural and count/mass-noun characteristics. Type 1 (generic) contexts

sometimes allow a choice of article (Ø with a plural or mass noun or a/the with a singular) but

in other cases require more restricted use. Finally, there are conventional uses that require a

knowledge of the original motivation for using the article (which may be obscured in the

present form). In sum, unlike many aspects of grammar that have a clear form/function

relationship, L2 English learners are facing a range of choices that have lesser or greater

restrictions in different contexts.

2.2.3 Evaluation of the effect of article errors

Article errors are often invisible and may be often considered less relevant in spoken English.

However, there seems a convincing argument that learners who wish to write academically or

professionally require more accurate use of articles (Master, 1987). There are two main

problems often caused for students who make many article errors. Firstly, the reading audience

may negatively perceive their message, equating their grammatical accuracy with their subject

knowledge (Master, 1995). Secondly, although the majority of errors are superficial – causing

only annoyance rather than confusion – some article errors can cause a barrier to effective

communication. The following example is taken from a student’s assignment in which she

first used (1) but later saw that she was meant to use (2) after teacher feedback.

The WTO has developed a high level of influence and consequently has had a large impact on

(1) …the number of trade-related issues …

(2) ...a number of trade-related issues….

(unpublished – author’s student)

In another published example Berry (1991) illustrates how article choices can have an effect

on the readers’ understanding:

(3) I want to know the English.

(4) I want to know English.

A further example from Berry shows how the article error can have a more nuanced but

nonetheless important effect on the readers’ construal of the message:

(5) Police investigating the crime have discovered a gun.

(6) Police investigating the crime have discovered the gun. (i.e. the one used

in the crime)

It could be counter-argued that the vast majority of article errors have a negligible effect on

communication, being only superficial or what are sometimes called ‘local errors’ (Burt &

Kiparsky, 1974) which seldom lead to miscomprehension. There are certainly academic

English competencies that should take a higher priority in the EAP teaching curriculum at all

learning levels and a high emphasis on all elements of the rule based article system in the most

intense courses or lower level classes will not be recommended at any point in this thesis. On

the other hand, when time permits incidental attention to this area, the neglect of the most

frequently occurring word in the English language seems less reasonable. This attention seems

particularly important in the context of international students paying high fees in university

Presessional EAP courses to learn how to avoid any misunderstandings of any sort in their

writing. Regardless of world English arguments, with their eyes set on the submission of

academic assessments, most international students expect EAP tutors at university to provide

scaffolded support towards accurate and genre appropriate writing skills.

Clearly, given that students frequently set themselves unrealistic and unnecessary goals such

as ‘native-like accuracy’, part of an effective EAP tutor’s role is to make students understand

that the occasional slip or marked overuse of articles will not prevent them achieving their

goals. The frequent overcorrection of article use that shall be demonstrated in this thesis is

likely to represent a recognition of such demands without adjustment of expectations, however

unhelpful such practices are shown to be. Equally, however, given the financial/time

investment these students have made in a presessional programme, it would seem disingenuous

to pretend that excessively inaccurate use of determiners in a piece of academic writing will

not have any effect at all on the reader. At some point in the learners’ proficiency, the neglect

of the article system in the syllabus becomes more difficult to defend. It may be more related

to the complexity of the grammar itself, teachers’ concerns about discouraging learner

motivation or a gap in available resources for the effective teaching of this language area.

On the one hand, difficulties with articles may be experienced by all learners of English,

regardless of L1 background. On the other hand, it has often been assumed that differences in

the way L1 languages express definiteness may have crosslinguistic interference on their

learning of the article system in English. The background to this assumption is explored in the

next section.

2.3 Contrastive Analysis

The claim has historically been made, both strongly (Lado, 1957) and in more nuanced forms

(Gilquin, 2008), that L1 background contributes to language inaccuracy in some way (the

Contrastive Analysis Hypothesis). A hypothesis has thus been made (Díez-Bedmar & Papp,

2008) that students will be more challenged by article use if their first language does not 1)

grammaticalise the article system at all or 2) grammaticalises the system in a different way.

This thesis will investigate a weak form of the Contrastive Analysis Hypothesis, but it will not

make the simplistic uncritical assumptions of linguists from the Contrastive Analysis era (e.g.

Lado, 1957) in which L1 negative transfer effect was thought to be the dominant if not singular

determiner of error. On the other hand, it will not take the equally extreme assumption of the

dominant SLA theories of the 1970s-1980s which perhaps underestimated crosslinguistic L1

interference.

Indeed, in the 1970s the widespread rejection of all forms of the CA hypothesis in both the

SLA and teacher training fields would have made this thesis impossible to write. Fortunately,

this thesis is written in an era of less doctrine and greater plurality of linguistic assumptions.

Cognitive linguistics in particular has been responsible for ‘reopening’ the issue of

crosslinguistic interference (Langacker, 1986). Figuratively speaking, applied linguists today

are more unshackled in their assumptions and can investigate all the possible determiners of

L2 development. This has given opportunities for a diversity of new approaches including

learner corpus-based integrated contrastive analysis approaches (Granger, 1996). Meanwhile,

the affordances of modern technology, combined with an acceptance of a range of factors

affecting L2 development have allowed the empirical examination of learner data (Ellis &

Larsen-Freeman, 2009) with a renewed focus on L1 transfer effects in general (Gilquin, 2008)

and an interest in crosslinguistic interference in article acquisition (Master, 1997; Lu, 2001;

Ekiert, 2004; Díez-Bedmar & Papp, 2008; Snape, Mayo & Gürel, 2013; Crosthwaite, 2012,

2016a).

2.3.1 Referential and non-referential noun phrases in Chinese

The high frequency of citations (Robertson, 2000; Díez-Bedmar & Papp, 2008; Crosthwaite,

2016a) to Li & Thompson’s (1981) ‘Chinese grammar reference’ show that it is widely

respected and useful for researchers investigating Chinese systems for showing definiteness

and indefiniteness. All examples in this section are also taken from Li &Thompson (ibid.)

unless otherwise stated.

As a [-Art] language, Chinese does not grammaticalise definiteness formally like a [+Art]

language such as English. However, writers can mark definiteness optionally if they wish

through word order, since definiteness ‘is partially signalled by preverbal position for topics,

subjects and objects.’ (p.20). The following examples (p.21) juxtapose how ‘book’ used before

the verb suggests definiteness, but after the verb commonly implies a notion of indefiniteness.

According to Li & Thompson, like English, nominal groups in Chinese have a head noun which

is always final, whether or not modified by other words in the noun phrase. As explained by Li

& Thompson, referential noun phrases in Chinese thus provide the writer with a degree of

choice. The writer can provide readers with a classifier, numeral or demonstrative for added

emphasis. Very often, alternatively, no such emphasis is added and readers are asked to infer

the definiteness or indefiniteness of an entity from context (ibid.). However, a key difference

in Chinese (that will be highlighted in Chapter 6) is that non-referential noun phrases, in

contrast, ‘never take classifier phrases’ (p.130). In the Type 4 indefinite sentence below

(p.127), ‘gōngchéngshī’ is non-referential in that it expressed a quality rather than a person.

Similarly, Li & Thompson (ibid.) demonstrate that generic noun phrases in topic position also

have no classifiers or demonstratives that could act to determine the noun in Chinese:

The challenge of understanding the grammatical constraints that determine how meaning is

mapped is also argued to be the main cause of error rather than the conceptual system itself

(Master, 1997; Robertson, 2000; Milton, 2001; Chuang & Nesi, 2006), but these challenges are

also influenced by L1 differences. More basically, plurality in Chinese is mapped by numerals,

classifiers and suffixes rather than final -s and there is no orthographic indication to

differentiate proper and common nouns in Chinese (Shi, 2016). Thus, the English article

system involves choices that Chinese learners must make without the contextual clues they

gain in their L1. As a consequence, Chuang & Nesi (2006: 266) attribute many article errors

they recorded to L1 transfer effects:

The Chinese language does not distinguish between count and noncount nouns, and does not

have a rigid, formal distinction between singular and plural (plural markers are not required);

this makes the concept of countability problematic for Chinese learners.

Faced with article choices, L1 English writers draw on their knowledge almost unconsciously

of how choices are constrained by the count/non-count/plural/mass noun distinctions of the

head noun (in each context). Thus [-Art] L1 learners face not only a new system of denoting

definiteness, but also need to understand how new lexis constrains their choice.

Based on the marked differences of English and Chinese in noun phrase determination Díez-

Bedmar & Papp (2008: 154) make a very generalised hypothesis of L1 transfer effects:

On the basis of the description of the lack of an article system in Chinese, we hypothesise that

Chinese speakers will underuse the definite and indefinite articles in English, overuse the zero

article, and even misuse them until they manage to learn to restrict the use of them to the specific

semantic contexts and pragmatic functions in which they are employed in English.

However, some commentators (Robertson, 2000; Xu, Shi & Snape, 2016; Crosthwaite, 2016a)

argue that Chinese is a system more ‘article-like’ (Crosthwaite, 2016a: 94) than article-less,

and the above hypothesis requires more nuance. In fact, definiteness can also be denoted by

demonstratives (e.g. zhèi ‘this’ and nèi ‘that’ ) while yi- ‘one’ is can also be optionally used

with referential nouns to signify indefiniteness. As explained by Li & Thompson (1981: 131-

132), Chinese learners of English will already be somewhat familiar with grammaticising

definiteness:

Finally, we should point out that Mandarin does not have words that correspond to the English

words the and a. The demonstrative nèi ‘that’, however, is beginning to function as ‘the’ if it

is not stressed and the numeral Yī ‘one’, if it is not stressed, is beginning to function as ‘a’.

This is logical because this is how languages such as English developed the definite article

(Bybee, Perkins & Pagliuca, 1998). The use of demonstratives leads Crosthwaite (2012, 2016a)

to argue that Chinese Mandarin students have an advantage (a positive transfer effect) over

Koreans and Japanese (other languages with grammaticised articles) when learning English.

Moreover, given that non-referential noun phrases can never be modified with classifiers,

Robertson (2000: 143) makes the point that Chinese learners are familiar with understanding a

referent as indefinite when used with a numeral but not a classifier while known to the writer

and specific (definite) through the use of such a classifier:

In Mandarin there is an interaction between referentiality and the use of nominal classifiers. Li

and Thompson (1981: 130) point out that nonreferential NPs never take classifier phrases. It

follows that a noun phrase with a classifier phrase must be referential. Furthermore, if the

classifier phrase includes a demonstrative (i.e. zhèi ‘this’ or nèi- ‘that’), then the NP must be

definite.

However, the extent to which Chinese can be depicted as ‘article-like’ is controversial. The

view that there are equivalents of some articles (mixtures of ‘one’ + classifier words) in

referential noun phrases in Chinese is shared by Chen (2004) only to the extent of indefinite

articles. While agreeing that demonstrative zhe and na are occasionally used in Chinese in

similar ways to definite articles, Chen is not convinced that occasional uses are evidence of the

grammaticalization of definiteness in Chinese. In fact, while not all uses are purely pure deictic

(referring to physical or linguistic co-presence), Chen (2004: 1156) highlights that their use to

demonstrate anaphoric or shared knowledge is rare in writing compared to speaking:

In sum, while demonstratives in Chinese have developed some functions which are typically

served by deﬁnite articles in languages like English, they are, generally speaking, still much

closer to demonstratives than deﬁnite articles on the path of grammaticalization.

It should also be finally noted that demonstratives in Chinese may have a slightly nuanced

notion of definiteness, which could also explain some L1 transfer effects. Xu, Shi & Snape,

(2016) claim that Chinese learners sometimes confuse specificity with definiteness and may

overuse the in a/an contexts as a result. Moreover, researchers such as Ionin (Ionin, Ko &

Wexler, 2004; Yang & Ionin, 2008; Ionin et al., 2011) have argued that article-less L1 learners

of English L2 often make this interlanguage ‘fluctuation hypothesis’ error of misusing articles

during their gradual L2 development, confusing [+/- definiteness] with [+/- specificity].

In sum, the contrastive analysis of Chinese and English shows that there are both similarities

and differences in the methods of expressing notions of indefiniteness and definiteness in the

two written languages. These differences appear most marked in the way the two languages

treat non-referential noun phrases since English displays greater choice (a/an, Ø) than the

Chinese bare nominal form (without a classifier) in Type 4 contexts and even greater choice

(a/an, Ø,the) in the way generic Type 1 noun phrases are determined. Beyond this consensus,

however, this literature review has shown some divergent views about the extent to which the

pseudo grammaticalisation (use of demonstratives and classifiers) in Chinese will create

opportunities for positive transfer effects. Various hypotheses are developed in Chapters 6 and

7 which address a central Research Question in this thesis:

To what extent does the Chinese students’ L1 background affect (1) Target-like use and (2)

their types of article error?

2.4 Empirical research into Chinese L1 learners’ written L2 English errors

2.4.1 Studies of [-Art] L1 background students’ use of articles

Much research has focussed on whether [-Article] L1 students have greater problems mastering

accurate use of articles than [+Article] L1students in English. A number of earlier studies

certainly claimed that such ‘article-less’ background students show less accuracy with

indefinite and definite articles (Yamada & Matsuura, 1982; Master, 1995; Master, 1997; Odlin,

1989; Jarvis, 2002). Investigating the universal language acquisition hypothesis, much effort

in SLA studies was also focussed on the effect this ‘[+/- article] parameter’ had on English L2

learners’ order of acquisition. According to some (Huebner, 1983; Master, 1997; Thomas,

1989) article-less L1 learners were observed to go through stages of ‘flooding’ language with

definite articles at beginner-intermediate stages of development.

Other studies by Liu & Gleason (2002) and Young (1996) however conclude the opposite:

early and accurate control of the indefinite article is observed before definite article accuracy.

In addition, in their study of the ‘natural order’ of article acquisition (excluding generic uses),

Liu & Gleason (2002) found that learners had the greatest problems inserting the obligatory

use of the in texts where the definite article had been omitted in contexts of conventional, or

what they termed ‘cultural’ use of the definite article. As Figure 3 shows below (comparing

average errors made on the vertical axis with levels and types of article error), this not only

seemed to be the most marked problem for learners but also the most persistent, less affected

by level of proficiency.

Figure 3: Comparison of means of missed the by category (Liu and Gleason, 2002: 13)

For Liu & Gleason this is less surprising given what learners perceive as arbitrary uses of

articles (2002: 7):

…such use of the is often not framed by situation but is determined, to a large extent, by

conventional practice. For example, we use the with some but not all disease names. Similarly,

we place the before some geographical names (such as rivers) but not others (such as most

lakes). We also use the with the musical instruments we play but not with the sports we play,

for we can say play the piano but not play the basketball.

One clear limitation of all the studies above was their approach of grouping Chinese students

with Japanese/Korean/Russian students to form ‘article-less’ background L1 groups.

Moreover, few earlier studies properly accounted for level and most have little relevancy for

the study of university student writing. Certainly, the findings of any studies that conflated the

analysis of Chinese learners’ article use with other L1 background students or observed just

one student such as Heubner (1983) need to be treated with a great deal of caution. In fact,

there are few valid justifications for studying the errors of such groups of students mixed

together since, as shown in the contrastive analysis in Section 2.3, some claim that Chinese to

some extent grammaticalises definiteness relative to other article-less languages (Robertson,

2000; Xu, Shi & Snape, 2016; Crosthwaite, 2016a). This chapter now turns its attention to

more robust recent studies that used corpus-based approaches and differentiated Chinese L1

groups from others to make more convincing claims about Chinese university students’ English

L2 article use.

2.4.2 Corpus based evidence from university EAP contexts

Using a corpus-based approach, Milton ( 2001) found that Chinese L1 students at a Hong Kong

university made frequent article errors and argued that these often resulted from their problems

distinguishing between countable and uncountable nouns. Making the point that article errors

were difficult to differentiate from final -s noun errors, Milton chose not to differentiate

between errors in singular/plural distinctions and article error.

In Chuang & Nesi’s (2006) study of undergraduate level Chinese students academic writing,

grammatical errors (85.9%) were far more common than lexico-grammatical (5%) or lexical

errors (9.1%). Of these, article errors (27.6%) were by far the most frequent category of

grammatical errors identified. This large corpus study was one of the first to present

meaningful findings based on authentic student work (50 pieces of authentic academic writing

of 1,500-2,000 words by Foundation level students at a B1/B2 level). Chuang & Nesi’s (2006)

finding that omission of definite articles is the most frequent error, closely followed by

overspecification of definite articles is therefore of significance to this research into

Presessional students’ article use (though Chuang and Nesi’s students were slightly younger).

According to Chuang and Nesi, the omission of a/an was the third most frequent error and

misselection or ‘a for the’ and vice versa was a rare error.

Another important corpus-based study is Lee & Chen’s (2009) keyword analysis of

undergraduate Chinese L1 English major students’ dissertations and some shorter texts from

the BAWE corpus. This study reported that the definite article was ‘overused’ compared to the

authors’ corpus of L1 academic writing. Indeed, Lee & Chen suggest that it is closed

class/function words and common words that students overuse rather than the academic words

typically focussed upon in EAP classes. Their finding that Chinese students oversupply the

definite article appears to support the claim that Chinese learners go through a ‘flooding’ stage

of definite article use in their acquisition process (Heubner, 1983; Thomas, 1989; Master, 1997;

Lu, 2001), even at university level.

Lee and Chen (2009) illustrate the frequent overspecification of definite article use with the

example from a Chinese L1 English major’s undergraduate dissertation submission, shown in

Figure 4. This demonstrates what they claim is ‘a tendency to use the even when talking about

general rather than definite subjects’ (ibid, 2009: 287).

Figure 4: Example of definite article ‘overuse’ from Lee & Chen (2009: 287)

The example above serves to illustrate what some researchers have identified and claimed to

be a rather ‘spoken like’ characteristic in Chinese university students’ academic writing (Cobb,

2003; Hinkel, 2003; Granger, 2004; Mayor, 2006; Gilquin & Paquot, 2007). While out of

context it is not easy to decide if all the alleged overspecifications are errors, the uses of definite

article in ‘the confidence’, ‘the language ability’ and ‘the text-based questions’ do show (in the

researcher’s opinion) an excessively conversational tone for the genre in question (an Applied

Linguistics dissertation), particularly as the student was seeking to make a generalised claim

about a topic (as evidenced from their use of Ø + students in the topic sentence). If Lee and

Chen’s example is representative of the overall dissertation style, there seems to be a plausible

argument that such overspecification would create at best an informal tone and at worst a

misunderstanding in meaning. In turn, this again shows that article errors can sometimes go

beyond mere surface level error and can affect the perceived register (and appropriacy within

a genre).

One frequent criticism of such generalisations (i.e. that Chinese students write too informally

when producing academic texts) is that the data analysed often comes from semi-formal IELTS

type prompts and Leedham (2015) makes a convincing case that many of the cited studies

alleging a conversational tone failed to sample the students’ true abilities in academic writing

because they examine short IELTS type essays which themselves promote an informal tone.

Lee & Chen’s use of authentic (assessed) 5,000 word dissertations, however, makes their claim

of a register effect of article oversupply in academic writing more convincing.

Another critique of such broad generalisations of ‘oral tone’ or ‘informal language’ is the huge

variability across disciplines and even university departments about what constitutes

acceptable academic style (Hyland & Milton, 1997; Hyland, 2004; Swales, 2004). Admittedly,

EAP tutors and even researchers can struggle to make judgements about register appropriacy

for different genres and, as Gardner & Holmes (2009) point out, most come from a humanities

background which has conventions that differ from those of than the students they are teaching

which can somewhat invalidate such judgements. It is indeed essential that any researcher

makes judgements specific to a given genre and discipline. As will be highlighted in Chapter

6, a redundant definite article might be marked but acceptable in one genre (a case study) but

wholly inappropriate in a journal article introduction or conclusion (when the writer is creating

a frame of reference to the general rather than the specific actors in the discourse community).

On proviso that judgements are kept discipline specific and focussed on a specific genre that

the researcher understands, this thesis makes the assumption that nuanced judgements of article

use and misuse can be made and, moreover, that such a research focus is essential in the EAP

community if it is to help novice academic writers develop register awareness in an L2. The

role of EAP tutors in helping students conceptualise their audience (Ede & Lunsford, 1984) is

long acknowledged and more effective EAP tutors are often argued to be those who teach

through genre (Flowerdew & Peacock, 2001; Hyland, 2007b). The ideal EAP programme

would therefore appear to be one that trains students so that they leave their course ready to

investigate the genre requirements in their own disciplines for themselves (Paltridge, 2001).

Focussing on small words such as articles can occasionally be essential to teach genre and

small morphemes in language frame the writer/reader relationship arguably defeats the central

purpose of EAP in training students in the interplay between language, register and genre

variation. By showing students that different genres within their discipline may prefer or

discourage the informal tone created by definite article use (as opposed to a bare Ø plural),

students can be shown what Leedham calls ‘the preferred ways of making meaning’ (2015: 7)

in the genre. The issue of how students conceptualise their reader is clearly the underlying

variable (Ede & Lunsford, 1984) but students might usefully be focussed upon how article use

invokes their audience distance in writing composition. In short, Lee & Chen’s (2009)

argument that EAP needs to be ‘making a bigger deal of the smaller words’ (their paper title)

appears persuasive.

One more justifiable criticism of the approach of the three corpus studies discussed above could

be of their ‘deficit approach’ (Leedham, 2015) in problematising Chinese students’ errors. It

is difficult to argue against this accusation, and it must be acknowledged that all such error

studies make the assumption that L1 norms are a valid target. In fact, by focussing on article

error frequency, such studies risk ignoring the accurate use of articles. However, two corpus

studies summarised in the following section (Díez-Bedmar & Papp, 2008; Crosthwaite, 2016a)

have overcome this emphasis by using ratio measures that allow a more holistic picture of

students’ accurate use in addition to their errors.

Using Granger’s (1996) Integrated Contrastive model and the Bickerton Framework (see

Section 2.2.1), Díez-Bedmar & Papp (2008) compared the errors made by Chinese L1

international students at Portsmouth University with the writing of L1 English writers and

Spanish L1 students of L2 English. Their use of the framework in addition to Pica’s (1983)

Target-Like Use (TLU) ratio has been replicated by other studies including those of Chapter 5

and 6 in this thesis. Using Pica’s index (which measures correct use as a ratio of incorrect use),

Díez-Bedmar & Papp found that their Chinese L1 students used Ø article most correctly

(75.9%), followed by a/an (73.2%) and lastly the definite article which was used least

correctly (67.4%). This low Chinese learner accuracy contrasted with the Spanish learners who

were least accurate with zero articles and more accurate with a/an articles than the articles.

Chinese learners were also reported as having particular problems with generic articles and

Type 3 (see Section 2.2.1) context articles.

It is very difficult to compare the findings reported in raw frequency terms (Chuang & Nesi,

2006; Lee & Chen, 2009) with those that use ratios such as ‘Target-Like Use’ (Díez-Bedmar

& Papp, 2008; Crosthwaite, 2016a) because the former analyses are reporting the under/over

specification of articles in terms of frequency (the most frequent errors), while the latter are

taking the accuracy rate in terms of how often the student made the right choice in obligatory

contexts. In other words, the former is answering the question ‘what are the most frequent

article errors made in writing?’ while the latter is asking ‘given an equal number of obligatory

contexts, which article’s use would challenge the learner most?’. As Ø is the most frequent

free morpheme in written academic English with many more obligatory uses in most genres

than a/an, Díez-Bedmar & Papp’s (2008) conclusion that Chinese students’ ‘Target-Like Use’

of a/an is lower than that of Ø cannot be interpreted as showing Ø to be less of ‘a problem’

in frequency of error terms, but does helpfully acknowledge the high accuracy of Chinese

students’ use of Ø when its frequency of use is factored into the discussion.

Crosthwaite (2016a) investigated differences between the accuracy in article use of different

article-less L1 background students (Chinese, Korean and Thai) and the divergences of

accuracy order among four different language competency levels (beginner/low-

intermediate/intermediate/upper-intermediate). Compared to Díez-Bedmar & Papp’s 2008

study, Crosthwaite’s findings suggest a higher accuracy Target-Like Use among Chinese

learners. Indeed, at Upper-Intermediate levels equivalent to a Presessional EAP programme’s

entry levels, he reports that such levels of learner do not struggle significantly with any of the

articles (the, a/an, Ø), apart from occasional challenges with generic use contexts. In fact, in

the two short essay prompts set in the ICNALE corpus that Crosthwaite used, students at

Upper-Intermediate level showed almost perfect Target-Like Use (TLU) ratios (statistically)

in most article contexts and a fairly equal order of accuracy in all articles apart from generic

contexts.

However, at intermediate levels (slightly below Presessional entry levels), Crosthwaite (2016a)

does report that the TLU of definite articles was significantly lower than that of indefinite

articles. Moreover, he acknowledges that Chinese learners struggle with generic contexts,

oversupplying definite and indefinite articles when they do attempt to use them (but the median

measures he uses removes such outliers from his headline results). In another study of the errors

made longitudinally over three data-points in essays and reports by Chinese L1 university

students (while students received error correction) Crosthwaite (2016b) found that the most

frequent article error in both essay genres and reports was definite article omission. In

addition, but to a lesser extent, students oversupplied definite articles in generic and non-

referential contexts. As a ratio of response length, Crosthwaite also reports that article accuracy

was higher in the reports than the essays.

2.4.3 The contrasting hierarchy of difficulty reported in the research

To summarise, in the four most important corpus-based studies summarised in this chapter, as

shown in Table 2, there are divergences in the reported findings pertaining to the order of

difficulty. Discussing this order in terms of the contexts that students find most difficult to

master, it can be seen that the two studies showing definite article omission to be the most

serious error for Chinese learners (Chuang & Nesi, 2006; Díez-Bedmar & Papp, 2008) conflict

with the other two studies. This area of disagreement motivates the next Research Question

developed at the end of this chapter:

What are the linguistic contexts of obligatory article use in which international students are

most likely to make errors?

Table 2: A comparison of important corpus findings

Study

Student level

Discipline

Hierarchy of difficulty

(lower accuracy first)

(Chuang & Nesi, 2006)

Foundation

B1-B2 level

Business studies

Foundation students

The> Ø> >a/an

(Lee & Chen, 2009)

Undergraduate

Unspecified English level

Linguistics/Applied

linguistics

Ø>the (no mention of

a/an)

(Díez-Bedmar & Papp,

2008)

University students in 2

year or later UG/PG level,

IELTS 4.5 – 7.0

Various

The >a/an> Ø

(Crosthwaite, 2016a)

4 levels: (A2-B2/C1)

Various

The =a/an= Ø at Upper-

Intermediate levels.

2.5 Other factors influencing English Accuracy

To make conclusions about how article accuracy could be promoted by EAP pedagogy and to

differentiate L1 transfer effects, it is clearly necessary to consider a number of various

developmental factors. L1 transfer effects are just one of many possible factors affecting

learners’ academic writing accuracy. Firstly, there is general consensus in the EAP community

(Master, 2002; Hinkel, 2003; Alexander, Argent & Spencer, 2008) and SLA field (Larsen-

Freeman and Long, 1991) that L2 learners of English make errors as novice academic writers

as a process in their natural acquisition and that, regardless of L1 background or instruction,

students’ surface writing accuracy improves with sufficient comprehensible input in the

discipline discourse of their new academic community. Empirical studies into Chinese

learners’ article accuracy all therefore acknowledge general proficiency as a factor that must

be controlled in any measure of L1 transfer effects (Lu, 2001; Díez-Bedmar & Papp, 2008;

Díez-Bedmar, 2015; Crosthwaite, 2016a) or teaching/error-correction (Master, 1995, 1997;

Crosthwaite, 2016b). There is more generally a natural order of acquisition of L2 grammar

forms (Larsen-Freeman and Long, 1991; Ellis & Larsen-Freeman, 2009). Research into L2

acquisition of English articles by different L1 groups has shown that this order of acquisition

of English articles is slightly divergent across different L1 backgrounds of English L2 learners

(Luk & Shirai, 2009; Murakami & Alexopoulou, 2016), but these studies have shown that

article accuracy improves at higher proficiency levels. Moreover, it has been shown that not

all contexts and forms of article use are affected by proficiency equally and the effect of

proficiency level on different contexts of article has been the focus of some studies (Liu and

Gleason, 2002; Crosthwaite, 2016a).

Different parts of the EAP community agree and disagree about the extent to which academic

writing accuracy in general is affected by instruction. The section below discusses how article

use accuracy may or may not be accelerated by:1) Reading into writing methodologies, 2) a

focus on lexis and phraseology, 3) error correction and/or explicit teaching of the article

system, and 4) current approaches to teaching EAP.

2.5.1 The ‘reading into writing’ paradigm

A strong argument can be made that students increase their accuracy of article use in their own

discipline’s genres by reading extensively in their field. In broader linguistic terms, the main

evidence for this consensus view come from SLA studies (Dulay, Burt & Krashen, 1982;

Richards & Rodgers, 1986; Nunan, 1991; Savignon, 1991; Larsen-Freeman & Long, 1991)

which showed that acquisition is most closely accelerated or hindered by the requisite student

access to comprehensible input. EAP theorists meanwhile argue that academic texts (through

a genre approach) give students models for academic writing (Hyland, 2007a; Hirvela, 2016).

From passive input it is argued that students learn about genre interplay with language to frame

their own output (Carrell & Carson, 1997; Hirvela, 2016). Therefore, by passively seeing

how new terms colligate with articles in their discipline’s lexis, students could be expected to

gain confidence in accurately determining these concepts in noun phrases within their writing.

While the focus in EAP is normally upon meaning from input to output (Nation, 2007), students

in parallel also acquire patterns of how new vocabulary is paired together with other lexical

words (collocations) and frequently partnered with grammatical words (colligations). Through

reading, students therefore acquire many idiomatic uses of articles in fixed colligations within

phrases (lexical bundles). Although the acquisition of many article uses within lexical bundles

can to some extent be framed as vocabulary development, research has shown that knowledge

of vocabulary and appropriate control of fixed terms within formulaic phraseology are

accelerated through reading rather than list learning (Li & Schmitt, 2009; Hyland, 2008). By

reading discipline specific texts students see the vocabulary in their field and the pragmatic

ways of appropriately referring to concepts in different genres. Returning to Lee & Chen’s

(2009) example of student writing in Figure 4, this student might read further Applied

Linguistics texts that show how key concepts and actors (students, vocabulary etc., ) have

divergent references in specific and generalised examples. Extensive reading might illustrate

to students how articles are used or not used in the generic references of the key concepts in

their discipline. This key importance of new vocabulary further demonstrates Nation’s (2007:

5) argument that ‘Knowing a word involves many different aspects of knowledge’.

2.5.2 Knowledge of lexis and phraseology

The different aspects of knowledge of a word to which Nation alludes include its

singular/plural/mass noun characteristics and its colligations and collocations. As accurate

article use requires such knowledge, the students’ knowledge of vocabulary is inextricably

linked to their accuracy of articles in a given topic. In recalling Sinclair’s (1991) finding that

lexical choices are sometimes ‘closed’ as well as ‘open’, EAP tutors also have a role in

focussing students on the use of articles in fixed phraseology. A review of 35 most frequently

used ‘lexical bundles’ (e.g. ‘on the basis of’; ‘at the time of’) across academic disciplines

reported by Byrd & Coxhead (2010) shows that only 8 phrases do not contain an article. Byrd

& Coxhead (2010) argue for the importance of drawing learners’ attention to such

‘prefabricated and often repeated language’ (p.50). Li & Schmitt (2009) examined one EAP

student’s acquisition of phraseology over 12 months and found that after first contact with new

terms the learner went through stages of using the new vocabulary inappropriately before

mastery. In light of this aspect of the EAP approach, the main input to such acquisition of

articles is therefore reading, but instruction may also play a part in correcting inappropriate

use.

2.5.3 Error correction

Aside from form-focussed corrections, there is some support for feedback and corrections in

the genre approach to EAP when focussed on content or the way lexico-grammatical features

affect genre. In fact, within a genre approach, some argue it is an essential tool for correcting

inappropriacy and scaffolding learning:

In genre classrooms feedback is a key element of the scaffolding provided by the teacher to

build learner conﬁdence and the literacy resources to participate in target communities.

(Hyland & Hyland, 2006; 83)

As outlined in the introduction, error correction of all surface errors in students’ writing without

focussed purpose is generally argued to conflict with the ‘EAP paradigm’ (Alexander, Argent,

& Spencer, 2008). Yet, what teachers have read or learned in their training does not always

determine their classroom practices. In fact, research by Etheringon & Burgess (2002) into the

practices of 48 UK based EAP tutors in centres affiliated with the British Association of

Lecturers of English for Academic Purposes (BALEAP) showed that over half the EAP tutors

regularly corrected grammar errors even when they did not affect meaning. Moreover, over

half the teachers believed grammatical accuracy in writing was improved by form-focussed

corrections.

Empirical studies often claim to show significant effects on writing accuracy in general from

error correction (Polio, Fleck & Leder, 1998; Chandler, 2003; Ferris & Kurzer, 2019). There

is a broad consensus that correction has risks and can often be ineffective (see Truscott, 2004

for example), but some voices in the literature claim that such risks can be overcome under the

right conditions with sufficient teacher training (Ellis, 1994; Ferris, 1999; Bitchener, Young &

Cameron, 2005; Russell, 2009; Gass & Selinker, 2010; Rahimpour, Salimi & Farrokhi, 2012).

Even more consistently, studies show students value corrective feedback (Lee, 2005; Ferris,

2006; Hyland & Hyland, 2006) and prefer a teacher’s error focussed corrections to peer

corrections (Zhang, 1995). As a rule-based area of language accuracy, it has been argued that

article use is more amenable to error correction than other surface errors (Ferris, 2006; Dale,

Anisimoff & Narroway, 2012). However, the empirical claim that error correction improves

writing accuracy over the long term continues to be disputed in general (Truscott, 1996, 2004)

and in the more particular case of article use accuracy (Crosthwaite, 2016b). An evaluation of

this evidence in particular reference to article use will be presented in Section 2.7.

2.5.4 Explicit teaching of the article system

Although there has long been a broad consensus that EAP teaching should be content-driven

(using authentic texts wherever possible to provide the comprehensible input for acquisition),

there is less agreement about the extent to which language instruction about grammar should

be overt and form-focussed (Brinton & Holten, 2001). On the one hand, tutors report they do

not believe decontextualised grammar teaching is appropriate in EAP when surveyed

(Etheringon & Burgess, 2002). On the other hand, as Chapter 4 shall highlight, what teachers

say they do and what they actually do in the classroom has been shown to be divergent (Phipps

and Borg, 2009). In addition, when they are asked students appear to want more grammar

teaching than their tutors feel it is necessary to provide (Schultz, 2001; Ferris, 2006). This

may be in part due to the background of Chinese students and the way they have been

continually tested in decontextualised grammar terms in the Chinese education system

(Leedham, 2015). Ironically, Lu (2001: 43) points out that Chinese students spend a lot of time

doing tests and grammar exercises at lower levels, when they have little possible hope of

applying the rules, but less time on such errors when they are more ready to acquire them.

2.5.5 How articles can be taught in EAP

Based on the assumption that a focus on form in terms of selective text based noticing or

consciousness-raising activities can be appropriate as long as any focus on form is not

decontextualised from the text and overall discourse ( Hyland & Milton, 1997; Hinkel, 2003;

Hyland & Hyland, 2006; Alexander, Argent & Spencer, 2008; Hyland, 2008), this section

develops the thesis that a focus on articles can be integrated within the pedagogical methods

often recommended in EAP literature such data-driven learning and a genre-based approach.

Peter Master (1990; 1995; 1997; 2002) has long been the fiercest advocate of focussing

students on the effects of articles in texts and his later approach calls for a combined focus on

juxtaposed uses of articles with error correction (student self-correction and teacher

corrections). Another advocate was Johns (1991), who argued that such complex issues of

grammar could be covered within ‘data-driven learning’ in which the teacher tells the students

“I’m not sure: let’s find out together” (p.2) in contrast to the ‘spoonfed’ approach of grammar

books the data-driven approach involves asking students to deduce colligation and collocation

patterns by themselves through concordance lines (i.e. examples searched from a corpus of

academic texts). For Johns (ibid.) framing students as researchers rather than learners makes

the approach particularly suitable in the university context. Furthermore, for Johns: (ibid.; 3)

only data-driven approaches allow teachers to address the most complex grammatical problems

in student work:

There are large areas of English syntax which have traditionally been neglected in the standard

descriptive and pedagogic grammars not because they are unimportant but because they are too

difficult (e.g. Article Usage) or because they have simply been overlooked (e.g. Transitivity).

Hinkel (2013:7) argues that higher level learners need tutors to focus on ‘effects of grammatical

features on context, discourse and text’. There are certainly many calls for more form focussed

instruction with regards to nominalisation in EAP (Halliday & Martin, 1993; Fang,

Schleppegrell & Cox, 2006; Biber & Gray, 2010; Bennett, 2011; Parkinson & Musgrave,

2014) and this is inextricably linked to article use. Helping students to ‘pack meaning into

the noun phrase, and to make their text nominally rather than clausally complex’ (Parkinson &

Musgrave, 2014; 48) also requires a parallel focus on article use (or non-use).

In Chapters 6 and 7 this thesis advances the claim that the ‘compressed’ nature of academic

noun phrases in academic writing obscure the grammatical characteristics (-/+ number, -/+

countability) of the nominal group, just as they have been argued to obscure complex implicit

meaning relations (Halliday & Martin, 1993; Biber & Gray, 2010) in academic writing. The

hypothesis will be examined that any challenges that article-less L1 background international

students face in learning the grammatical properties of new lexis are heightened when they

are asked to use them in noun groups with pre-modifiers that increase the cognitive load of

applying new grammatical mappings onto language. Thus, among the many causes for the

omission of articles in Chinese students’ writing, one factor that may amplify their potential

for error is this compression of the linguistic context of article use.

Indeed, according to Biber et al (1999), 60% of nouns in academic writing are pre-modified

compared to only 15% in conversational English. Biber & Gray (2010: 19) note that

compressed noun phrases help experts write and read academic discourse, but challenge

students new to the genre:

In contrast to the academic professional, the compact, inexplicit discourse styles of research

articles are difficult for novice students. This is because students lack the specialist knowledge

that would allow them to readily infer the expected meaning of compact, inexplicit

constructions.

Lamenting that learning materials typically ignore the issue of nominalisation, Biber & Gray

(2010: 19) also criticise materials and pedagogical grammar books for presenting over

complicated elaborations less relevant to EAP in their place:

Traditional grammar practice for advanced academic purposes tends to focus on elaborated

structures, often dependent clauses like relative clauses, adverbial clauses, and complement

clauses. Phrasal modification especially noun phrase structures with multiple levels of phrasal

embedding is usually given much less attention.

Unfortunately, when they do pay attention to noun phrases, teaching materials and grammar

reference materials tend to present articles in prototypical sanitised contexts, not in the messy

academic contexts in which head nouns are obscured by pre-modification. For example, it has

been observed that grammar textbooks pay particular attention to simplified anaphoric forms

of indefinite and definite article and little attention to Ø or the generic definite article (Yoo,

2009). Rather than the simpler aspects of the article system ‘rule’ there is arguably a greater

need for the teaching of the article system in relation to such dense academic nominalisation.

The corpus-based analyses of the spoken and written grammar of academic English conducted

first by Quirk & Crystal (1985) and in the 1990s by Biber and colleagues (Biber, et al, 1999;

Biber, Leech, & Conrad, 2002) are frequently used to juxtapose students’ L2 English use with

the norms of use these frequency counts can be claimed to represent. Comparing the writing of

expert academic writers and students in ‘Freshman writing composition classes’ in the UK,

Bennett (2011) highlights the greater frequency of pre-modified noun phrases in academic

writing which novice writers need help in noticing and developing within their own writing. It

has also been shown that L2 English EAP students, challenged by compressing nominal groups

in the first instance, go through a stage of pre-modifying noun phrases with attributive

adjectives rather than classifying nouns (Parkinson & Musgrave, 2014). In later stages of

development, the students must learn to use other noun classifiers, participle phrases,

possessive nouns and appositive noun phrases to qualify the head noun (ibid). The relevance

for EAP tutors, for Parkinson & Musgrave, is that this feature of academic writing needs more

focus on EAP courses and they propose ‘a focus on nouns as premodifiers and prepositional

phrases as postmodifiers’ (p.49).

2.6 English articles within the current EAP pedagogical paradigms

As discussed in the introduction to this thesis, the extent to which a focus on form is compatible

with EAP has often been challenged with simplistic rejections of decontextualised grammar

teaching and assumptions of a ‘core knowledge’ of grammar or lexis that is useful across

disciplines. In fact, new teacher entrants to EAP might be forgiven for thinking that a focus on

grammatical accuracy had been consigned to history by more recent approaches to pedagogy.

However, this thesis argues that current pedagogies allow for a more meaningful focus on

language forms such as article use. A focus on how the writer uses articles to move the readers’

focus from specific examples to general statements or when and how knowledge is assumed

would seem have obvious synergies with the genre approach to EAP in which emphasis is put

on contrasting different writing purposes and appropriate writing styles (Swales, 1990; Hyland,

2002a; Swales, 2004; Swales & Feak, 2004; Hyland, 2007b).

The genre approach focusses students on different text types (essays, research reports, case

studies, lab reports ect.,) that they may meet in their academic community (Paltridge, 2001).

Hyland (2007b) argues that the genre approach allows students to see models of how language

choices are made in authentic text types. It would seem logical to argue that the way in which

article use varies across genres and how usage itself shapes the register and framing of the

audience-writer relationship build a strong case for a focus on articles. As shall be shown in

Chapter 6, uses of articles may be different from one genre (a case study report) to the next (an

essay) as the conventions of providing evidence and the purpose and assumed audience of the

writing changes. Rich language content in such authentic texts provides many opportunities

for a focus on grammar and form (Master, 1997; Brinton & Holten, 2001), as students can be

focussed on article use within texts and between genres.

As exemplified by the example of definite article overuse by a student in Chen & Lee (2009),

article use can affect the academic style of writing. The same authors argue for the importance

of focussing students on lexico-grammatical features and their effects upon register. However,

such ‘register’ awareness sessions can have little meaning out of context since the appropriacy

of a determiner will be greatly affected by framing of the reader-writer relationship and the

degree to which specific examples are being referenced in the genre. In fact EAP learners

need to be encouraged to start writing compositions by considering their audience (Flowerdew,

1993; Dudley-Evans & St John, 1998; Swales, 2004; Swales & Feak, 2004) and EAP tutors

also have a role in developing students’ consciousness of how they invoke the audience (Ede

& Lunsford, 1984). Teaching students about articles within a genre approach is naturally

somewhat contingent on how specific the focus of the EAP instruction is, and this review of

the literature would be incomplete without discussing the issue of whether students are in

specific disciplines or more general academic English courses.

The controversy in EAP that has occasional bearing on the issue of teaching article use in EAP

is the issue of whether students are taught in English for Specific Purpose (ESP) groups by

discipline or can learn the academic English they require effectively in more general mixed

disciplinary English for General Academic English Purposes (EGAP) classes. Some argue very

strongly for the former (Dudley-Evans & St John, 1998; Hyland, 2002b, 2012) while others

argue that EGAP can be effective (e.g. Hutchinson and Waters 1987; Spack 1988; Alexander,

Argent & Spencer, 2008). Those arguing for discipline based teaching often claim that the

writing requirements are extremely divergent across different disciplines (Swales, 1990;

Hyland, 2002b, 2012) and only through grouping students by discipline can a proper focus be

given to the lexis and genre of each student’s specific field (Dudley-Evans & St John, 1998).

Those supporting EGAP argue that there is a core body of academic English vocabulary, skills

and grammar needed by all students and that practical impediments make such ESP classes

unrealistic in real world contexts.

While attention to article use in EGAP contexts is possible, it is likely to be especially effective

in ESP contexts due to the importance of lexis and colligational knowledge for article accuracy.

The assumption made by EGAP advocates that students acquire specialised vocabulary fairly

quickly in their later studies is sound, but each discipline has not only a specialised lexis but

its own preferred ways of making generic references in writing, by using definite generic,

plural count nouns or mass noun references (Master, 1990). Drawing attention to such

differences can arguably help students understand the genres in their fields. Thus, when the

opportunities for learning lexical bundles that colligate with articles are also considered,

discipline specificity is indeed helpful for a focus on English article use. While English for

General Academic Purposes can focus on the colligations with the Academic Wordlist, there

are clearly more opportunities in ESP and in the words of Coxhead (2016: 186):

When it comes to specialisation through academic studies, it is important to find out at what

point it is better for learners with special purposes to start focussing on learning the specialised

vocabulary of their chosen field.

On the other hand, most summer presessional programmes in the UK are built around EGAP

principles. A focus on article use and its effect on genre is not impossible within a ‘core skills’

approach in which an EAP tutor is addressing a mixed discipline class of English for general

purposes. While less than ideal, tutors teaching on EGAP presessional programmes can

nevertheless give students the autonomous learning skills to consider genre and its effect on

article choice. EAP tutors can thus be thought of as facilitators who train students to find about

the genre requirements in their own disciplines for themselves (Flowerdew & Peacock, 2001;

Paltridge, 2001).

2.7 Empirical research into the effect of teaching and correcting articles

Having further developed the thesis that a focus on English article accuracy is compatible in

principle with modern approaches to EAP pedagogy, this section will further evaluate some of

the empirical claims in specific reference to 1) article teaching and 2) correction in pedagogical

practice.

2.7.1 The distinction between teaching and error correction

In this thesis, error correction denotes the term which the SLA field describes as ‘negative

feedback’, the fourth option within VanPatten’s (1993) focus-on-form model of 1) structured

input; 2) explicit instruction; 3) production practice and 4) negative feedback stages. It has

been pointed out (Ellis, 1998) that teacher practitioners sometimes conflate all these stages as

‘teaching’, for example corrective practices with explicit instruction, while researchers attempt

to separate processes into analysable components. In fact, it must be conceded that some error

correction is often combined when tutors check concepts and interact with students

spontaneously. However, this thesis generally distinguishes teaching (structured input and

explicit instruction stages) from the written error correction provided after students have

produced and submitted written work.

2.7.2 Empirical research into explicit instruction

Peter Master reported that the teaching of articles produced significant improvements in

students’ article accuracy (1986, 1990). For Master, the key was simplifying the many complex

usages and rules of articles into a simpler ‘binary system’ in which students were asked to

construe noun phrases as either classifying or identifying a referent. However, Master’s

research methods lacked control groups or post-study measures and many of his manuscripts

making the strongest claims were not peer reviewed.

2.7.3 Empirical research into error correction

Far more research has been published into the efficacy of error correction than explicit

teaching. A number of studies have claimed a significant effect upon article accuracy from

instructing ‘article-less’ students about article use through ‘consciousness raising activities’

involving error correction (Master, 1995; Master, 1997). However, notwithstanding the

mixture of Chinese L1 with other groups, these studies all failed to use control groups or

properly account for other variables that contribute to accuracy/error. A lack of control groups

in much of the literature is in fact a critique made by the main proponents of error corrections

(Ferris, 2006; Ferris & Kurzer, 2019). Nevertheless, a number of studies evaluating error

correction with control groups have shown significant improvements in article use (Chandler,

2003; Bitchener & Knoch, 2008). The latter two researchers found significant improvement in

the progress of lower level learners’ use of the definite article for anaphoric reference with

highly focussed oral and written corrective feedback on this error.

A major criticism of grammatical error correction research is that the extremely focussed and

considered correction of errors in the research environment has little relation to the real world

process of error corrections (Truscott, 2004). Truscott (1996) has long argued that grammar

error correction is ineffective, incongruent with theories of L2 learning, overly time-

consuming and even ‘has significant harmful effects’ (Truscott, 1996: 328). Truscott (2004)

indeed continues to critique other flaws in the methodology of Chandler’s (2003) and Ferris’

(2006) studies, most notably the difficulty in attributing causality to observed decreases in

errors. It is indeed extremely difficult to show clear evidence of causality, as highlighted in a

study by Crosthwaite (2016b) who found a significant improvement in the accuracy of articles

among students doing an EAP course, yet concluded that task prompt effects, which may have

affected accuracy rates, made the results inconclusive.

Given the evidence that feedback on grammatical form errors can be discouraging (Truscott,

1996; Polio, Fleck & Leder, 1998), in addition to the inconclusive evidence regarding the

effectiveness of correction (Krashen, 1992, Truscott, 2004), there is a compelling argument

that indiscriminate article error correction may be a less deserving of tutor and students

valuable time. There are clearly circumstances in which a tutor is obliged to correct occasional

errors as part of the teaching process (Hyland & Hyland, 2006). However, while there is a case

for focussed article correction and there will be contexts in which a tutor is compelled to

correct articles errors, such as in preparation for high stakes assignments or paper submissions

in which error correction is justifiably ‘product focussed’, this thesis makes the assumption that

overcorrection of article errors is ineffective. Thus, when dealing with more regular writing

tasks in EAP there is insufficient evidence to support the hypothesis that mass article error

correction improves accuracy, so teachers may wish to spend their limited time and resources

on more proven areas of pedagogy (Truscott, 1996, 2004; Crosthwaite, 2016b; Crosthwaite

2017). On the other hand, there is a clear need for more highly focussed and effective conscious

raising of English article functions (which may include error correction as a modelling for peer

and self-correction), particularly if it such focus can be integrated within a contemporary text-

based/data-driven student centred approach.

2.8 Research Questions

This chapter’s review of the literature has highlighted the imperative of addressing four

important Research Questions, as now summarised.

2.8.1 Research Question 1

To what extent do EAP teachers in Higher Education programmes in the UK explicitly

teach or correct English article use?

While a strong case has been made for incidental focus on article use through the ‘EAP

paradigm’ and against indiscriminate mass article error correction, there are clear indicators

that much pedagogy related to article accuracy falls within more traditional approaches to

grammar teaching. The review of the literature showed some controversy about the

teaching/correction of grammatical errors. Based on this this literature review and anecdotal

reference, it has been suggested that there may be a tension between EAP theory and the actual

practice of teachers. Anecdotally, the researcher has witnessed the wide-spread error correction

of article errors at a UK university, but only one UK based survey was found to have

investigated EAP tutors’ methods or attitudes to grammar teaching and this study is now 16

years old, and thus pre-dates the huge influx of Chinese students into UK higher education.

Whether and how EAP teachers teach the article system is therefore the first Research Question

addressed in the survey presented in Chapter 4.

2.8.2 Research Question 2

What are the linguistic contexts of obligatory article use in which Chinese L1 students

at Upper-Intermediate levels of English L2 are most likely to make errors?

Given the conflicting research findings presented in this chapter, the extent to which Chinese

L1 university students at B2 level (the minimum level for university entrance in the UK)

continue to make article errors will be investigated. Crosthwaite's finding (2016a) that Chinese

L1 learners displayed fairly accurate use from intermediate levels was contrasted with some

studies showing less accurate use even at high levels. It is critical that material

designers/syllabus designers have a better picture of the accuracy rates of university students

in various contexts. In addition to simple frequency, the second Research Question also

requires an examination of the relative accuracy with which the three article choices (a/the/Ø)

are made. More specifically, beyond the issue of a morpheme hierarchy of accuracy, the thesis

will investigate the extent to which Chinese students arriving for a Presessional EAP

programme (with Upper-Intermediate L2 English level) display ‘Target-Like Use’ in various

contexts of obligatory article use as presented by the Bickerton Framework. In addition,

following Sinclair’s recommendation (1991) to use more corpus-driven inductive research

without framework assumptions, Chapter 6 will also investigate the simple linguistic context

in which the article is found less accurately produced by Chinese L1 students at clause level,

with a particular focus on the pre-modification of nominal groups showed to be characteristic

of academic writing.

2.8.3 Research Question 3

To what extent does the Chinese students’ L1 background affect (1) Target-like Use

and (2) their types of article error?

As discussed in this chapter, accuracy of article use may be linked not only to developmental

proficiency, instruction and phraseology, but also to L1 interference. While the research

evidence for an effect of L1 interference is often presented as overwhelming in studies that

compare accuracy among [+Article] and [-Article] L1 background students, the literature

review has shown a more complex number of hypotheses for L1 effects. As shall be presented

in later chapters, hypotheses centred on the largest differences between Chinese and English

shown by contrastive analysis, i.e. the absence of any classifiers, demonstratives or quantifiers

in Chinese non-referential contexts (including generic) contrasted with the many different

article forms used in English in such contexts.

2.8.4 Research Question 4

Does the explicit teaching or correcting of English article usage lead to more target-like

use of articles by Chinese L1 university students whose L2 English displays

inaccuracy?

While it has been shown that students value and expect grammatical corrections, the evidence

regarding whether long-term effects are achieved through this lengthy and time-consuming

process is mixed. Research Question 4 is an empirical question (requiring an experiment).

Building on the findings in Chapter 4 (showing the extent to which EAP tutors teach/correct

article use), it needs to be asked whether such teaching or correction serve any purpose in EAP.

2.8.5 Conclusion

This literature review has revealed areas of both agreement and disagreement in the literature

regarding the place for a focus on English articles in EAP. The argument has been developed

that a focus on article forms could be fully compatible with an integrated genre approach to

EAP if teachers had a better understanding of the challenges faced by different learners and

familiarity with methods that were more effective. Answers to these research questions could

afford understanding of this type with regards to Chinese learners, the overwhelmingly

largest single L1 group at most UK universities. If it can be shown that the article system can

be taught/learned, pedagogy and materials development could then focus selectively where

most needed/helpful for such L1 groups most represented in university presessional EAP

courses.

3 METHODOLOGY

The aim of this chapter is to introduce the background to the methods of the studies referred to

in this research with particular attention to the methodological considerations of error analysis

using a learner corpus. A detailed presentation of research methods is provided for each

individual study in Chapters 4, 5, 6 and 7. However, methodological issues are more critically

developed in this chapter in reference to Corder’s (1967) five steps of error analysis. As shall

be shown, decisions about whether to include dispreferred forms such as marked

overspecification of definite articles were particularly important due to the ‘covert’ nature of

many article uses and genre effects in all judgements (Ellis & Barkhuizen, 2005). Through the

presentation of a preliminary study at the end of this chapter, it will be demonstrated how the

methodology evolved. More reliable and valid methods of error identification, annotation and

reliability checking were facilitated by the trialling of alternative solutions with a small group

of students before using the methods presented in later chapters. While the steps taken to ensure

methods were ethical are presented in each individual study, this chapter also discussed the

critical issues needing consideration. Most fundamentally, one issue faced was achieving

informed consent with fee paying students who were sometimes taught by the researcher. As

discussed at the end of this chapter, the University of Birmingham Ethics Board helped find

resolutions to these issues. The guiding principles of giving students enough time to refuse

consent and not disadvantaging students by non-participation informed the design of all

methods presented in this thesis.

3.1 Critical methodological issues in Error Analysis

Error analysis is affected by decisions of data collection, error detection, error classification,

error explanation, and a consideration of error gravity (Corder, 1967). Thewissen (2015)

provides a critical summary of the many methodological errors that can invalidate research by

reference to Corder’s (1967) five steps in error analysis and inferences from Thewissen’s

survey of issues that can invalidate error analysis are presented below in Table 3 before being

further evaluated in Sections 3.1 to 3.4.

Table 3: Error Analysis Steps/common methodological issues

Error Analysis Step (Corder, 1967)

Key methodological issues identified by Thewissen

(2015)

1. Data collection

• Under specification can invalidate research:

importance of reliably reporting genre, task-type, text

numbers, text lengths, corpora size).

• Research sometimes does not transparently and

reliably report the proficiency level.

2. Error Detection

• Does not account for complexity of error concept – a

relative concept with blurred border between actual

errors and dispreferred norms, mistakes/errors or

does not convincingly show how covert errors are

dealt with.

3. Error Classification

• Is a descriptive (taxonomies or surface structures) or

explanatory error classification system appropriate?

• How errors are quantified (e.g. obligatory occasion

analysis, T-unit analysis, proportion of part-of-

speech)?

4. Error Explanation

• Not accounting for Errors of performance (knowing

whether the ‘errors’ are simple mistakes or genuine

competence error).

• Lack of control of other factors causing error

(proficiency).

5. Error Gravity

• Are errors important?

• Do errors affect comprehensibility? How is non-

acceptability distinguished from minor irritation?

3.1.1 Data collection

Traditional error analysis methods relied upon judgement tests of grammatical acceptability

using invented prototypical examples, whereas computer-aided error analysis (CEA) can use

more authentic corpora of learner data (Granger,1996). The main criticism of the former

traditional methods is that they only allow the researcher to find predicted answers to questions

defined by assumptions of the researcher which in linguistic contexts may be less representative

of authentic academic language. Acceptability judgement tasks are therefore generally less

valid than corpus-based approaches. Nevertheless, the testing approach does have a number

of strengths, as outlined below, which is why the method was used to complement corpus-

based approaches in the latter stages of the research.

One major strength of the acceptability judgement task approach is the ability to focus on

defined areas of L2 language exposure in addition to the possibility of controlling factors such

as proficiency level (Ionin, 2012). Granger (2011) argues that large learner corpora can

overcome problems of variable control and allow for the study of rare forms of language use,

but this assumes corpora are over 1 million words in size which was only true of all the subsets

of the International Corpus of Learner English (ICLE) corpora combined together. Moreover,

comparing and aggregating findings from different corpus studies can be problematic since as

yet very few learner-corpus based studies transparently report variable specifications

(Granger, 2003; Díez-Bedmar, 2015; Thewissen, 2015; Crosthwaite, 2016a). Without

precisely controlling and reporting genre, task-type, text numbers, text lengths, and corpora

size the ability of other researchers to replicate or compare studies is limited. In particular,

Thewissen (2015) stresses the need to accurately define and measure proficiency levels and

notes that some stated proficiency levels (such as those in ICLE) have been shown to be

unreliable. This caveat is shared by Díez-Bedmar (2015) who points out that many error

studies are limited by using only institutional membership as a proxy measure of L2

proficiency. Unfortunately, these issues of participants’ actual proficiency can also affect the

validity of results, since later explanations of error differences between groups are less

meaningful. As Thomas (1994 as cited in Thewissen, 2015: 32) highlights:

It is vital to control for levels of proficiency in studies of the acquisition of a given L2 by

different groups of learners […] If the two groups do not start out with comparable skills in L2,

the research pay spuriously attribute differences in knowledge or performance between the two

groups to their differences in L1.

The next variables to consider in choosing the data (the ‘corpus’) in which to investigate errors

are genre, task-type, text numbers, text lengths, and conditions in which texts are written. To

compare one study with another or generalise from a small study across the wider population

requires consideration of these variables as well as careful design and clear reporting. For

example, many lexical-grammatical features found in error-analysis may be subject/topic

specific. For this reason ICLE (Granger, 2003) standardised the tasks across studies meaning

that each new researcher could replicate these variables when applied to other L1 language

background students. Yet this approach can sometimes be criticised for the lack of authenticity

given to the texts investigated in learner corpora (Leedham, 2015). Considering task type also

encompasses the question of whether students planned their work to the same extent, faced

similar writing conditions (e.g. reference materials or help were controlled), and had similar

access to technological aids. Thus the extent to which a researcher can generalise any findings

from a corpus study to the wider community depends on controlling a number of variables

(which has resource and time implications) while a tension arises between ‘control’ and

‘authenticity’.

3.2 Error detection

The first major challenge related to any analysis of article errors is that an error can sometimes

not only be attributed to an ‘error’ of omitting the article but also the alternative possibility of

being a simple ‘slip’ (careless mistake). This issue is heightened in the analysis of English

article errors because an incorrect use can occur from using a singular rather than plural noun.

For example, students may have missed a final –s on a noun.

Assuming that language is used by a learner consciously, the identification of errors is also

affected by the way the researcher defines an ‘error’. Corder’s (1967) notion of

‘grammaticality’ suggests that researchers focus on language that would be impossible for a

native speaker to accept. The first controversial issue is therefore which L1 speaker is chosen

in light of the polemic debate about the linguistic imperialism against emerging forms of

English. However, rightly or wrongly, all published studies seem to return to the notion of

‘native’ speakers’ correct use as the measure of accuracy and this thesis makes this same

assumption in full acknowledgement of the debate as to what actually constitutes ‘native’ status

(Davies, 2003, Brown, 2014). Combining both ‘grammaticality’ and ‘acceptability’, Lennon

(as cited in Ellis & Barkhuizen, 2005: 56) offers the following definition of error:

A linguistic form or combination of forms which, the same context and under similar conditions

of production, would, in all likelihood, not be produced by the speakers’ native speaker

counterparts.

3.2.1 Dispreferred forms

The notion of ‘dispreferred’ judgement (Thewissen, 2015) relates to whether an error is

marked but possible, or categorically impossible and inaccurate. This is a challenging issue

because of its subjectivity at more advanced levels of English. Moreover, in terms of the issue

of ‘academic style’ (when students use overly informal English), this is greatly affected by the

given genre. That is to say, it is agreed in the EAP field today that there are huge variations of

genre that may be broadly referred to as ‘academic writing’, indeed differences in register and

conventions often occur within report/assessment genres of the same university department

(Hyland, 2012; Nesi & Gardner, 2012). To agree on the identification of error, annotators will

therefore need not only a similar view of grammaticality but a similar appreciation of the genre.

3.2.2 Competence and performance errors

Error analysis should ideally be focussed on true systematic errors that are being made because

of insufficient knowledge (competence errors) of the L2 rather than slips on the keyboard. To

some extent, this relates to the control of variables such as motivation or time allowed on a

task, but a thorough methodology may also involve checking with students about whether they

only made a careless ‘slip’. All L1 language users make ‘mistakes’; any meaningful error

analysis must therefore be focussed on true systematic errors that are being made because of

insufficient knowledge of the L2 rather than slips of the pen/tongue (Corder, 1967). Some

authors define error as only one that learners simply cannot comprehend. Sowton (2012), for

example, distinguishes between ‘slips’, mistakes and errors, relegating inaccuracies that the

learner can understand with teacher prompt as a ‘mistake’.

Three possible solutions to this challenge (of confusing mistakes with errors) would appear

possible. The first would be to give the study participants the opportunity to check their

errors/mistakes and self-correct (assuming that mistakes would by definition be corrected on

review). Another would be to use large quantities of data in the hope that the ‘mistake’ effect

on results will be insignificant. Student could finally be made to write texts in highly controlled

conditions. Admittedly, this last solution is likely to become more necessary in today’s world

of online grammar checkers and improved word-processor functionality. However, in the early

years of the data collection for this research project these online tools were less sophisticated

(2010 – 2014). In fact, no solution to this challenge is perfect, since controlled conditions

without grammar tools and dictionaries are less representative of the modern realities of

academic writing. Checking of errors (for example by interview) is time consuming and is also

problematic as errors might often be changed simply by the attention being focussed upon

them. The frequency solution is also imperfect as it ignores the fact that some mistakes may

be systematic and frequent especially when using a keyboard to type written language. Some

efforts are made in Chapter 6 to tentatively examine this issue through a limited number of

interviews, but the main solution adopted in this research is the first (allowing students time to

review their work). The basic assumption made in this thesis is that students proofread their

work and used word-processing tools to the extent that most errors are genuine.

3.2.3 Overt versus Covert errors

One issue in checking the reliability of the first researcher’s identification of error, as shall be

highlighted in Section 3.7.4, is that annotators cannot make judgements in decontextualised

sentences. The whole text may need to be studied to look at possible implicit first referents or

implied meanings, especially in the case of anaphoric reference in articles. Thus, Ellis &

Barkhuizen (2005: 56) point out that the annotator is not looking only at overt errors, but covert

errors

An error is said to be overt if it can be detected by inspecting the sentence/utterance in which

it occurs. An error is covert if it only becomes apparent when a larger stretch of the discourse

is considered.

3.3 Error classification

The researcher is faced with two choices when analysing data, either choosing to apply a pre-

determined classification/explanatory system, i.e. a deductive research approach, or adopting

a more open-minded inductive research approach, i.e. simply looking for patterns of error

without any prior assumptions. Sinclair (1991) argued that corpus-driven inductive approaches

allow for analysis of patterns that were not predicted in a previously delineated framework. As

shall be presented in Chapter 6, the use of computer programmes such as the freeware

AntConc toolkit for concordancing and text analysis allows for the analysis of the key word

in context (KWIC) view and the application of simple left and right searches around the

keywords (or error tags) to look for common error patterns. As Groom and Littlemore (2012:

164) point out:

The appeal of the concordancer is that it allows you to sort data alphabetically to the right or

the left. This in turn allows you to analyse your data more efficiently and to see otherwise

hidden patterns.

Another advantage of inductive approaches over the application of a previously published

framework is that it is simpler. Classifying semantic environments of noun phrases inevitably

involves an element of subjective judgement. In Butler’s (2002: 455) words, the application

of a deductive method, for example using the Bickerton Framework to authentic student work,

is not without challenges:

Although some contexts may be quite straightforward to classify, others are not. Quite often

researchers have to make assumptions or guesses about semantic contexts and then try to judge

whether or not the articles were used correctly based on these assumptions.

Nevertheless, as will later be discussed in the preliminary study, this research showed that the

Bickerton Framework could be reliably used with sufficient annotator training.

3.4 Error explanation

In terms of explanation of errors, as discussed in the preceding chapter, the earliest traditional

error analyses in the era of Contrastive Analysis (during which time behavioural linguistic

assumptions informed all explanations) simply assumed that all errors were caused by language

transfer effects. Yet by the 1970s most linguists agreed that many errors made by L2 language

learners were developmental (intralingual), whether or not the order of acquisition of L1 and

L2 learners of a language was identical. The modern approach taken by this thesis is to make

more cautious/tentative speculations about L1 interference within the greater interplay of

proficiency level, genre, task, instruction and other such variables affecting accuracy.

3.5 Evaluation of recent methods in English article studies

Before evaluating the various choices of method, it is worth quickly reviewing the

methodological choices made by other researchers. Table 4 overleaf shows some of the

methods in recent influential article error studies, which include ethnographic study, testing

through prompts, testing through grammaticality judgement tasks, contrastive interlanguage

analysis with learner corpora, and experimental/control studies (to test the effect of teaching).

The evaluation which follows focusses on the risks and limitations of each study in the context

of English article acquisition research.

3.5.1 Ethnographic study methodology

In the context of English article acquisition research, this qualitative method is the most

effective way of explaining learner errors and the stages of their development. The insights

can be extremely rich as shown by Huebner’s (1983) hypothesis of a ‘flooding’ stage in the

English article acquisition of [-Article] L1 background students. However, this method is

extremely time-consuming, and the conclusions can seldom be generalizable without further

investigations. Another difficulty is understanding student utterances and the laborious process

of transcribing data into a manner that can be interpreted.

3.5.2 Testing through prompts

Employed by the 1970s child acquisition studies, these methods focus on what the learners

are confident to use at the current stage of development (recording, transcribing and then

analysing naturalistic conversation arising from the prompts). Results can be reliably

Table 4: Review of influential article acquisition/corpus-based studies

Researchers

Learners

Level

Method

(Yamada and

Matsuura, 1982)

Japanese

Upper –

Intermediate

level

Cloze tests in which participants asked to identify the

missing articles in obligatory contexts

(Huebner, 1983)

Hmong

Complete

beginner

Study of the natural acquisition of English by one Hmong

speaker (Ethnographic)

(Thomas, 1989)

Various

Various levels

30 learners given oral elicitation interviews with picture

prompts

(Master, 1995)

Chinese and

Japanese

Advanced

Measured effect over 3-4 weeks of consciousness-raising

techniques (interviews, written corrections and students

asked to keep journal of errors)

(Butler, 2002)

Japanese

Various levels

Investigated accuracy with gap fill test & learners’

metalinguistic knowledge through interviews

(Ekiert, 2004)

Polish

Various levels

Compared learner corpus compiled from ESL and EFL

settings

(Bitchener, Young

and Cameron, 2005)

East Asian

Various levels

Treatment and control groups used to compare students’

progress when given various forms of corrective feedback

(Díez-Bedmar and

Papp, 2008)

Chinese and

Spanish

Upper-

Intermediate

Learner corpora compiled from university students’ essays

(Crosthwaite, 2012)

Madarin

Korean

Various levels

Analysed referring expressions used in L2 narrative

discourse in Written Learner corpora

(Crosthwaite, 2016a)

Chinese

Korean

Thai

Various levels

Written corpora

generalizable, but an extremely expensive and time-consuming method if strict controlled

conditions are to be effectively applied. On the one hand, understanding and transcribing

participants’ responses is very time consuming. On the other hand, as Brown reflects (1973:

28), the difficulty of keeping up with answers can be mitigated by recording the students, and

such analysis has an important role:

It turned out, however, that a great deal could be learned from meticulous records of

spontaneous speech in combination with answers to simple questions.

3.5.3 Interviews

The interview method involves choices of whether interviews are structured, semi-structured

or unstructured and whether the data is analysed using an emic perspective analytical

framework developed before the interview process or an etic perspective analysis that evolves

from the interaction itself (Groom & Littlemore, 2012). As shall be further discussed in Chapter

6, the limited number of interviews conducted for this research lent itself to more student-led

interviews that could not be minutely structured. The approach was therefore taken to record

discussions about errors that students found most confusing, transcribe them and search for

patterns that could be analysed and quantified.

3.5.4 Testing through grammaticality judgement tasks

Judgement tasks that ask participants to rate the acceptability of grammatical forms have the

advantage that the researcher has control over variables and can provide measured

semantic/genre/linguistic contexts to participants. Thus, the test designer can focus on the

aspect of the English article system that they wish to investigate and control other factors. Such

control also means that the learner cannot use avoidance strategies and rarer grammatical

patterns can be investigated while such features may not naturally occur frequently enough for

study in corpus-based approaches. On the other hand, such tasks are less authentic than learner

corpora approaches which allow for analysis of the actual writing contexts that students face

in real life. For this reason, a mixed corpus-based and testing approach was used in this

research. Corpus methods are the only way of validly achieving a full picture of students’

control of English articles, but it became clear that several hypotheses required investigation

through a tighter control of the linguistic context of article use. This advantage of the testing

paradigm was particularly useful in the investigation of article overspecification and the effect

of pre-modification of noun phrases (as will be presented in Chapter 7).

As Chapter 7 will show, the testing approach can complement corpus-based research by

addressing issues such as dispreferred forms (i.e. overspecification of articles). Many

redundant uses of definite article found in natural student writing cause disagreement among

researchers/tutors. As shown by the example of a student’s ‘overuse’ (see Literature Review

Figure 4) of definite articles, subjective judgement means that it is difficult for researchers to

agree on which errors are completely inappropriate and which are simply marked. The testing

approach can present errors that are unanimously considered as errors and also avoid challenges

of ambiguous error (caused by ungrammatical language) that can also be found in the authentic

writing of many students.

3.5.5 Corpus linguistic approaches

Sinclair (1991) defined any corpus as ‘[a] collection of naturally occurring language text,

chosen to characterize a state or variety of a language.’ Through studying authentic learner data

electronically (a learner corpus) corpus-based research is often argued help EAP materials

writers develop better EAP pedagogy/resources (Johns, 1991; Sinclair,1991). Indeed, Chuang

& Nesi (2006: 251) conducted their analysis of Chinese undergraduate writing ‘… with a

view to creating remedial grammar materials for Chinese students studying in the medium of

English.’

In the Sinclair tradition, an inductive corpus-driven approach analyses data with fewer a priori

assumptions. Other more deductive approaches can involve the use of a previously established

framework or hypotheses. Whether deductive or inductive in approach, corpus-based research

clearly offers the most valid insights into student writing when using authentic student data.

The advantage of using truly authentic data (Chuang & Nesi, 2006; Lee & Chen, 2009) is that

genre effects on error rates can be analysed. Particularly for post-graduate level students, the

longer word length of the dissertations in data collected for these studies is more representative

of the academic writing such students must submit in their academic programmes.

3.5.6 Contrastive Interlanguage Analysis

Granger’s Integrated Contrastive Model (ICM) (1996) methodology is a corpus-based

approach that involves the analysis of authentic data through computerised corpora. In short,

this more deductive method involves the study of the L1 and building of hypotheses before the

analysis of learner data. To go beyond the identification of errors, Granger argues that the

explanation of errors can be achieved by a combined contrastive analysis (CA) and corpus-

based approach (the study of L1, L2 and even L3 comparative use). Without interlanguage

analysis through corpus-based studies, CA analyses are simply predictions with no evidence.

However, vice-versa, little interpretation of interlanguage studies can be made without some

form of CA (ibid). Borrowing from Jarvis’s hypothesis of transfer effects (2000), Gilquin

(2008) presents a Detection-Explanation-Evaluation (DEE) transfer model which requires

stages of L1/L2 Contrastive Analysis, stages of interlanguage analysis, and stages of analysing

other language uses of the L2 in order to quantify and explain the crosslinguistic interference

process.

Whether or not the approach preferred is deductive or inductive, comparisons of any corpus-

based research requires a critical understanding of the data collection and error identification

methods used. For example, a review of five recent corpus-based studies (see Table 5) shows

some marked differences in the authenticity, genres, text length, corpora size and backgrounds

of students. The advantage of the use of short essay prompts (Diez Bedmar & Papp, 2008;

Table 5: Five learner corpus studies compared

Study

Data

Word

length

Corpus size

Student level &

Discipline

Error

Classification system

Chuang &

Nesi (2006)

Authentic

EAP

Writing (essays)

1500-

2,000

words

50 essays

(75,000 –

100,000

words)

Foundation

B1-B2 level

Business studies

Foundation students

Own error categorisation

system:

(omission

overinclusion

misselection)

Lee & Chen

(2009)

Authentic EAP

writing:

dissertations

5,000

words

dissertations

(390000

words)

Undergraduate

Unspecified

English level

Linguistics/Applied

lingusitcs

Keyword analysis of

overuse and underuse)

Díez-

Bedmar &

Papp (2008)

ICLE

argumentative

essay prompts

Mean

average

540

words

74 Essays

(39, 663

words)

University students

in 2

year or later

UG/PG level,

IELTS 4.5 – 7.0

Various

Bickerton Framework

Crosthwaite,

(2016a)

Two

argumentative

short essay

prompts from

ICNALE corpus.

Mean

word

length

217 –

250

words

575 essays

(125. 588

words)

4 levels: (A2-

B2/C1) Various

Bickerton Framework

Crosthwaite,

(2017)

Essays or reports

(assessed EAP

compositions).

800 – 1500 words

794

words

mean

length

257 essays and

reports

(205,683

words)

year

undergraduates

Various

Annotation of multiple

language errors. System

adapted from Dahlmeier

et al. (2013, as cited in

Crosthwaite, 2017)

Crosthwaite, 2016a; Crosthwaite, 2017) is that more reliable comparisons can be made across

proficiency levels by students who have answered the same prompt. There is also variation in

the measures of proficiency and Crosthwaite’s (2016a) study can be seen to be the only study

that reports accurate proficiency rates across four levels (the ICNALE corpus required all

students to do an Oxford Quick Placement test).

3.5.7 Experimental/control studies

The empirical question of whether explicit teaching or correction of articles helps learners to

improve (Research Question 4) can only be investigated by experimental methods, as shall be

detailed in Chapter 5. As shall be discussed, studies that use an experimental paradigm must

control all variables apart from the ‘treatment’ and then seek to show the effect was the result

of the ‘treatment’ through the use of control groups. Ionin (2012: 31) stresses the importance

of controlling variables in order to investigate learning over time:

By keeping constant such factors as native language, proficiency level, type of L2 exposure,

and age of acquisition, among others, an experimental SLA study can reduce variability among

learners, and determine how learning proceeds within a given population.

3.5.8 Critical evaluation

The main advantage of corpus-based contrastive interlanguage approaches is that correct use

can be measured (and annotated) in addition to incorrect use. When combined with coefficient

measures such as Pica’s TLU (Pica, 1983), which factors both omission and overuse in addition

to correct use, quantitative measures can become more meaningful and comparable (Díez-

Bedmar, 2015). Moreover, by using full texts written by the participant on topics relating to

their academic subjects and in authentic genres found in academic writing, error analysis can

have greater validity. In contrast, grammaticality testing techniques typically use contrived

language that may neglect the language and choices that learners actually make in academic

writing.

The main advantage of simple cloze or multiple-choice testing approaches is that they

circumvent many of the methodological problems of ambiguous error, mistakes/error issues

and ungrammaticality of responses. Since the test takers are typically focussed on accuracy,

this avoids some of the issues of competence/performance errors. Moreover, the researcher

can give choices that he/she knows are categorically correct, avoiding some of the issues of

error identification that can arise, especially when analysing lower proficiency learner data

(which may be too ambiguous/ungrammatical to categorise). The greatest advantage, however,

of a simple testing approach is that the researcher can investigate large groups of learners with

lower resource implications.

When comparing the validity of corpus-based and testing methods, however, it is clear that

testing methods tend to exaggerate the inaccuracy rates found, since the test designers will

focus on error types that they know are challenging, while many uses of articles that the learners

find straightforward will not be reflected in the ‘rate of accuracy’. On the one hand, Butler

concludes (2002: 455) that testing allows for research into areas of grammar that students avoid

in their own writing. However, Granger (1996, 2011) makes a convincing argument that the

study of learner corpora provides greater validity as it allows the analyst to study natural

authentic language that students actually use at university. Although a testing approach allows

large-scale studies that can produce significant results, cloze or multiple-choice questions use

contrived language that may neglect the language and choices that learners actually make in

authentic academic writing. In fact test questions represent the ‘tricky’ words and findings from

a testing approach may be dramatically affected by the researchers’ choice of question in a

parallel way to the effect of corpus-studies from choice of prompts.

Given the issues regarding both methods discussed above, this research used a mixed Corpus-

based and testing methods approach. A Corpus-based methodology that used both deductive

and inductive approaches was first employed to answer the question ‘what errors do Chinese

L1 students make?’. A testing approach was then employed to further investigate effects of

linguistic context and proficiency. While a full explanation of errors was outside the time

constraints of this research, student interviews were also planned to give qualitative insights

into the data.

3.6 The need for a preliminary study

In the light of the critical importance of the control of variables, the decisions to make in error

identification, and the ethical review of all procedures it was necessary to conduct a preliminary

study to ensure compliance and reliability. Methods in the research of a learner corpus can be

seen as an evolving process (Chuang & Nesi, 2006) as shown in Figure 5 as preliminary studies

Figure 5: Reproduced from Chuang and Nesi (2006: 257): 3 stages of developing a tagging

system

can investigate and evaluate various methods in the error identification process through

trialling alternative approaches with a small amount of data from a small number of students.

3.7 Preliminary study

Participants in the trial study were drawn from three classes in a 42-week Presessional course

at the University of Birmingham. A small number of students (mostly Arabic L1 and only one

Chinese student) participated in five extra ‘article use’ lunchtime lessons and extra online

quizzes in the first year of the project to trial several approaches to error identification and

other methodological procedures. The effect these workshops had on the students’ accuracy

of article use was measured by the analysis of their essays over the course of the term. Although

this data was collected and analysed, the findings of this preliminary study are not presented in

this thesis (which focusses on Chinese students). However, the way in which the

methodological approaches to error identification, classification, reliability checking

procedures and ethical recruitment evolved as a result of this first trial (with impact on later

studies) is outlined below.

3.7.1 Evolution of ethical safeguards

The main ethical concern in all studies in this thesis centred on the use of students as

participants. Serious consideration was given to the risks of demotivating the students;

impacting the students educationally from the point of view of participation or non-

participation; and students’ rights to confidentiality, integrity, respect and privacy (Phakiti,

2014). Any study recruiting study participants from among the researcher’s own teaching

cohort adds risks and limits the degree to which participation can be voluntary if 1) the research

takes place during normal scheduled teaching hours (at which times the student is expected

to attend) or 2) the students find it difficult to refuse their teacher’s request.

After gaining permission from the Director of the English for International Students Unit

(EISU), the next step in the preliminary study was to apply for ethical review. All research

involving human participants must go through the three stages of the University of

Birmingham’s Ethical Review procedure (University of Birmingham, 2016): 1) online initial

assessment stage; 2) application for Ethical review 3) responses and adaptation of methods to

gain approval from the Ethical Review panel. In fact, the first original contact with the review

board was the start of many negotiations as the methods in the research project evolved and

amended applications were submitted for later studies ( ERN_10-0375/amended 14-0375,

ERN_14-1455, ERN_16-1493 & ERN_17-1549) and this evolution will be discussed in greater

detail in the next section.

3.7.2 Ethical review of Corpus/experimental studies

Recruitment

The main ethical concern in this study centred on the use of students as participants. In light

of the risks of demotivating the students, the educational impact of participation and fairness,

the ethics board asked that all research activities took place outside the teaching timetable

(normally on Friday and Wednesday afternoons or at lunchtime). A small number of students

were 17 years old and it was not clear if they would legally be allowed to give informed consent

without the involvement of their parent(s)/guardians. The decision was therefore taken to only

seek participation from students who were over 18.

Consent

In order to ensure that consent was ‘informed’, students had to be given full information about

the study, its objectives and their right to non-participation. To achieve informed consent, the

decision was made to 1) tell students about the research projects in person and distribute an

information sheet, 2) give the students 24 hours, and 3) ask students to sign consent forms only

after they had fully understood the research project and their right to non-

participation/withdrawal. Based on requests from the ethics board, the students’ right to not

participate was made more prominent (see consent forms in Appendix 4). Participants were

free to withdraw from the project at any time before or during the study. Participants were also

given the freedom to withdraw their results within a year after the experiment was conducted.

Information given to participants

Students were not deceived in any way. They were told the focus of the research, how they

would be asked to help in the research and about their rights to withdraw/refuse consent at all

times. The university Ethical Board took a special interest in the feedback measures.

Highlighting the risk that students could be negatively disadvantaged by not participating in

the project, the board asked for an additional session to be made available to all students

regardless of participation at which the main points and preliminary research findings were

shared. Therefore, all students on the course (regardless of participation) were invited to end

of study briefings and sent reports of preliminary findings.

Storage of data

Recognising the student’s rights to privacy (and the researcher’s responsibilities under data

protection law), all data was stored electronically on an encrypted network drive backed up on

an encrypted flash disk held in a secure filing cabinet on university premises.

3.7.3 Ethical review of online testing with cloze/multiple choice tests

A number of quizzes were presented to the students on the same online learning environment

(Canvas) used for their main assessments throughout this research. Thus, there was a risk that

students would either not see participation in research as voluntary or do the online quizzes

accidentally. The Ethics Board’s main concerns were 1) that no confusion occurred between

what students were mandatorily required to do as part of their programme and the research

project and 2) that they could do the online quizzes and receive feedback regardless of their

participation. A suggestion to put the consent box at the start of the quiz was rejected because

students might have felt they could not do the quiz and benefit from its feedback. The solution

found to these concerns in the preliminary study and all future ethics applications (ERN_16-

1493: ERN_17-1549) was to make it clear to students that quizzes were non-mandatory and

request consent to use their data some days after completion (giving the students informed

consent without implying their non-participation would stop them seeing valuable quiz

feedback).

3.7.4 Evolution of error identification, classification and reliability checking procedures

The presentation of the procedures for reliability checking is postponed until Chapter 5, in

which Section 5.6 provides details for how three colleagues checked a 10% sample of errors

identified. However, the preliminary studies contributed to the development of both the

reliable methods of error classification and this reliability checking system.

At the early stages of the research, an overly complex classification procedure involved the

annotation of article use according to both the Bickerton Framework and Quirk & Crystal’s

(1985) descriptions of cataphoric reference and post-modification, logical use with adjectives,

unique reference, anaphoric reference and immediate situational reference. However, when

the first researcher’s classifications were compared with three other researchers, disappointing

levels of inter-rater agreement were found. This highlights the problems in applying a complex

explanatory system to language (Taylor, 1989). Given that the application of the Bickerton

Framework was challenging in of itself, the main studies eventually focussed only on this one

Framework (without further categorisation of 2DA definite articles).

The preliminary studies discussed here also led to the evolution of more robust inter-rater

checking procedures. Preliminary efforts with the Bickerton Framework alone failed to

produce sufficiently high Fleiss Kappa coefficients of inter-rater reliability. However, these

poor initial results were shown to be the result of the tagging review interface. For the first

attempt, an online interface had been built on a Virtual Learning Environment in order to

organise the training and annotation checking. This online interface presented the sampled

noun phrases (from the preliminary corpus dataset) in a standardised form to the three

annotation checkers. These colleagues were asked to apply the Bickerton Framework/judge

correctness using the online interface without sight of the researcher’s judgements. However,

this online system resulted in low Kappa coefficients of inter-rater agreement. On closer

analysis, it was discovered that the checkers 1) lacked sufficient training and 2) were finding it

hard to make judgements without the full context of the student response. In fact, checkers did

have the opportunity to review the whole essay/response with a click of a button in the first

online interface, but they found this procedure time-consuming and cumbersome. In later

studies, a simpler paper-based approach achieved more reliable results. Checkers were given

more training and the whole texts in printed form with highlighters to show their coding

decisions. The inter-rater agreement went up substantially (see 5.6), which shows that

decisions about semantic/pragmatic meaning are best made holistically without the

decontextualization that had mistakenly been applied in the first approach.

3.7.5 Challenges identified for the experimental study

The first methodological challenge highlighted by the preliminary study was the control of

variables such as class teacher pedagogy and error-correction. Participants in the preliminary

study were drawn from three classes which gave students 9 hours of academic English and 12

hours of general English each week, delivered by between two and four teachers for each class.

As the students were using General English textbooks which covered article use, it is likely

that some of the students were receiving instruction from their normal classes in article use.

Thus, participants’ article accuracy rates were being measured while taking part in five extra

lunchtime workshops, but it was impossible to attribute their progress to this extra treatment in

the preliminary study as the corrections and teaching in the students’ classes had not been

controlled. For this reason, the main studies presented in Chapters 5 and 6 made calls for

volunteers from the 15-week and 10-week Presessional courses which had only one class

teacher, far more structured syllabi more focussed on EAP and not general English. This

enabled the researcher to quantify and establish the effect of the normal teaching that the

participants received.

Another key challenge experienced at the pre-study stage was the issue of how English

proficiency level of students could be measured. With the participants’ permission (from the

consent form), students’ results in recent secure English tests could be accessed, which

normally related to Cambridge International English Language Testing System (IELTS).

Table 6 below (overleaf) presents descriptors and rough equivalents to the Common European

Framework for Reference to Languages (CEFR) from the British Council’s IELTS Guide for

Teachers (British Council, 2012). Most students entering university programmes at the

University of Birmingham are required to hold at least a B2 CEFR level of English proficiency,

equating to a recent IELTS score of 6 - 6.5 (Competent user) or IELTS 7 grade (depending

on programme). At Presessional level, students normally therefore enrol at the B1/B2 threshold

with IELTS 5 or 6 band scores and the presessional programmes aim to accelerate students’

progress to reach the equivalent IELTS level. While IELTS was a robust test taken under

controlled conditions, the fact that so many students on the 42-week Presessional were grouped

along a very narrow range of bands (generally 5.5 and 6) made it difficult to estimate the effect

Table 6: IELTS score descriptions (from IELTS, 2012)

Band score

CEFR

Description

9 Expert user

Has fully operational command of the language: appropriate, accurate

and fluent with complete understanding.

8 Very good user

Has fully operational command of the language with only occasional

unsystematic inaccuracies and inappropriate words. Misunderstandings

may occur in unfamiliar situations. Handles complex detailed

argumentation well.

7 Good user

Has operational command of the language, though with occasional

inaccuracies, inappropriate words and misunderstandings in some

situations. Generally handles complex language well and understands

detailed reasoning.

6 Competent user

Has generally effective command of the language despite some

inaccuracies, inappropriate words and misunderstandings. Can use and

understand fairly complex language particularly in familiar situations.

5 Modest user

Has partial command of the language, coping with overall meaning in

most situations, though is likely to make many mistakes. Should be able

to handle basic communication in own field.

4 Limited user

Basic competence is limited to familiar situations. Has frequent problems

in understanding and expression. Is not able to use complex language.

3 Extremely limited user

n/a

Conveys and understands only general meaning in very familiar

situations.Frequent breakdowns in communication occur..

2 Intermittent user

n/a

No real communication is possible except for the most basic information

using isolated words or short formulae in familiar situations and to meet

immediate needs. Has great difficulty understanding spoken and written

English.

1 Non-user

n/a

Essentially has no ability to use the language beyond possibly a few

isolated words.

0 Did not attempt the

test

n/a

The test taker did not answer the questions.

of proficiency on small sample sizes, as statistical measures could only show significant results

with an adequate number of students across different bands.

3.7.6 Preliminary study conclusions

The students in the preliminary study (including one Chinese L1 learners) made fewer errors

when given extra sessions on English article use. However, the aforementioned issues in

controlling all variables made it difficult to attribute causality to these extra sessions.

Moreover, it was not known whether students were making fewer article mistakes due simply

to their increased focus on ‘monitoring’ their language. The ambiguous nature of the

grammatical errors that students made in their writing occasionally made classification of

errors more challenging, but most nouns could be reliably annotated for error. One concern had

been that the involvement of B1 or borderline B1/B2 students would involve language so

ungrammatical as to make the intended meaning (including definiteness/ specificness) too

ambiguous to categorically state as correct or incorrect. However, the trial showed that such

occurrences (just eight noun phrases in this study) would not be a major problem with the

higher level proficiency of students who were to be invited to the subsequent main study.

Nevertheless, there were some particularly difficult decisions to make when the

overspecification of the definite article had created an issue of ‘register’ (academic style),

rather than grammatical accuracy. This indicated that a greater formalised training would have

to be given to annotation checkers to show how error had been defined. After replacing the

overly complex explanatory framework with a simplified Bickerton Framework, giving

annotator checkers more training and allowing them to make judgements of the whole texts,

the trial showed that researchers could agree a concept of ‘accuracy’ in grammatical terms and

apply it fairly consistently.

In addition to highlighting many challenges to be overcome, the preliminary study showed that

authentic student data from real students could inform the analysis of article errors made by

Presessional students. Reliable and ethical ways of collecting data and identifying errors in

were developed on the basis of this pre-study trial.

3.8 Summary of methodological approaches in main studies

Research Questions 2 and 3 were investigated with two main methods: a learner-corpus

approach and a grammatical judgement testing approach. As shown in Figure 6 below, after

identifying errors in the student writing (step 2), the annotated errors would need to be checked

for reliability (step 3). Step 4 then involved both a deductive corpus analysis (testing of the

hypotheses) and an inductive analysis (using a concordance). Step 5 involved a multiple-choice

test of a larger population to mitigate some of the constraints in the corpus-based approach and

further investigate some insights from the inductive approach that refined the hypotheses. The

aforementioned advantages of the testing approach (increased sample size, control of variables

and less ambiguous errors) allowed the last stages of the research to investigate some complex

questions less amenable to study in small corpus-based studies. For example, the effect of

proficiency could be investigated (the small corpus-sizes made statistical measures less

significant). However, given the limitations of any testing approach, this last stage in the

methods should not be seen as a test of the corpus findings themselves. Only the use of

authentic student data can give a picture of students’ accuracy in academic writing with any

validity as these tests would be affected by the limitations discussed in Section 3.5.4.

Figure 6: Summary of mixed method approach to Error Analysis

3.9 Conclusion

This chapter has critically evaluated some of some key methodological issues experienced in

the research project’s main corpus and testing approach studies. The main methodological

concerns relating to error identification (ambiguity/ ungrammatical language, dispreferred

forms) and annotation checking appeared to be resolvable. By applying for Ethical Review

and amending procedures to satisfy the university’s Ethics Board, it was possible for informed

consent to be gained ethically and several principles for safeguarding students’ rights were

established for the main studies (for which ethical approval was also obtained). The

methodological procedures of each study are further developed in each of the next chapters.

1.Preliminary

study to develop

error analysis

systems, control

of variables and

ethical review

2.Corpus led

identification of

errors

4. Testing of

hypotheses and

more inductive

approaches

3. Error

annotation

reliability

checking

5. Further

investigations of

hypotheses

using grammar

judgement tasks

(larger numbers)

Before investigating student error, however, the next chapter moves attention to the issue of

EAP tutors’ attitudes and practices with regard to English article use in academic writing.

4 SURVEY OF TEACHER ATTTITUDES AND PRACTICES

This chapter presents a survey undertaken to investigate the extent to which EAP teachers in

the UK Higher Education system teach the English article system. Before applying the corpus-

based and testing approach outlined in the previous chapter, the unknown context of whether

and how tutors teach the article system in EAP required investigation. As was highlighted in

Chapter 2, there could be a theoretical tension between the ‘EAP paradigm’ of teaching and

teaching/correction of English articles, especially when examples given are prototypical and

decontextualised from academic writing. The ‘EAP paradigm’, as outlined in the BALEAP

competency framework (Alexander et al., 2008), promotes at all costs the use of authentic texts

and a focus on the function of language rather than its form. There would therefore assumedly

be little support in this paradigm for isolated decontextualised exemplars from invented

language to show prototypical uses of English articles. With regard to written feedback,

Chapter 2 also showed how divisive the issue of written error correction is among the teaching

community. Certainly, mass indiscriminate English article correction is argued against by

many leading contemporary EAP researchers (Truscott, 1996, 2004; Ferris, 2006; Hyland &

Hyland, 2006; Alexander et al., 2008; Crosthwaite, 2016b). In order to investigate whether and

how tutors correct and teach article use, a survey and some qualitative interviews were

conducted within the EAP community.

Before presenting the survey itself the first part of this chapter briefly goes over some

definitions and highlights evidence that there may be a gap between the theoretical aspects of

the contemporary recommended practices in the EAP literature and what is actually happening

in many classrooms.

4.1 Background and recap of Research Question 1

Evidence was provided in Chapter 2 that the ‘EAP paradigm’ may not support either the

decontextualised teaching of articles or regular corrections of written article errors. However,

if EAP teachers are grouped together with other language teachers, the literature shows that

their practice is influenced by many factors beyond pedagogic theory and published best

practice. From a meta-analysis of 22 studies which investigated second language teachers’

beliefs and practices regarding grammar teaching, Borg (2003: 91) found that ‘language

teachers’ are shaped by a wide range of interacting and often conflicting factors’. In the EFL

field, in which many EAP tutors are trained before moving into EAP, Phipps and Borg (2009)

cite an example of a teacher using a highly decontextualised grammar ‘choice of article’

exercise using overly simplified examples even though the teacher himself believed a context-

based approach was more valid and appropriate. Several teachers cited in Phipps and Borg’s

study stated that they often used oversimplified controlled approaches such as gap-fill exercises

in grammar teaching in conflict with their own beliefs because the approaches were expected

by students. Furthermore, some tutors in Phipps and Borg’s study reported that their students

seemed to be more motivated when grammar was presented in such a way or even that they

used grammar to calm students down after discussions.

It is important to avoid assumptions about how EAP tutors teach article use in the classroom.

The fact that many experts of EAP teaching discourage excessive correction of surface errors

in student writing and insist that all teaching of grammar is incidental and based around text

meaning does not necessarily mean that such recommendations are followed. Investigating this

difference between stated beliefs and practice more specifically in EAP grammar teaching,

Etheringon & Burgess (2002) conducted a survey of 48 UK based EAP tutors in centres

affiliated to BALEAP about the tutors’ attitudes to a grammar teaching in EAP. Most teachers

in interviews expressed a theoretical preference for more discourse orientated approaches to

integrated grammar teaching (i.e. a focus on form when needed within a meaningful context),

rather than a decontextualised approach. However, these authors report that grammatical error

correction in writing was shown to be widespread even when it did not affect meaning. The

majority of tutors in Etheringon and Burgess’ study believed that error correction and formal

instruction helped students to improve their written accuracy, irrespective of their reported

belief in more content-based teaching approaches. Repeating the same questions in New

Zealand, Barnard and Scampton (2008) surveyed 32 EAP tutors and reported that four out of

five teachers believed that ‘form-focused correction helps students to improve their

grammatical performance’ (ibid: 59). Together with other answers that showed traits of highly

traditional approaches to writing composition instruction, the authors (2008: 59) concluded that

EAP teachers in New Zealand showed ‘a preference for more extensive treatment of

grammatical issues than is usually suggested by proponents of a strictly incidental Focus on

Form approach’.

On the one hand, teachers’ continued correction of errors may potentially have resulted in a

perceived improvement in the process and methodologies of giving feedback since error

correction is another example of changing practices in EAP. In addition to moving from a

‘product’ orientated summative approach of the past to a ‘process orientated’ formative focus

(Ferris, 2006; Ferris & Kurzer, 2019) which is currently favoured, many new approaches such

as peer assisted marking, guided self-correction, changes in EAP writing pedagogy and

computer assisted feedback have transformed the way in which formative writing feedback can

be given (Hyland & Hyland, 2006). On the other hand, if new technologies and approaches

have facilitated mass correction that is demotivating for students, overcorrection again requires

research attention.

The present study followed Etheringon & Burgess’ (2002) investigation into the actual

grammar teaching practices in EAP classrooms with a focus on the teaching/correction of

English articles. Anecdotal evidence in the researcher’s own institution suggested that English

article error correction was frequent and widespread, while articles were taught more often in

‘hygienic’ prototypical examples somewhat divorced from the reality of academic writing.

Such observations suggested the need for more rigorous empirical research in order to

investigate Research Question 1:

To what extent do EAP teachers in Higher Education programmes in the UK explicitly teach

or correct English article use?

4.1.1 Definitions of the notion of grammar teaching

Grammar teaching today refers to the broad range of ways that teachers focus on grammatical

form, which means not only the traditional focus upon decontextualised isolated grammatical

examples of one structure with a presentation/practice paradigm, but any activity that focusses

students on a discrete part of a language’s grammar. In the modern sense, grammar teaching

includes; asking students to notice patterns; asking students to self-correct or peer correct each

other’s work for grammatical accuracy, and studying concordance lines from an academic

corpus and identifying grammatical patterns. While the issue of corrective feedback is in itself

an area of academic and pedagogical interest, written/oral/direct/indirect corrective feedback

is a necessary feature of teaching. Thus, grammar teaching involves any focussing of learners

upon ‘correct’ grammatical form, or as Ellis explains (2006: 84):

Grammar teaching involves any instructional technique that draws learners' attention to some

specific grammatical form in such a way that it helps them either to understand it

metalinguistically and/or process it in comprehension and/or production so that they can

internalize it.

Since this is a very broad definition it will also be useful to distinguish between explicit or

implicit grammar teaching. Implicit knowledge can only be held or expressed unconsciously

and is acquired for communicative purposes by interactions with the environment. As Ellis

(ibid) notes, explicit grammar teaching does not necessarily imply that metalinguistic reference

is made to the language element being focused upon and it should also not be confused with

inductive or deductive teaching styles which are both types of explicit grammar teaching. Thus,

for example, asking students to reflect on why/how the definite article is used as a cohesive

device in a text (or looking for other patterns of use) would still be ‘explicit’ grammar teaching,

even if the teacher had not prefaced the activity with a focus on forms, the objectives of the

exercise or specialised metalinguistic terms such as ‘cataphoric or anaphoric reference’. The

distinction in explicit grammar teaching is that the teacher moves the students’ whole focus to

form and its related meaning or function, rather than the text’s meaning alone. Therefore, the

notion of grammar teaching in this thesis will include any explicitly conscious raising activity

aiming to increase language accuracy.

4.2 Methodology

A survey questionnaire (followed up by interview) was distributed to a small number of

teaching staff in the university’s English for International Students Unit (EISU), at that time

responsible for Presessional and Insessional EAP programmes at the university. The same

survey was then distributed (as detailed in 4.2.4) among two larger populations of English

teachers recruited via a first email distribution lists of BALEAP teachers and a second email

distribution list of teachers involved in an MA TEFL Distance Learning Programme. This

sampling approach allowed an investigation of EAP tutors’ beliefs and practices at both the

local and national level.

4.2.1 Survey design criteria

In designing the survey, the aim was to create an instrument with a mixture of closed and open

questions. With a view to extending the survey to further populations of teachers, a second

design criterion was to include as few questions as possible in order to maximise participation

and reduce drop-out rates. No biographical information was collected in order to fully

anonymise the survey and ensure that responses would be as honest as possible.

4.2.2 Questionnaire instrument

As can be seen in the Questionnaire shown in Appendix 1, the first 3 questions were closed

response multiple-choice questions. Question 1 (see Figure 7) prompted respondents with

various article errors and asked tutors whether the examples were used accurately, were errors

or somewhere between (correct even if unnatural). Question 2 aimed to investigate how often

participants taught article use with options ranging from ‘never’ to ‘most weeks’. Question 3

returned to the errors introduced in the first prompt and asked participants about the likelihood

with which they would provide written corrective feedback in different scenarios. Finally

question 4 allowed open responses to a question asking tutors to reflect on how learners could

be made to improve English article accuracy.

The four questions were first piloted with colleagues. As a result of their ideas, the definitions

of ‘sometimes’ and ‘occasionally’ in question two were expanded to illustrate and define the

terms which initial respondents found confusing. The examples given in prompt 1 shown below

in Figure 7 were provided following feedback that questions about errors needed

contextualising with some actual examples of student article errors. These examples were

authentic errors from a previous student’s particularly error filled essay (consent given for use

when anonymised) were added as a prompt, as shown in Figure 7:

Figure 7: Survey Question 1

4.2.3 Interview structure

In addition to the online surveys, six EAP colleagues at the University of Birmingham

volunteered to participate in semi-structured interviews to allow for some more qualitative

insights into their survey responses. At interview, a paper copy of the survey questionnaire

was shown to participants. In order to avoid overly leading questions, all tutors were asked the

same two identically worded questions at the start of the interviews:

1. Question 2 asked about how often you teach the article system to your students. Can

you talk a little more about when you teach them and how you teach them? When are

you most likely to teach article use?

2. Question 3 asked about how likely you’d be to highlight the errors/give feedback in the

example. In general, do you give corrective feedback? How often and when do you give

corrective feedback? How do you do it?

All the teachers were asked to bring to interview some recent examples of written work they

had corrected. All tutors agreed to be recorded (the recordings are transcribed in Appendix 3).

4.2.4 Participant information

English for International Students Unit (EISU)

EISU has since been merged with the university’s Foundation programme and rebranded the

Birmingham International Academy. At the time, EISU mainly delivered a range of

Presessional and Insessional programmes. As a centre accredited by the British Council (BC),

teachers are required to hold TEFL Q (Masters level of education and/or a recognised English

teaching qualification) level qualifications. At peak summer periods the unit employs over 100

EAP tutors to teach the large summer Presessional cohorts. However, at the time of this initial

survey (in the Autumn term) there were 20 full-time EAP tutors employed at the unit (both

open-ended and fixed-term contract staff). All staff were teaching on one (or two) of the main

programmes in addition to seeing international students who request further support with their

academic English in one-to-one tutorials.

The British Association of Lecturers of English for Academic Purposes

BALEAP has 120 institutional members (from over 73 British Universities) and 130 individual

members. The association has an active JISCMAIL forum discussion list on which members

discuss issues in EAP, upcoming conferences/meetings/groups/ training events, and various

other topics. This email distribution list was used to invited participants to respond the four-

question online survey.

The ELAMA Mail Discussion List for Distance MA students

The University of Birmingham’s Department of English Language and Applied Linguistics

(ELAL) runs two distance course for teachers/applied linguists working abroad /off campus

wishing to follow a UK Masters programme: MA Teaching English to Speakers of Other

Languages (TESOL) and MA Applied Linguistics. The programmes attract a diverse range of

students, most of whom are practising English teachers. Each year there are between 200 and

300 students enrolled on the programme (and enlisted on the mail discussion list). Most are

L1 English, but some are L2 English speakers. Their experiences range from being initial

entrant teachers (in language school, college or university settings) to teachers/institution

managers and directors with many years of experience. Most are in broad TEFL contexts (even

if working at university level). The programme’s students are all automatically registered onto

the ELAMA mail Discussion forum which is used to discuss a range of course and TESOL

related topics. In addition to the 200-300 current (and active) students, there were reported to

be approximately 400 further past students, course lecturers and staff enrolled on the list at the

time of the survey.

4.2.5 Recruitment/Response rates

As shown in Table 7, the first small survey was a convenience sample of the English for

International Students’ Unit, the researcher’s own institution. 13 online responses were

Table 7: Response rates

Population

Responses

Response

Rate

English for International Students Unit (n=20)

65%

BALEAP JISCMAIL (n =250)

129

52%

University of Birmingham Distance MA ELAMA List (n=600)

completed (65% response rate). In the email sent to colleagues with the link to the anonymous

online survey, teachers were asked to respond by email if they would be happy to participate

further by following up the 5 minute questionnaire with a ten minute one-to-one recorded

interview at which they were also asked to bring recent written corrective feedback (marking)

of student work they were currently doing or had recently done (with the incentive of being

entered into a small prize draw). Six of these colleagues further agreed to an interview/audit of

their teaching materials/corrective feedback.

Survey of the wider EAP field

The online survey was copied and redistributed to the BALEAP JISCMAIL Discussion List

(129 responses) and the University of Birmingham Distance MA ELAMA List (54 responses)

to achieve a more representative sample of the field. There are no known response biases that

explain the higher BALEAP mailing list response rate (52%) response rate to the ELAMA

Distance MA List response rate ( 9%). However, it is known that the high number of calls for

participation in the ELAMA list is responsible for ‘response fatigue’ among tutors.

4.3 Results and Discussion

The quantitative results of the survey responses are now analysed (Sections 4.3.1 - 4.3.3) before

the qualitative results are discussed in Sections 0 - 4.3.11.

4.3.1 Question 1

As can be seen in Figures 8 – 10 below, there was general agreement in Question 1 (multiple-

choice response), from 91 percent to 100 percent, that the student examples provided in the

question prompts could be classified as errors. All EISU tutors in the survey identified the

exemplar as containing errors in overspecification and omission of definite article. Only a small

number of teachers disagreed with the categorical classification as ‘errors’ in the BALEAP

(Figure 9) and ELAL samples (Figure 10). The ‘middle’ choice allowed the respondent express

their view that the student’s use was ‘correct’, even if not ‘natural or native-like’ (6.3 percent

in BALEAP list, 5.7 percent in MA List and 0 in EISU). From reviewing the BALEAP open

Figure 8: EISU EAP Tutor Responses to Question 1 (n=12)

choice questions (Question 4), it can be seen that two teachers believed the sentences might be

acceptable in another World English (with which they did not want to interfere).

Figure 9: BALEAP mailing list responses to Question 1 (n=127)

Figure 10: ELAL MA Distance programme responses to Question 1 (n=53)

4.3.2 Question 2

Figures 11 - 13 show the responses to Question 2 which enquired about the frequency with

which they taught articles in the classroom (with a list of possible examples to stress that this

was not limited to traditional methods of teaching). The frequency descriptors were designed

to help teachers to reflect on their practices with a given class. For an EAP Presessional tutor

this class contact is typically a fairly intensive period of 6, 10 or 15 weeks, but in some instances

lasts for two or more terms. For example, at the time of writing, most of the EISU Presessional

tutors had just finished teaching EAP for ten weeks (18 hours with a full timetable). However,

the other survey responses reflect a broader diversity of other course periods and timetables.

The most striking point observed in the data collected is the surprisingly low number of

teachers who reported never teaching article use. An assumption had been made that many

teachers, especially from the BALEAP list, might be teaching highly advanced academic skills

classes that do not focus on grammar. Yet no EISU teachers (and only 5.5 percent in the

BALEAP list and 9.4 percent in the MA list) chose ‘never’. The small difference between the

BALEAP and MA list tentatively suggests the possibility that EAP teachers are more (not less)

Figure 11: EISU EAP Tutors’ responses to Question 2 (n=12)

Figure 12: BALEAP list EAP Tutors’ responses to Question 2 (n=127)

100

Figure 13: ELAL MA Distance programme responses to Question 2 (n=53)

likely to teach articles than TEFL orientated teachers (the MA list had more TEFL teachers

subscribed than the other two groups). On the other hand, while the majority of tutors are

teaching English articles one, two or three times a year/term/programme, very few tutors were

teaching the article system more than three times.

4.3.3 Question 3

Question 3 (Figure 14 below) aimed to assess how likely the teachers were to provide corrective

feedback to the errors presented in the first prompt. It had been expected that most teachers

would follow the general recommendations across the EAP literature and not correct all such

errors in certain contexts (such as classwork or in homework tasks). However, reflecting the

anecdotal experience that some tutors rightly or wrongly see themselves as proofreaders for

101

students’ higher stake assignments, it had been predicted that tutors would be more likely to

correct such errors in certain contexts.

Going back to this piece of writing (see below) how likely would you be to highlight any written

errors you saw in your feedback to students (written or verbal)? This could mean

annotating/marking errors, underlining errors or simply pointing them out to students in a

tutorial. Please consider the situations below:

Figure 14: Question 3 prompt

Indeed, as had been predicted, the vast majority of teachers in all groups would correct such

errors in a ‘first draft’ or ‘conference paper’. However, it was surprising to see not only that a

large majority of teachers in all groups would correct the errors in written homework, but also

that over half of teachers in all groups were likely or very likely to correct the errors in a quickly

written piece of work in class time (58.3 percent of EISU tutors, 60.7 percent of BALEAP

tutors and 51.9 percent of the ELAL MA list).

Figures 15-17 illustrate the responses to the question of whether tutors would correct article

errors in writing produced quickly in class.

102

Figure 15: EISU EAP Tutors Question 3 (in class) n=12

Figure 16: BALEAP EAP Tutors Question 3 (in class) n=120

Figure 17: ELAL MA Distance programme Question 3 (in class correction) n=52

Figures 18-20 below illustrate the teachers’ responses as to whether they would correct the

article errors in a writing task set for homework. Again, for weekly homework it had been

expected that many tutors would not be likely to correct article use or would be unsure without

further knowledge of the students’ general writing ability. However, across all groups most

tutors were ‘likely’ or ‘very likely’ to correct such errors. The group with the smallest number

103

of teachers answering ‘likely’ or ‘very likely’ was the MA list (79.3 percent), with 87.1 percent

and 91.3 percent for BALEAP and EISU teacher groups respectively.

If this was in a weekly writing homework Going back to this piece of writing (see below) how

likely would you be to highlight any written errors you saw in your feedback to students

(written or verbal)? This could mean annotating/marking errors, underlining errors or simply

pointing them out to students in a tutorial.

Figure 18: EISU Tutors Question 4 (weekly homework) n=12

Figure 19: BALEAP EAP Tutors Question 3 (weekly homework) n=127

104

Figure 20: ELAL MA Distance Programme Tutors Q.3 (weekly homework) n=53

There was less surprise about the first draft of an assessed piece of work (as illustrated in

Figures 21-23 below) since at EISU for example, the teachers were actually expected to mark

the first page of the process writing extended writing assignment. Finally, as shown in Figures

24-26, it can be seen that most teachers were ‘likely’ or ‘very likely’ to correct such errors in

when they were co-authoring a conference paper with the student.

Going back to this piece of writing (see below) how likely would you be to highlight any written errors

you saw in your feedback to students (written or verbal) in the first draft in process writing? This could

mean annotating/marking errors, underlining errors or simply pointing them out to students in a

tutorial.

Figure 21: EISU Tutors Question 3 (in first draft) n=12

105

Figure 22: BALEAP EAP Tutors Question 3 (in first draft) n=127

Figure 23: MA list tutors Question 3 (in first draft) n=52

In a conference paper you plan to submit (and co-present) with the student. Going back to this piece

of writing (see below) how likely would you be to highlight any written errors you saw in your feedback

to students (written or verbal)? This could mean annotating/marking errors, underlining errors or

simply pointing them out to students in a tutorial.

Figure 24: BIA EAP Tutors Question 3 (conference paper) n=12

106

Figure 25: BALEAP EAP Tutors Question 3 (conference paper) n=127

Figure 26: MA List Tutors Question 3 (conference paper) n=52

4.3.4 Qualitative findings

This section now presents the qualitative insights gained in this study which includes the

respondents’ answers to the survey open question (Question 4) and the interviews with six EAP

tutors. The full answers from the tutors who responded to the open response question are

shown in Appendix 2 (n=101). The analysis below starts instead with the insights gained from

the answers given in six tutor interviews carried out following the survey. In the interviews the

teachers were asked to give more information about their answers. Their full answers were

transcribed and are presented in Appendix 3. By analysing their answers and comparing them

to the wider survey responses to the open question (Question 4) asking tutors to reflect on how

learners could be made to improve English article accuracy, this analysis provides a deeper

indication of when, how and why tutors correct or teach article use.

107

Interviews/audit of feedback.

On request, all teachers brought a class set of marked papers (or gave access to a class set the

week after the interview) with the aim of establishing whether the tutors regularly corrected

errors and contextualising the interviews with specific examples. A review of these papers

showed that only two teachers had not made annotations of article errors in their marking. One

teacher of the four who corrected articles had made only two article corrections, but all the

others had corrected at least one example of what was perceived as an article error in each

essay. In brief, the picture was again one of correction of articles for four out of the six teachers

even when the task was not assessed.

Sections 4.3.5 to 4.3.10 below summarise and interpret the answers of the six interviewed

teachers in turn (their full responses are presented in Appendix 3).

4.3.5 Tutor #1

Tutor #1 was an English L1 tutor with three years’ experience of Presessional teaching.

Analysing her answers at interview, it can be seen that this tutor saw her role as correcting

article misuse rather than teaching correct use in the classroom. Her initial responses suggested

that she did not see the teaching of articles to be consistent with EAP best practices or possible

in the time constraints of the Presessional. However, arguably in conflict with this asserted

rejection, it was clear from the examples she brought to interview that she was regularly micro-

marking article use. From her answers (see transcribed extract in shown in Figure 27 below),

it is also clear that she believed such article corrections helped learners improve their accuracy:

108

Question 2 asked about how often you teach the article system to your students.

Can you talk a little more about when you teach them and how you teach them?

When are you most likely to teach article use?

Teaching the article system? To be frank, I don’t have enough time to teach it in the classroom

in one session of its own. With the time pressure and everything we cover, I don’t have time

to dedicate time to this. […] It might come up in cohesion and I might send them to a grammar

reference for homework. But I don’t formally teach it in EAP. I think it’s quite a trivial surface

error. But each time when I correct writing…when I check students’ writing, I will check and

point out any errors in articles most of the time.

Question 3 asked about how likely you’d be to highlight the errors/give feedback

in the example. In general, do you give corrective feedback? How often and when

do you give corrective feedback? How do you do it?

I would like to say I do peer correction and things like that…but I just don’t have time. Each

time… I correct the errors – I often use codes for all grammar errors. Students will get articles

marked two or three times…sometimes more if there’s a high stakes assignment. Maybe that’s

not ideal, but we are so short of time. So, each time if I spot any error I will just mark it. And

then in the tutorial time I will talk through it. And…if..actually if there are more than one

students making similar mistakes…in sort…maybe in of in similar usages, I will collect

examples and talk through very briefly in the classroom as a group.

How often do you correct their errors?

Most weeks I’ll micro-mark a few paragraphs of their writing. For the important

assessed work I’ll probably mark two pages of the first draft of the assignment.

Does that help?

Definitely. Without correction they wouldn’t normally realise that they‘ve used the article in

the wrong way and after I’ve picked it up they normally ask questions and try to reason…find

the correct way and they ask questions. OK, no I wouldn’t be able to evaluate how effective

this practice this. But I think in my opinion they do get better over time.

Figure 27: Transcribed extract Tutor #1

While the view that absolutely all article errors should be indiscriminately proofed and

corrected is likely to be a minority view across EAP teaching community, this tutor’s

109

assumption that correction led to improved accuracy was representative of a small number of

responses in survey Question 4 (open response), four of which are shown below for illustration:

• Correcting surface errors like articles are part of our role as EAP tutors. I feel this is

best done using error codes and the student’s own work rather than abstract sentence

level grammar exercises.

• Firstly, students need to be aware that they have a problem with their own use of

articles…

• ...constantly flagging it up in corrective feedback may help over time.

• Yes, we can. Written and oral feedback on written work.

Such tutors show an underlying belief that accuracy improved over time through regular

correction. It appears that such tutors see themselves as ‘proofreaders’ for their students.

While most teachers will perceive a need to micro-mark parts of the early drafts of high stakes

assignments in a justifiable ‘product-oriented’ approach, some teachers appear to believe they

should correct all article errors in all written assignments. However, their underlying

assumption that teacher led indiscriminate error correction of written work automatically leads

to learning was not fully supported by the literature reviewed in Section 2.5.3.

4.3.6 Tutor #2

Tutor #2 was an English L1 tutor with an EFL background, very new to EAP teaching

(interviewed after her first year of Presessional teaching experience). Her answers are

representative of a type of teacher who regularly focusses students on the grammar of articles,

both within and outside of recommended practices in the EAP literature. In fact, (as shown in

Figure 28) she reported teaching the article system 5-6 times during a 15-week presessional.

This started with one very specific 45 minute article focussed session followed by more

impromptu 5 minute incidental sessions and whole class reviews of common errors made in

articles (in feedback on writing). On the one hand, many of this tutor’s approaches (focussing

110

on articles in presentation skills classes, peer corrections and a focus on proofreading skills)

were well within the frequently recommended pedagogy in the EAP literature. She also clearly

thought article errors represented a lexical issue in addition to a conceptual ‘rules’ problem for

students. However, it was clear from her answers that she would have spent longer on the rules

of article use with greater time and she was still adopting somewhat of an EFL paradigm in her

view that teachers should return to grammar points like articles in a ‘cyclical way’.

Question 2 asked about how often you teach the article system to your students.

Can you talk a little more about when you teach them and how you teach them?

When are you most likely to teach article use?

[…] So, thinking about what I normally do….In the 15-week Presessional I did this - I had

one session which was...like a 45 minute session where we look specifically at articles and then

there were two or three times when it came up and we had 5-ten minutes each time. And then

there’s the ARP feedback, so maybe I help them with articles 5 or 6 times on a longer

presessional. But I didn’t do this so much with the 10-week students.

[…]OK, last question. In question 4 you were asked for

recommendations/suggestions about helping students improve their accuracy of

English Articles in writing….(do you have any further comments?). Do you have

any more tips? Can we help students improve their English article accuracy and

if so, how?

OK, so from what I have…from what I’ve understood. OK look I’ve got my Diploma but I’ve

not done …I’m not someone who’s done a lot of reading on this. But from what I’ve

understood, articles are late acquired. So, to a certain extent, it’s one of those things that is

cyclical. It comes up at various points in what you’re teaching. It’s just a complicated thing.

So, what I’ve done in the past with …because I have taught Russians and East Europeans…so

for speaking, making it part of the rhythm. I teach it more in the speaking skills. You can make

them understand that there’s a space for the article, even by clapping it out. For example ‘as a

result’…you clap on the syllable….include them in a group. The focus shouldn’t be on ‘why’,

it’s just look it’s there in that chunk of language. As with everything, for some students it really

helps them, for some students it’s like water off a duck’s back.

Figure 28: Transcribed extract Tutor #2

Reviewing survey Question 4 open responses (n=101), it appears that a number of EAP

teachers assume that continued references to rules or a novel method of helping students

111

understand the general purpose of articles is the key to improving accuracy. The following

responses typify what could be characterised as an ‘optimistic rule giver’ type of tutor:

• I try to teach them some basic 'rules'. 'The' is used for something specific, one of a kind,

unique. 'A, an' are used for one of many.

• Simplify article use by showing that the definite article is an audience oriented particle

which signals to the reader - you know what I'm talking about.

• Some basic rules e.g. with plural country names (the UK, USA etc.) are helpful. Ideas

which are harder to grasp quickly, e.g. talking about generic versus specific items, can

be illustrated through reading and recognising article use.

• There's no magic wand - they just have to learn the rules and develop an ear for how

articles are used.

• Learn about articles ourselves.

Having established that many tutors continue to believe in rules-based grammar teaching, as

also suggested by the quantitative findings, it was important to examine whether such pedagogy

was effective in improving accuracy (as discussed in the next chapter).

4.3.7 Tutor #3

Tutor #3 (English L1) had eight years of ESOL experience and five years of EAP Presessional

experience. Of the six interviewed teachers, this tutor’s reported beliefs and classroom

practices (see Figure 29 overleaf) appeared most aligned to the pedagogy most usually

advocated in the EAP literature. He had not marked article errors in the student feedback

examples he brought to interview. However, he did focus students’ attention on article use.

Rather than decontextualised prototypes, this teacher used a modified authentic academic text

to highlight how article was used in definitions and later how articles helped the writer change

the focus from general to specific references. The teacher used a springboard session to focus

students on correct use within an academic article, but then stopped teaching and emphasised

the use of peer-marking and self-correction to promote learner autonomy. When asked about

112

error-correction, the only articles that this teacher reported correcting were those in high-stakes

assignments. As shown in the transcribed extract below, this tutor did not correct articles in

most homework submissions as he saw this as ‘spoon-feeding’. This teacher again seems to

be practising contemporary EAP best practices advocated in most literature.

Question 2 asked about how often you teach the article system to your students. Can

you talk a little more about when you teach them and how you teach them? When

are you most likely to teach article use?

Yes, I do introduce it more formally at the start. I often look at correct use first, maybe at how

it works with definitions of generalisations in the introduction, which is normally a plural form

without article. This is important because I can’t assume they’ve been taught it [the article

system] in the way I want them to understand it. It fitted in quite nicely in the 10-week course

this year at the beginning – I brought forward the stuff about proofreading. But rather than

focus on errors, I first looked at how correct use worked. Then I put a Powerpoint slide up with

a research article I’d chosen but simplified a bit. I got them to focus on how articles determine

specific examples at some points and generalisations in others – I think I got this idea from

Swales. Then I gave them some proofreading exercises with correct and incorrect article use.

Then I got them to start peer marking their own work. I actually do that with every written task

– they peer mark it every time…. before it comes to me. And that first time it was with articles.

Question 3 asked about how likely you’d be to highlight the errors/give feedback

in the example. In general, do you give corrective feedback? How often and when

do you give corrective feedback? How do you do it?

[…] I think there’s a lot of teachers overmarking articles. The problem is, the student is just

going to depend on your spoon-feeding. And do they actually learn? You’re not going to be

there in their main programme – they have to learn to do this themselves. And we have these

really amazing motivated students in EAP which is so great – I come from an ESOL background

where I spent so long trying to get students to take responsibility, but our Presessional students

get this more quickly.

Figure 29: Transcribed extract Tutor #3

While the majority of respondents in the survey appeared to error-correct students work more

indiscriminately, many answers to survey Question 4 (open response) appeared to follow these

more data-driven approaches to teaching article use:

113

• I haven't found generic exercises such as 'correct the errors' in random sentences help

really. However, I have found that activities which help students notice their own

specific errors in their own writing does seem to make some difference.

• I think it is important to sensitise students to the issue of article use and encourage them

to self-correct in revision drafts.

• I generally ask students to proofread and peer check. I think that students can generally

avoid article errors when they concentrate on articles specifically. However, when

grammar is not the focus I think there is a tendency to translate from the student's L1

which means they make errors with articles.

• At lower levels, raising students' awareness of their use of articles can make a

difference, with peer correction being useful in this regard.

• My response to this is to get students to closely analyse reading texts which have either

been introduced as part of the syllabus or which they independently chosen and to look

for specific language use issues - in this case the use of the article (pre-teaching article

use if appropriate) and to compare in pairs

• Awareness-raising will be useful for some students, allowing them to improve their

understanding/use over time. Short focused activities (e.g. correction exercises, use of

corpora lines to note patterns) may also help (some) students appreciate

when/where/why the use/omission of an article may to blur meaning.

Stressing the importance of autonomous study skills (self-correction, peer marking) and

focussing students on article use through data-driven learning approaches the tutors above

appear to be following the best practices of the recommendations found in the EAP literature.

These best practices show that focussing students on article accuracy certainly does not need

to be decontextualised or teacher centred. For the sake of simplicity, within this fivefold

typology, such teachers are thus labelled ‘EAP paradigm tutors’.

4.3.8 Tutor #4

Tutor #4 was an (L1 English) EAP teacher with over 15 years’ of experience. At interview she

stated an assumption of a strong L1 transfer effect on Chinese students’ presumed inaccuracy

with articles. While mentioning several practices (see Figure 30) that could fit within the

114

pedagogy and learner autonomy most advocated in the EAP literature (the ‘EAP paradigm’),

she also showed a strong belief in error-correction and rules from grammar reference guides.

More similar to tutors #1 or #2 than #3, this tutor might therefore best be characterised as a

combined ‘proofreader’ and ‘optimistic rule giver’.

Question 2 asked about how often you teach the article system to your students.

Can you talk a little more about when you teach them and how you teach them?

When are you most likely to teach article use?

Yes, we need to teach articles in EAP if the students are IELTS 6.5 or below. For higher level

students, I think they’re sorting themselves out alone, but most Presessional students from

China need help. The Chinese students really struggle with articles because their first language

doesn’t have them. There isn’t much time, but I’ll try at least twice to pull some of the students’

errors that I’ve noticed and put them on the screen. They need to see how important articles

can be especially in definitions or about how generic articles work in an engineering topic and

the difference between when they generalise about a term in one paragraph but may need to

refer to specific and known things in another paragraph. And apart from that I send them to

grammar reference books. I tell them they should all have Hewing’s grammar book so if they

get a correction they don’t understand they need to look it up and learn the usage. I correct

their errors too so they can see what they need to learn.

Figure 30: Transcribed extract Tutor #4

Tutor #4’s confidence in the effect of grammar rules and mass error-correction was not widely

shared among the surveyed responses. However, a small number of the open responses given

to survey Question 4 (open response) were equally confident in such practices:

• Chinese speakers will need to be continually supported with frequent exercises and

focused correction of their writing.

• [Recommendation of book]. This book outlines 125 rules for articles.

• Yes, teach, practice with paragraphs of error-filled text, etc.

• Access to a grammar reference source such as McCarthy and Carter's Cambridge

Grammar.

• Yes, but it's very difficult. Awareness raising, examination of rules, practice exercises.

115

The responses above suggest that some EAP tutors wish to keep away from the issue of article

use in whole class teaching, but see some responsibility for making sure students are noticing

their errors through correction and then assuming the students’ independent use of grammar

references materials can help them improve their accuracy levels.

4.3.9 Tutor #5

Tutor #5 (L1 English) had an ESOL background and 3 years of EAP Presessional teaching

experience. This teacher showed a strong commitment to data-driven learning, learner

autonomy and peer/self-correction, but framed this focus through a rule-deduction approach.

He had only marked two article errors in all of the feedback he brought to the interview. As

shown in Figure 31 below, like Tutor #3 many best practices were reported that seemed

harmonious with EAP literature recommended practices (the ‘EAP paradigm’).

1. Question 2 asked about how often you teach the article system to your

students. Can you talk a little more about when you teach them and how you teach

them? When are you most likely to teach article use?

From the get go I would try to introduce them to the idea of proofreading and learner autonomy.

So in the first week I’d expose them to the idea of how to correct some of these problems. I

created proofreading prompts… (checking countability, plurality). For articles I get them to

deduce three basic rules and that single countable noun phrases needs a determiner. I take

them through these intensely in this one session with this proofing prompt (he demonstrated).

So I’d expose the issues to them like this and make them aware of it, but I wouldn’t keep

pressing it. I wouldn’t keep micro-marking it. If they wanted to, in tutorials I’d direct them to

further resources, but it’s something they have to do autonomously. OK, if it was the ARP

(extended writing assignment) I feel there’s too much at stake so normally I micro-mark the

third page, or I would micro-mark a page or two, but they’d have to go over the rest of it.

Figure 31: Transcribed extract Tutor #5

On the issue of error-correction, this tutor argued the case for occasional highlighting of article

116

errors, but without telling students the correction itself:

Question 3 asked about how likely you’d be to highlight the errors/give feedback

in the example. In general, do you give corrective feedback? How often and when

do you give corrective feedback? How do you do it?

I didn’t correct much on the 6-week course this year. But on the 10-week course last year, I

marked them …I would just put ‘G’ for grammar and I would leave it up for the student to work

out what the mistake is. I would leave it up to them to come to me if they didn’t’ understand

what this issue was. I really want to reinforce learner autonomy. So, another thing I do is peer

marking. I think er…I’ve especially learned this with learner autonomy, it’s more about

promoting their confidence. So with peer marking, it doesn’t matter if they’re not 100%. It’s

ok, but they will learn from their mistakes from making mistakes. And obviously I will do some

quality control.

Figure 32: Tutor #5 continued

In many ways, Tutor # 5 appeared to best characterise the attitudes, beliefs and practices most

frequently mentioned in the BALEAP survey responses. These teachers teach the form of

language, but always usually with an authentic text driven approach with emphasis on students’

own noticing of patterns. Figure 33 below presents an analysis of the BALEAP answers

(n=101) clustered into group types. As can be seen, by far the most frequently recommended

teaching strategy was the general notion of asking students to ‘notice’ article use (32 mentions).

These 32 mentions did not explicitly make reference to the third most popular recommendation

of asking students to deduce the function-form meanings (18 recommendations) from studying

texts/examples, but it can probably be assumed that many of the 32 respondents recommending

‘noticing’ strategies’ implicitly suggested some sort of data-driven learning approach. As can

be seen below, the more frequent approaches (apart from correction of errors) are not

incompatible with the frequently recommended pedagogy in the EAP literature.

117

Figure 33: BALEAP list pedagogical recommendations (number of mentions, n=101)

4.3.10 Tutor #6

Tutor #6 was an L2 English speaker of (L1 Chinese) at the end of her first year of teaching

EAP in the UK, following five years of teaching college English in China. Tutor #6 had

corrected two article errors in the high-stakes assignment (first draft), but no article errors in

the other homework examples that she brought to interview. As can be seen in her full interview

transcript (see Appendix 3), this tutor reported that she rarely corrected English articles in

regular homework feedback because 1) she preferred to focus on ‘the way they structure their

ideas’ and 2) she did not ‘feel confident correcting them’. She therefore rather reluctantly

corrected article usage only in the most important high stakes assignments, mostly because she

felt it was what the students wanted rather than seeing any acquisitional benefit from doing so.

As a teacher with experience of teaching academic English in China, this tutor explained (see

Figure 34) that Chinese students often wanted more grammar in lessons because of the way

118

English is taught in the compulsory ‘college English’ courses in China. She further speculates

that Chinese students can feel frustration at their lower accuracy in real production compared

with their better performance in multiple choice tests:

Could you think of any cultural reasons for misunderstanding article use? As

you’re Chinese, perhaps you can help me understand whether it’s the students’

background or their misunderstanding of academic genres that is the problem.

I was teaching English actually at university in China for five years. You know in

China it’s compulsory for the students to study English for the first two years. We call

it ‘College English’ and we have a dedicated coursebook that they will study no matter

what their subjects are. On reflecting on these practices, we did actually do a lot of

grammar practice. In fact, more often or not we’re using multiple choice questions with

texts, so you only need to pick out the correct answer. So you’re not really producing

your own writing. It’s very different…picking out the correct one…to writing your

own one. This college English is more about grammar. A major part of it is to

consolidate grammar points and a major part of it is to show them exam papers and 60-

70% of the exam is multiple choice. I don’t know why students make more errors in

their writing than they do in the tests, but I think it’s like….like being on a simulator to

learn to drive a car. However long you go on a simulator, it cannot prepare you for

driving your own car on the road. And English for Chinese students is like this. I

actually use this analogy when I teach essay writing. Knowing the rules is quite

different to using the rules accurately. Chinese students spend so much time on those

multiple-choice questions, but their knowledge is not useful in real language situations.

I notice some students overuse definite articles. It is sometimes that Chinese students

want to reinforce something. But it is really confusing.

Do you think it can have any negative effects when we correct students? Are there

any dangers of overcorrecting students? I worry what would happen if I tried to

learn Chinese writing and had lots of corrections on parallel forms like measure

words or something.

I wouldn’t have thought so, to be honest. They are all very keen. They can see that

attending this course can help them to improve. Actually, one of the feedback I had

from students at the end of the 6-week Presessional was ‘why don’t you teach more

grammar’ and ‘why don’t you teach pronunciation’. I think they would like more

grammar.

Figure 34: Transcribed extract Tutor #6

119

Tutor #6’s attitude resonates with two other responses in the survey. In response to Question

4 (the question which allowed tutors an open response), two respondents commented that they

found the rules too challenging to explain:

• I find it difficult to explain and believe that extensive reading on part of the student is

the best solution

• I'm a second lang [sic] learner myself at PhD level and make mistakes with articles

and prepositions. I'd teach the rules which are sort of vague anyway but in general I'd

let it go.

While two BALEAP tutors indicated a reluctance to teach or correct articles due to their own

lack of confidence in understanding or explaining article use, 99 other respondents made no

reference their own lack of abilities or understanding. This result was surprising as the English

article system is extremely complex to fully comprehend. It is possible that some tutors felt

embarrassed to discuss their own abilities even in an anonymous survey so, like any self-

reporting survey, this result must be treated with some degree of caution. For whatever reasons

they would prefer to avoid correcting article use, this typology characterises these tutors as

‘reluctant error correctors’.

4.3.11 Principled ignorers of English articles

None of the interviewed EISU teachers showed a principled outright objection to focussing on

article form (either through noticing techniques or correction). However, 5.5 percent (n= 127)

of the BALEAP questionnaire respondents reported that they never taught English articles and

39.4 percent only taught them once or so during the programme. As shown below in their

answers to the survey open question, there is therefore evidence that a small minority of

teachers have principled objections to focussing on articles in any way. These teachers can be

characterised as ‘principled ignorers’ who may see the EAP best practices as incompatible with

120

the focus on minor surface forms. Five example answers to survey Question 4 (open response)

are shown below:

• I'm unconvinced that beyond pre-academic study it's really worth explicitly teaching

this point at all, for the little return which may come from doing so, not least in terms

of its impact on meaning. By the time students reach undergraduate or postgraduate

level, I doubt if it's more than a minor irritation to their subject lecturers (with perhaps

a few exceptions in language-related subjects). With more and more English-medium

research published (and edited) by non-native speakers of English, occurrences of

article 'errors' may well become increasingly frequent.

• We used to talk about 'error gravity' a long time ago, and I believe that inaccuracy in

this particular area is seen as 'error lite'.

• For a more confident tutor, it's much less important. In a less parochial sense, English

is becoming much more global (cf English Next et al), with washback into "Centre"

countries from other users.] Is it practical?

• I would say that I am less likely to address it explicitly as an EAP teacher because it is

less important than other features of EAP.

• Not sure. I think articles are one of 'final errors', if you like, NNS that have been in an

English country for many years can still make errors with articles. It may be improved

with greater exposure to language and language environment than direct teaching in

class.

Further analysis of these 101 BALEAP teachers’ responses (specifically the principled

ignorers) involved grouping their responses into 7 categories, as shown below in Figure 35.

These categories paraphrase the gist of the respondents, rather than show verbatim responses.

As can be seen, eight of the 101 BALEAP respondents made reference to the fact that articles

could be left to acquisition processes with enough naturalistic contact with academic English.

Other reasons for not teaching the article system included; the ineffectiveness of teaching the

system; the low return for the effort; the higher perceived level in EAP, and World English

arguments. Nevertheless, these views were rare among the surveyed respondents.

121

Figure 35: Reasons given for not teaching articles explicitly (n=101)

4.3.12 Fivefold typology of EAP tutor approaches to English articles

The analysis above has shown five ‘types’ of EAP tutor: 1) tutors who teach English articles

within the data-driven learning and student-centred recommendations found in the EAP

literature (the ‘EAP paradigm’), 2) optimistic rule givers, 3) proofreaders, 4) reluctant error

correctors and 5) principled ignorers. The tutors using the ‘EAP paradigm’ made mention of

noticing techniques, text-based pattern focus, guided proofreading, self-correction and peer

correction. Some EAP tutors see themselves as proofreaders whose main role is to correct

every error they see on the assumption that students learn from frequent correction. The

‘optimistic rule givers’ appear to see the problem as conceptual, the assumption being made

that students have previously misunderstood article use and some sort of clearer or more

accessible explanation can help the students’ understanding. A small number of tutors are

‘reluctant correctors’, overcoming their low confidence or principled objection to focussing on

articles because of perceived student demands. Finally, a small group (principled ignorers) see

any focus at all on article accuracy as contrary to best practices in EAP, irrelevant or

unnecessary or simply too trivial to merit attention.

0 2 4 6 8 10 12

Teaching rules is ineffective/useless

References to natural acquisition/naturalistic…

There are more important things to focus on in EAP

Lack of understanding of article rules (by teacher)

World English/language in flux arguments

Teaching rules only useful at lower levels, not…

A lot of effort for a little or no return

Reasons given for not teaching articles

explicitly

122

4.4 Conclusion

The anecdotal observation that many presessional teachers at the researcher’s institution teach

article use was confirmed. Survey and interview data showed that all respondents at EISU had

somehow focussed their students on article use at least once in the preceding term. Moreover,

across the wider sector, the survey of BALEAP tutors suggests that the overwhelming majority

of EAP tutors ‘sometimes’ or ‘occasionally’ focus on article use in their classroom. One

unexpected finding was that EAP tutors appeared to be marginally more likely to focus on

English article use than EFL tutors (as represented by the ELAL MA list). A small minority of

‘principled ignorers’ see any attention to articles as inconsistent with contemporary EAP

approaches, but the overwhelming majority appear to be teaching article use within text based,

data-driven and student-centred approaches consistent within the ‘EAP paradigm’ outlined in

Chapter 2. A smaller number of EAP tutors across the UK appear to also be optimistically

looking for a simpler/clear rule to teach article use with decontextualised examples, which is

slightly inconsistent with the ‘EAP paradigm’.

However, while most teachers mentioned focussing and teaching approaches that contribute to

learner autonomy, many appear to be overcorrecting article errors in a more teacher-centred

manner more divergent from the recommendations found in the EAP literature. When asked

if they would have corrected the article errors in regular low stakes homework tasks, the

overwhelming majority of EAP teachers surveyed responded positively. This confirmed the

researcher’s earlier anecdotal observations and appears to confirm previous research that has

shown a marked difference between EAP tutors’ stated beliefs and their actual practices

(Etheringon & Burgess, 2002; Barnard & Scampton, 2008). Based on the responses at

interview it is likely that many teachers (termed ‘proofreaders’ in this study) continue to believe

that correcting all students’ article errors will lead to improved accuracy over time. It is also

123

likely that many are ‘reluctant correctors’, basing their behaviour more on what students appear

to want or peer behaviour than their own beliefs about the efficacy of such methods. These

findings confirm Borg’s (2003) proposal that a broad range of factors beyond the published

best practices affect teacher pedagogy.

Regarding error correction, the practices described by two teachers at interview (modelling

proofreading skills and training students to self-correct and peer correct) would appear more in

tune with the commonly recommended pedagogy in the EAP literature. Judging from the

BALEAP survey responses, many EAP tutors across the UK know about these methods and

see their worth. Such practices certainly appear to pose less dangers of confusing or

demotivating learners and allow students to leave the EAP presessional ready to monitor their

own writing without the teacher’s help. It could be argued, figuratively speaking, that most

EAP teachers embrace these principles and ‘talk the talk’ but are not ‘walking the walk’ since

they correct their students’ article errors more indiscriminately than most experts would

recommend.

Due to the fact that ELAL MA participants are less likely to work in English for specific

purpose contexts, as many Masters students work in EFL contexts as well as English for

General Academic Purpose contexts, they had been assumed to be likely to hold a more

communicative teacher outlook than the BALEAP list. Although some of these tutors are

delivering academic writing lessons, it had been generally expected that they would show

markedly different attitudes either because of the widespread use of an integrated lexical and

grammar based curriculum in EFL (a reason article teaching might be more frequent) or

because of their greater focus on oral skills and reduced need to focus on long academic writing

texts (a reason for less teaching). It seems possible these two dynamics cancelled themselves

out. For whatever reason, the MA list teachers overall appeared slightly less likely to explicitly

124

teach or correct article errors. Moreover, there are a high number of these teachers working in

[-Article] L1 countries (the biggest two markets are Japan and Korea), but the impact of this

work setting was not possible to ascertain without further interviews.

There are remarkable similarities across the three groups of teachers surveyed, particularly so

given the diversity of teaching contexts from which the teachers came. However, some

limitations must be acknowledged in this study relating to this variance. The wide range of

backgrounds and failure to control variables such as level, programme type and the failure to

collect biographical information about the teachers, could be argued to be invalidating features

of this study. They certainly require a degree of caution to be exercised in the interpretation of

these findings. A second limitation of any survey-based piece of research is the tendency of

teachers to report their opinions and practice selectively. Moreover, examples of errors were

given to the survey respondents to help them frame their answers meaningfully, but it is

conceded that these errors had more gravity than the errors teachers typically face in EAP (thus

different examples may have gained slightly different results).

A third limitation is the simplistic framing of the topic of correction in questions without

nuancing how students are trained to use feedback or reference to new forms of electronic

feedback (Hyland & Hyland, 2006). This study has not researched how tutors might direct

students to use their feedback, for example using data-driven approaches (Crosthwaite, 2017).

Future studies would benefit greatly from investigating new forms of feedback regarding article

use, particularly since it may be this technology that is encouraging tutors to correct errors so

frequently. Finally, in retrospect it is acknowledged that the study could have gone deeper into

the question of why tutors appear to go against their own stated beliefs and occasionally teach

and correct articles inappropriately. Judging from the Chinese L1 tutor’s explanation of

125

Chinese students’ experience in China, this is likely to be linked to the teachers’ perceptions

of their learners’ expectations and this would be an interesting area of future research.

Notwithstanding the limitations, this survey of teachers’ attitudes and practices has suggested

some conclusions that provide a foundation for the future studies presented in this thesis. Most

EAP teachers clearly believe that article use can and should be focussed upon at some point in

an EAP programme. These teachers would benefit from a greater understanding of the

challenges faced by Chinese learners and how different contexts could be taught within the

pedagogic approaches most advocated in the EAP literature. If EAP teachers can be persuaded

to stop excessively marking article inaccuracies, it would be helpful to provide more evidence

to counter their assumptions and, moreover, to recommend an alternative range of peer-

marking / self-correction type activities that can satisfy their students’ demands for help in this

area.

126

127

5 LONGITUDINAL STUDY OF THE EFFECT OF TEACHING/

CORRECTION ON 24 (L1 MANDARIN) LEARNERS OF ENGLISH

ON A 15-WEEK PRESESSIONAL ENGLISH COURSE

The research reported in this chapter proceeded to investigate the effect of explicit teaching

and correction on English article accuracy in a longitudinal experiment in which control groups

and post-teaching measures were used to overcome the methodological challenges discussed

in Chapter 3. The previous chapter examined the extent to which tutors focus on English article

accuracy in EAP and, having established that many EAP teachers are teaching/correcting

article use, the question now turns to the effectiveness of any such approaches. As highlighted

in the preliminary study presented in Chapter 3, investigating the efficacy of pedagogy poses

many challenges. In the first preliminary study, students made fewer errors when given extra

sessions on English article use, but may have made fewer article mistakes due to their increased

focus on ‘monitoring’ their language. This necessitated the analysis of any effects after the

teaching/error correction had stopped. The preliminary study highlighted the difficulty of

investigating causal relationships between the teaching/correction of the article system and

improved accuracy. This study therefore sought to control variables with 1) an experimental

group which was given extra teaching and article corrections; 2) an ‘error-corrected’ group

who would have extra corrections of their article errors (but no extra teaching) and 3) a third

control group (with no extra attention focussed on articles through either error correction or

teaching).

It should be noted that this chapter focusses on the effect of teaching and error correction. After

reviewing the research question and the methodology, the discussion of the results is limited to

describing and statistically analysing the effects of the experiment. Thus, references to the

128

literature and the issues determining why Chinese students made the reported errors are

discussed in the deeper analysis of the different article types presented in Chapter 6.

5.1 Study overview

5.1.1 Recap of Research Question and hypotheses

This study focuses on Research Question 4:

Does the explicit teaching or correcting of English article usage lead to more target-like

use of articles by Chinese L1 university students whose L2 English displays

inaccuracy?

When designing this study it was predicted that teaching, through ‘consciousness’ raising and

encouraging ‘noticing’ of article forms, would improve students’ accuracy (Master, 1990,

1995, 1997; Bitchener & Knoch, 2008). Similarly, after being focussed on the single

grammatical issue of article use, it was hypothesised that the error corrected group would also

make significant improvements in their article accuracy (Bitchener, Young & Cameron, 2005;

Ferris and Kurzer, 2019). There nevertheless remained questions about whether such

conscious raising leads to sustained improvements or more temporary accuracy gains (Master,

1997), so a post-study analysis over a month after cessation of extra treatment was required to

examine whether any improved accuracy rates would be maintained.

5.1.2 Chapter Overview

Section 5.2 outlines the methods used to source and compile a longitudinal corpus of student

essays and how the different groups were taught during the a Presessional. Section 5.3 presents

the application of the ethical considerations negotiated with the Ethics Board. The final

evolution of the data analysis approach, as discussed in the preliminary study, is presented in

Sections 5.4- 5.5 and the steps taken to check the reliability of the annotation system are

129

discussed in Section 5.6. After reporting and analysing the results in Section 5.7, the

conclusions are drawn in Section 5.8.

5.2 Methodology

5.2.1 Experimental design

As presented in Figure 36 below, three groups of students received differing pedagogical

treatment. In addition to an experimental group (given extra instruction and corrections in

article use), responses to the same essays were taken from a group given extra error-corrections

and a control group that had no extra attention focused on English article use. The first data

collection was in week 1 of the Presessional (before teaching/correction began), the second

essay was submitted at the end of the extra attention focussed on articles in two groups (week

8) and finally six weeks after extra teaching/corrections ceased (week 14).

Figure 36: Three group experimental design

Data taken pre-treatment (week 1), week 8 and week 14

Analysis

Weeks 1 - 7: Three groups have different treatment

Teaching/correction Corrections only No increased attention

Recruitment

Experimental group Error-corrected group Control group

130

5.2.2 Data

As discussed in Section 1.2.2, in the UK a great deal of EAP teaching activity takes place each

summer on ‘presessional’ programmes which allow students who have not reached English

proficiency entry requirements to enter their university, without retaking an English language

test course, upon satisfactory completion of the programme. The next section will provide

details about this programme to contextualise the experimental study.

The Business Management English Presessional Programme

The intensive summer presessional programmes at the University of Birmingham are divided

into four strands: foundation, undergraduate, the Business Management English (BME)

programme and a more general postgraduate EAP programme (for all other disciplines in

which students are not streamed into more specific discipline-related classes until the 10-week

stage). Students can select from a range of presessional course durations, depending on their

English proficiency and their programme entry requirements: 42-week, 31-week, 20-week, 15-

week, 10-week or 6-week courses. The data in this study is taken from presessional students

on the 15-week BME programme, a large English for Academic Purposes Presessional

programme for future Business School postgraduate students that runs each summer at the

University of Birmingham. At the time of this study, the BME 15-week Presessional ran six

classes with between 11 and 12 students in each class and the researcher was teaching one

class.

The BME was chosen for three reasons as follows: 1) convenience of sampling (the researcher

taught regularly on the BME programme); 2) Business subjects are the most popular subject

among Chinese students studying in the UK (The Higher Education Statistics Agency, 2019);

and 3) the BME allowed for the tightest control of variables.

131

BME students in 2010 were predominantly from China (and Taiwan), in addition to a much

smaller number of other nationalities (Thai and Arabic L1 backgrounds). The students in this

study were all preparing to study on one of four main Business School postgraduate

programmes: MSc Marketing, MSc Business and Management, MSc Accounting and Finance

or MSc Marketing communication.

The officially published range of levels of students admitted to the programme ranged from

low B2 to middle B2 level English proficiency on the CEFR framework. However, some

students were accepted with 4.5 IELTS writing level (B1 writing level) on the basis of their

overall IELTS 5.0 band in all four IELTS tests, which means students who presented as high

B1/low B2 borderline were also accepted on the presessional (but excluded from this study).

The entrance requirements and eventual pass mark for students on the BME reflected the three

levels of entry to the university’s Business school (ranging from an overall IELTS score of 6.0

to a requirement of 7.0). Most Business programmes at the time of the study required an

IELTS entrance level of 6.5 overall, so the overwhelming majority of the 15-week programme

students entered with an IELTS score of either a 5.5 or 6.0. The 15-week programme attracted

no students with a 6.5 IELTS score or above because such students were accepted onto the 10-

week programme no matter the department.

In addition to IELTS grades, a writing test on entry was used to identify the students who might

be in most need of language support as the course progressed. This assessment comprised of

a 300-word essay completed in 30 minutes, marked by two programme coordinators using the

same BME criteria used for assessing the end of programme exam. The writing criteria were

132

designed to mirror IELTS criteria because all students were measured against IELTS

equivalents by the end of the presessional. Marks were allocated as percentage scores and a

mark of 55 was equivalent to IELTS 5.5 and a grade of 60 to IELTS 6.0.

Participants

Given the potential challenges of analysing grammatical errors of students below an IELTS

5.5 level of proficiency data was collected only from students with at least 5.5 IELTS overall

scores and no less than IELTS 5.0 in writing. Since the lowest scoring presessional students

had all been grouped into three lower level classes, this was achieved by recruiting participants

for the study only from the three higher groups. Twenty-four of the twenty-nine Chinese L1

students in these three classes consented to participate in the research.

As shown in Table 8, 15 participants were Chinese and nine were from Taipei/Taiwan, but all

declared Mandarin to be their first language. The ages of the students ranged from 22 to 31,

with an average English learning history of eight years (in their home countries). No student

had been in the UK for more than three months; 22 students had arrived in the UK the day

before starting the programme while two students had been studying in the UK for 5-10 weeks

on a previous presessional programme before joining the 15-week cohort. No students had

lived in the UK or other English-speaking country for extended periods prior to arrival. All of

the teaching groups also included other nationalities and several Chinese students who refused

to give consent, but these are not referred to further in this study.

133

Table 8: Participant information

Control group

Error-corrected

group

Experimental

taught group

Summary

Age

22-25

23-31

23-26

Median: 23

Male

Female

Male 8

Female 16

Chinese

Taiwanese

15 Chinese

9 Taiwanese

IELTs level 5.5

6.0

16 IELTS 5.5

8 IELTS 6.0

BME entrance

writing test score

Range: 56-61

Median: 57

AD= 1

Range: 53 -59

Median: 5.5

AD= 1

Range: 56-61

Median: 57

AD= 2

Mean:56.8

SD 2.158

Length in the UK

or English L1

countries at entry

7 students under 1

month. 1 student

under 3 months

All students under 1

month

Seven students

under 1 month, one

student under 2

months

No students over 3

months in the UK

English L2 proficiency level

Another reason for choosing the 15-week BME programme students was their narrow range of

proficiency levels, allowing this study to report findings based on a more homogeneous

proficiency level, with a smaller range of levels than those seen in studies such as Díez-

Bedmar & Papp’s (2008) in which participants had proficiency levels reported on a far wider

134

spectrum (IELTS 4.5 to IELTS 7.00 levels). Sixteen participants in this study had achieved (in

the previous two years) a 5.5 band score in IELTS tests while 8 students had achieved an overall

IELTS score of 6.0. However, as two students had studied five weeks of English immediately

before joining the 15-week presessional, the written placement test was used as the preferred

measure of written English proficiency. As mentioned, this test used marking criteria based on

IELTS bands and the two examiners of this placement were highly trained (both had been

IELTS and Cambridge examiners) and had undergone intense standardization.

The scores in the placement test across the groups ranged from 52 to 61 with a median score

of 57, and a variance of 4.3/standard deviation of 2.08 (see Figure 93, p.341 in Appendix 10

for further information). Non-parametric tests were applied to the data given the small sample

size and slight skew. At the time of this study on the BME programme, all students scoring 54

or under were offered extra 1:1 language support tutorials and were generally put into

separately streamed groups. Although there was some marginal variance in students’ writing

proficiency within and across groups, with one student scoring an IELTS 5.0 borderline grade

of 52, these placement test scores showed there were no significant differences in writing

ability by group on entry as shown by an independent samples Kruskal-Wallis Test (n=24, H

(2) =2.143, p = .284) at the .050 level of significance (hence P ≤ 0.05).

5.2.3 Data sample

As discussed in Chapter 3, choosing a learner corpus data source is always a balance between

the competing demands of authenticity and variable control. The main BME 3,000 word project

writing assessment was not chosen as a data source because the topic was selected by each

student individually. Although this data would have provided rich insights into the authentic

contexts causing challenges for article use to Business School students, comparisons between

students and groups would have been meaningless. One solution might have been to use the

135

placement test responses or set another short IELTS type 250-word essay in controlled

conditions but, as outlined in Chapter 3, such task prompts are arguably less realistic and limit

the inferences that can be made about performance in academic writing. The study therefore

needed to collect more valid representations of academic work which required the use and

synthesis of sources and more developed academic argument (with examples and citations).

On the other hand, the tasks could not overburden the students while they worked on their

writing project and other case study assessments, so minimum words counts were kept realistic

(although students were allowed to go over the suggested word limit without penalty).

The three essay prompts used are summarised in Table 9 overleaf. In total, 27,940 words were

collected for analysis. The prompts followed the weekly topics studied on the BME each week.

The topic of week 1 was leadership and motivation, weeks seven and eight reviewed topics of

marketing and accountancy and the students had read several case studies about loyalty

schemes and brand management in the weeks preceding the final week 14 task. The growing

demands of the tasks set reflect the growing skills that could be demanded of novice academic

writers. The course introduced referencing and the paraphrase of sources from week 1, but in

Task 1 it was decided to require only 350 to 400 words and general reference to the theories

they had studied (with or without citations). As can be seen from Table 8, this prompt gave the

students a somewhat familiar argumentative question type which they would have presumably

have faced in IELTS exams. Later in the course (for Tasks 2 and 3), however, students were

asked to write more critically and provide greater references to the literature.

136

Table 9: Essay prompts

Rubric

Texts

Total

number

of words

Minimum and

Maximum

lengths and

Median word

length

Task 1

(week 1)

Does fear motivate employees to work harder?

Illustrate your ideas with reference to theories

of motivation studied on the course. (350 - 400

words)

9030

Minimum: 302

Maximum: 440

Mean: 376.3

words

Task 2

(week 8)

An effective marketing strategy is crucial to the

success of a business. Critically assess the value

of segmenting, targeting and positioning as a

marketing tool. Illustrate your answer with

suitable examples. (400-600 words)

*option for the three accountancy students:

Summarise the responsibilities of accountants

and discuss how they can contribute to efficient

information flow both within and outside

business organizations.

11383

Minimum: 474

Maximum: 605

Median: 474.3

words

Task 3

(week 14)

Retailers compete for customer loyalty and

retention in many ways.

Examine the issues, illustrate with examples and

suggest ways in which retailers can promote

loyalty and retention. (400-600 words)

7527

Minimum: 403

Maximum: 667

Mean: 470.4

words

In each of the groups there was a student progressing to an accounting and finance course.

Therefore, these three students were allowed to do an alternative essay title relating to

137

accounting in week 8. Since the aim of the research was to measure the comparative progress

of the groups and since there was one accountancy student in each group, this did not

dramatically affect the homogeneity of task type. A further complication was that one student

was automatically referred to a language support tutor because of his low score in the writing

test, and another student was referred to a tutor due to her lack of confidence in speaking. To

avoid external influences on English article accuracy in the study, the two language support

tutors agreed not to pay any particular attention to the subject of English articles in any of their

1:1 sessions.

Despite these efforts to control the teaching input around article use, it could be argued that the

homework nature of the tasks make the essay responses less reliable than studies which went

to greater lengths to control the environment, i.e. the opportunity for external help and time

spent on tasks (Bitchener & Knoch, 2009). However, as concluded in Chapter 3, such

controlled conditions do not accurately reflect the process of academic writing in the real

world: students make many careless mistakes under time pressure and the planning/drafting/

proofing stages allowed in this study are more characteristic of academic writing. In short, lack

of control is acknowledged as a limitation, but the greater validity is considered to have

outbalanced this risk. Moreover, while all essays were written for homework with full access

to spell check/word-processing tools (and no control to stop third party help), it should be

stressed that these tasks were not ‘high stakes’. Students therefore had little reason to use essay

mill services or proof-readers. Furthermore, while the students theoretically had access to

electronic tools at home, the use of electronic technology to aid writing accuracy was less

advanced when this data was collected in 2010.

In fact, at the time of this data collection (2010), most students were using very basic word-

processing software. The University of Birmingham did not make a licensed version of Word

138

for windows available to students until 2012 and even the most up-to-date word-processor

packages with fully updated English language settings (which few students had at this time)

rarely corrected article errors. In addition, while today’s standard of technological tools are

clearly influencing the language accuracy of student work, it should be noted that in 2010 the

students did not have modern translation software of today’s standard or grammatical proof-

reading tools such as Grammarly.

5.2.4 The BME course content

The students on the Business Management English programme were chosen because they had

the most tightly prescribed teaching scheme of work (i.e. teachers had very little choice about

each day’s lesson materials). While this course cannot be defined as an ‘English for Special

Purposes’ course because of the broad range of disciplines studied at the Business School, it is

far more specialised than many other UK programmes which focus on more generic EAP

topics. With regards to this study the specialised nature of the BME meant that the student data

provides rich insights into the writing errors made by novice academic writers with Chinese

L1 backgrounds writing in English on business related topics and genres for the first time. It

secondly allowed most variables, such as how articles were taught, to be controlled.

The three teachers (the researcher and two further tutors) were experienced teachers with

DELTA/TEFL Q level qualifications. All had over 5 years of EAP teaching experience and

previous BME teaching experience. The main text input on the BME course is a weekly

Business case study and the teaching timetable was in 2010 divided into six main skill areas:

academic writing, working with Business case studies, presentation skills, Business feasibility

study research skills, listening skills, and seminar skills. To an even greater extent than other

presessional programmes, there was little time in this highly intensive programme for

reviewing basic language concepts. Indeed, as shown in an audit of the programme’s main

139

materials presented in Table 10 below, there was little specific attention to article use in the

scheme of work beyond proofreading exercises. It was clear that teachers might make

reference to articles in nominalisation activities, but this was a teacher choice (and could be

influenced in this study).

Table 10: BME Teaching materials

Programme Materials

Coverage of English Articles

Working with Case Studies Materials

No mention

Academic Writing Materials

Only mention of article use in Proofreading exercise (see

Appendix 7: 323). The use of articles was not explicitly

mentioned in the sections on nominalisations or noun phrase

density.

Academic Reading Guide

No mention

Feasibility Study (extended writing)

Materials

No mention

Presentation and Seminar Skills Guide

One mention of definite article stress in a speaking pronunciation

exercise

The only marginal attention to articles demanded by the programme curriculum was therefore

in the proofreading activities which were normally covered towards the end of the course.

Admittedly, all the teachers in 2010 reported that they corrected article use in their student’s

writing on a regular basis. Moreover, there was nothing to stop teachers using their own

supplementary materials or discussing their use in ad hoc mini-teaching situations in response

to student queries. Such possibilities presented variables that needed to be addressed in the

context of this study and the following section explains how these variables were tightly

controlled.

140

5.2.5 Study procedures

The two teachers who participated in this study (in addition to the writer) agreed to limit how

they taught articles and to allow the recruitment of participants. Homework tasks were

normally at the discretion of each individual teacher, but these colleagues also agreed to make

a shared essay title mandatory for three weekly homework tasks.

Experimental taught group

The experimental group consisted of eight Mandarin L1 speakers. This class was the

researcher’s own class, already part of the course. Five supplementary thirty-minute English

article sessions were delivered in five extra sessions (given in hours outside their normal

teaching contact time) in weeks 2 -7, as shown in Table 11 overleaf. As can be seen, students

received both teaching and extra article corrections in their homework feedback. All teaching

and extra correction of articles ceased from week 7 in order to measure (in week 14) whether

improvements could be sustained when students’ consciousness about article errors was no

longer being raised (by Task three there had been no mention of articles in class for four weeks).

The decision to revise the students’ article use using the Bickerton Framework was made

because the students in the preliminary study had a) seemed familiar with the [+/-SR] concept

and b) had given good feedback about the whole system when presented.

141

Table 11: Five extra English article sessions

1. Session 1 (week 2)

Introduction and review of process of proofing

for article errors and typical errors to check.

Brief introduction to the Bickerton Framework

Students were also asked to read reference

materials for homework.

2. Session 2 (week 3)

Students asked to analyse definite article uses

in a text already analysed in class for meaning

(uses underlined). Students then presented

with the concepts of [+HK] [+SR] from

Bickerton Framework. Students were asked to

proof their week 1 essays and discuss errors

found in week 4.

3. Session 3 (week 4)

Review of homework, leading to analysis of

each other’s errors (teacher cut strips of paper

with examples from week 2 homework).

Further analysis of text from course materials

– further uses underlined and then matched

with concepts of [+HK] [+SR] and [-HK]

[+SR].

4. Session 4 (week 6)

Special attention to generic contexts [+HK] [-

SR] through the Bickerton Framework. First

through prototypical examples and then

through looking at list of errors made in the

class. Further reading and grammar exercises

set for homework (Swales and Feak, 2004, p.

395-402, Tasks 2 & 3).

5. Session 5 (week 7)

Review of homework. Review of article errors

related to proper nouns and idioms.

6. Participants invited to voluntarily take

part in six follow-up post-study one-to-

one sessions.

In the term following the presessional, three

volunteers for the post-study sessions had

various article uses underlined. They were

then asked if they were correct/an error or a

mistake. The teacher and participant then

discussed the reasons for errors.

142

5.2.6 Written article correction group

The article-corrected group consisted of eight Mandarin L1 students. This group received no

formal teaching of article use but did have all their article errors corrected. Students in this

group attended an extra session with the researcher in week 3 where the importance of

correcting work for article errors was stressed and useful proofing strategies were revised. In

addition, by prior agreement with the teacher of this group, extra emphasis was placed on the

written correction of article errors (i.e. written article errors were simply underlined without

explanation at least once a week between week 4 and week 8). Corrections were paper based

(not electronic versions). The students did not have to re-write the sentences. No article

grammar instruction was given except a very brief recap of the article system in week 1. Error

correction stopped in week 8.

The teacher of this group had over seven years’ experience teaching EAP and three years’

experience teaching the Business Management English course. She reported that her normal

approach to correcting article errors was contingent on the length of EAP course, the number

of article errors (and students’ other grammatical errors), gravity of error and the purpose of

the writing task. However, for the duration of this study she agreed to provide written error

corrections for all article errors in her students’ writing for four weeks. The teacher described

her normal approach to teaching articles, if she had not taken part in this study, as ‘ad hoc’.

She did not normally devote a formal teaching session to their use because of the time pressures

of the course. However, in normal teaching conditions she reported that she would revise article

use if opportunities presented themselves more spontaneously in the classroom. She reported

that she most frequently reviewed them in previous years in tutorials if students had questions

(if they wanted to discuss their feedback). However, for the purposes for this study she agreed

143

to avoid teaching articles in the classroom and tutorials beyond the clarification ‘you have made

an error with the article’ if students wanted to discuss their homework feedback in tutorials.

5.2.7 Control group (1)

Control group 1 consisted of eight Mandarin L1 speakers. Overall, this group’s IELTS grades

were marginally higher than the other two groups at the start of the course, but this was mostly

due to the students’ speaking scores. More students in the control group (4) had higher IELTS

6.0 writing scores than in other groups (2), but their BME writing entry test scores were almost

identical to the other group ranges. No extra article teaching or written corrective feedback for

articles was given to this control group over and above the first page of the draft of one high

stakes assignment in week 9 (as per the teacher’s normal practice).

This teacher of this group reported that, apart from occasional ad-hoc clarifications, she would

normally refer to article use only in week 12 of the course (when she completed the end of

course proofreading activities). With regards to corrections, this teacher reported that when she

started her career in teaching, she had ‘overcorrected’ all errors, but now seldom corrected

article errors in homework feedback. She also gave students a lot of proofreading help at the

end of course when students submitted their extended writing projects and essays. However,

this was not a problem for the purposes of this study and she readily agreed to not correct or

teach article use in the time periods agreed.

5.3 Ethical issues

As discussed in Chapter 3 (Section 3.7), the Ethics Board had stressed the importance of not

disadvantaging students from not participating in the study, allowing for informed consent and

reporting back any beneficial findings to students in their responses to initial ethical review

144

applications (ERN_10-0375 amended 14-0375). This section will examine how the agreed

ethical safeguards were enacted in the main study.

Consent

Students were given an information sheet on the 2

day of induction and at least 24 hours to

read some information about the study (see Appendix 4). The researcher then visited

classrooms to orally present the study’s objectives. It was explained that students could receive

the same extra resources at the end of the study even if they did not choose to participate.

Students were then given a further 24 hours to reflect and were asked to bring their forms in

for their class teacher (signed or unsigned) the next day.

Information given to participants

Students were given full information about the research, but only informed of their group after

agreeing to take part in the study. They were reminded at this point that they could still

withdraw their consent.

Sharing of results with participants

The Ethics Board insisted on a workshop in week 14 at which point some of the researcher’s

preliminary findings (common errors identified) were presented to anyone on the BME who

wished to attend regardless of whether they had taken part in the study. At this lunchtime

session a worksheet about article use (see Appendix 5) was also shared with attendees.

Participants were sent a conference paper published on the university website that summarised

the study findings six months after the study.

Participant feedback

To check students had benefitted from participating in the study, all students were asked to

complete an end-of-study feedback form in week 14 (see Appendix 6).

145

5.4 Constraints of the study

As planned, at the end of the ‘intervention’ in week 8, in which one group was explicitly taught

article use rules and a further ‘correction’ group had all their article errors underlined, all

experimental pedagogical treatments ceased (students were not taught the article system and

no extra emphasis was placed on the correction of errors). After a six week period in which

students could ‘forget’ the study, a third essay was then set in week 14 in order to investigate

whether any learning was sustained. Unfortunately, due to a misunderstanding (the teacher set

a different essay for homework), it was not possible to collect a third third essays from one of

the control groups (the impact of this issue is discussed in Section 5.7.6).

5.5 Data analysis

For consistency in the distinction between head nouns and any nouns used as modifiers, head

noun phrases were extracted from the corpora automatically using a Java programme

developed by Oliver Mason at the University of Birmingham. Errors were then identified

manually by the researcher and tagged using macros in word (before conversion to text format).

In this study, every single head noun phrase was tagged for whether the it had a correct article,

an incorrect article or an alternative determiner.

5.5.1 Tagging system

The word macros were developed following the tagging system (see Table 12 and Table 13)

presented by Díez-Bedmar & Papp (2008) and many replicated studies (Díez-Bedmar, 2015;

Crosthwaite, 2016a) based on the Bickerton Framework as outlined in 2.2.1.

Following Ekiert (2004), who argued that articles for idiomatic and other conventional uses

are very different to other referential and generic uses, all fixed academic expressions and

articles used with proper nouns were denoted as a ‘Type 5’ error. Type 6 contexts (other

146

determiners used) will also be examined in the next chapter, when the underuse or overuse of

articles is discussed as compared to L1 use.

Table 12: Tagging system for correct uses (Díez-Bedmar & Papp, 2008)

Determiner used by the learner

Generic

Referential

Definite

Referential

Indefinite

Non-

referential

Idiomatic

use or

proper noun

Other

determiner

Definite article

1DA

2DA

5DA

Indefinite article

IIA

3IA

4IA

5IA

Ø articles

1ZA

3ZA

4ZA

5ZA

Table 13: Tagging system for incorrect uses (Díez-Bedmar & Papp, 2008)

Determiner used by the learner

Generic

Referential

Definite

Referential

Indefinite

Non-

referential

Idiomatic

use or

proper noun

Other

determiner

Definite article

1GADA

2GADA

5GADA

Indefinite article

IGAIA

3GAIA

4GAIA

5GAIA

Ø articles

1GAZA

3GAZA

4GAZA

5GAZA

Based on Bickerton (1981) and adapted by Thomas (1989) and then Díez-Bedmar & Papp (2008)

147

In addition to the tagging of the extracted noun phrases, the student data was also tagged with

‘^’ omission error and ‘$’ oversuppliance annotations. Misuse article errors (a/an errors, the

for a/an, a/an for the) were also annotated with a ‘$^’ code. 5383 noun phrases were inputted

onto a spreadsheet as the researcher annotated each text. In order to check for data entry error,

the data was later checked by running frequency counts on each text using a concordancer to

check the entries, and spreadsheet tallies were recorded.

5.5.2 Pica’s Target-Like Use

Following other researchers (Pica, 1983; Lu, 2001; Díez-Bedmar & Papp, 2008; Díez-Bedmar,

2015; Crosthwaite, 2016a) the study used a ratio measure that allowed for comparison of

accuracy across different text lengths and comparisons with other studies. Across the literature,

this equation is called either Pica’s (1983) coefficient of ‘Target-Like Use’ or ‘Target

Language Use’ (TLU):

𝑇𝐿𝑈 =

𝑛. 𝑐𝑜𝑟𝑟𝑒𝑐𝑡 𝑠𝑢𝑝𝑝𝑙𝑖𝑎𝑛𝑐𝑒 𝑖𝑛 𝑜𝑏𝑙𝑖𝑔𝑎𝑡𝑜𝑟𝑦 𝑐𝑜𝑛𝑡𝑒𝑥𝑡𝑠

(

𝑛. 𝑜𝑏𝑙𝑖𝑔𝑎𝑡𝑜𝑟𝑦 𝑐𝑜𝑛𝑡𝑒𝑥𝑡

)

+ (𝑛. 𝑠𝑢𝑝𝑝𝑙𝑖𝑒𝑑 𝑖𝑛 𝑛𝑜𝑛 − 𝑜𝑏𝑙𝑖𝑔𝑎𝑡𝑜𝑟𝑦 𝑐𝑜𝑛𝑡𝑒𝑥𝑡𝑠)

5.5.3 Chosen notions of grammaticality and acceptability

As highlighted in the preliminary study, it was important to establish a clear policy regarding

ambiguous grammar and the line between acceptable informal tone and inappropriate register.

The researcher was interested in overspecification, but the preliminary study had shown the

issue often caused disagreement among researchers (i.e. about whether a redundant definite

article was simply ‘marked’ or completely inappropriate). As the reliability checkers (see 5.6)

would need to know these policies, many examples were given in training and standardisation.

Figure 37 shows an extract from the training material that helped standardise this approach.

148

Instructions for annotators

Please mark noun phrases as incorrect <GA> only if they were impossible to accept as

grammatical or completely inappropriate in the academic style of the assignment task.

You are asked to ignore cases where the ambiguous meaning makes a judgement about article

choice impossible.

Omission errors

If a bare noun phrase could have been corrected by either plural form or an article with the

singular form, mark as incorrect (add “^” before noun and <GA> to tag).

Overspecification errors

There will be occasions when the appropriacy of the definite article will create an effect on

for the assignment task set or completely inappropriate (and therefore incorrect). If in any

doubt, please ignore only slightly ‘marked’ uses. For instance, in example 1 below the use of

the is used when many native speakers would have discussed a general statement with a bare

noun in plural form without article. However, as this student has referred to a ‘target market’,

this article could arguably be used as an inexplicit referential to this market and its consumers.

However, please mark as incorrect any overspecification of definite article when no inexplicit

reference can be reasonably inferred or when such reference creates a tone that would be

completely inappropriate for a Business Studies journal article (see example 2). Many

examples will be shown during training to standardise our approach.

(1): In a target market, brand image is very important for the consumers. (acceptable)

(2): For example, the young generation likes to follow $the new trendy stuff.

(unacceptable)

Figure 37: Policy of tagging for correctness

149

5.5.4 The issue of competence and performance errors

As explained in Section 5.2.3, the methods used ensured that most errors can be assumed to be

valid competence errors. Students were trained in how to proof their work and use Microsoft

Word for Windows to spell-check and grammar-check their work. The use of homework

assignments in a relatively ‘high stakes’ presessional course (the students could not enter their

Masters degree programme without success on the course) to some extent assured a level of

student motivation in the tasks.

5.5.5 The issue of final -s errors

The examples below show how singular count nouns (market, strategy, card) could have been

rendered correct either by adding a determiner before the noun or a plural final –s at the end

of the noun to make the count noun plural. In some cases it was completely unclear which

error was more likely.

[1 ] *Segmentation includes diverse points, mainly can separate into three aspects

which are ^ consumer market, business market and segmenting international marketing.

[ 2] * It refers to the process of dividing ^ whole market into smaller segments, targeting

one or certain segments to enter, and establishing a superior position in customers’

mind.

[3] *Organisations should select their professional field and facility compare with

competitors in a particular market, and select ^ concentration strategy or multisegment

strategy.

[4] *…^ Loyalty card is a common method to attract customers to go to the shops

Figure 38: Example 1: Identification of error problems

150

It could be argued that the parallel structures in [1] and [3] and the conjugation of the verb in

[4] display a motivation of the student to use a count noun in singular form (and therefore not

a mistake in omitting the –s). Moreover, most noun phrases seemed to give such ‘clues’ as to

the students’ intention. However, given that the homework tasks were word-processed, all

annotators were instructed to treat all such cases as errors (assuming that students had proofed

the work carefully using their word processors). With this training, all raters annotated the

above examples above as ‘incorrect use of English article or determiner’ with complete

agreement.

5.5.6 Straightforward cases

In fact, the majority of noun phrases appeared relatively straightforward to annotate as ‘correct’

or ‘incorrect’, particularly with regards to the omission of definite articles as shown below.

Errors [1] to [3] in Figure 39 below show examples of overt errors. These three noun phrases,

judged to be incorrect by the main researcher, were later verified during inter-rater reliability

tests when three further human raters independently annotated them as incorrect on two

separate sessions.

[1]* In ^ short time, it could increase customer turnover, but it is useless to attract

high-income customers.

[2]* Indeed, ^ business environment is complex.

[3] * Therefore, it would be hard-pressed to imagine that how these employees coud

work effectively in ^ a long term.

Figure 39: Example 2: Overt errors

151

5.6 Tagging reliability checks and inter-rater reliability checks

Reliability of the tagging was checked by the researcher and then by two L1 English EAP

teachers (DELTA Level qualified, with Masters in TESOL/Applied Linguistics). Three Task 1

responses, three Task 2 responses and two Task 3 responses were randomly selected and

checked by the two teachers. As discussed in the preliminary study, the approach taken to

annotation checking was text-based with annotators asked to check the whole text, since earlier

attempts to check the tagging at the sentence level had produced unsatisfactory results (with

teachers unable to understand the context of the sentences).

Before checking the text annotations, the teachers were given thorough training (prior reading

and a one-hour training workshop). As highlighted in the preliminary study, it was anticipated

that the parts of the framework most difficult to apply consistently to Academic English would

be 1) Generic use (and its distinction from non-referential uses) and 2) the difference between

Type 3 (indefinite) and Type 4 (non-referential uses). A further extract from the training

materials in Figure 40 shows how annotators were trained to use context and apply the

framework consistently. After going through some examples in the workshops, the annotators

standardised further by annotating one example text and comparing decisions.

Annotators were given the texts printed on paper in their entirety so that they could check the

context of each noun phrase context. The eight randomly selected texts gave a total number of

745 noun phrases which accounts for over 10% of the total dataset. Analysed with the Fleiss

Kappa inter-rater correlation formula (Fleiss, 1981), in order to account for chance agreement,

the pairwise agreement of 90.25% gave a coefficient of 0.87, showing strong reliability. This

shows that the text-based checking, training and standardization combined to give confidence

to the initial rater’s judgements.

152

Policy for difficult cases in the Bickerton Framework

On occasion, as shown in the example below, you may find it hard to make judgements between

a generic Type 1 context and a non-referential Type 4 context. For example:

A loyalty system helps keep loyal customers.

In this study please tag the above example as generic only if it occurred in the introduction of

an essay, particularly if the writer was defining their terms. If the example appeared in the later

body of the essay and was clearly not a known referent (but rather an unreal prototypical

example), as shown in the example below, please tag as a type 4 context. For example:

Many companies have not considered the best method of maintaining brand fidelity.

<4IA> A loyalty system could help keep loyal customers.

Decisions about whether a context is indefinite (Type 3) or non-referential (Type 4) also require

careful examination of context. The following examples suggest an approach to this sort of

decision:

Example out of context:

Marketing strategies are vital to companies which want to find a successful way to

increase their market share.

In this case, the distinction must be made by reviewing the position in the text and carefully

interpreting the context. If the context of this example was within a paragraph which had

mentioned specific referents, this example would be interpreted as an indefinite Type 3 context,

as shown in example below.

<3ZA> Cosmetics companies in Europe such as Unilever are spending increasing sums

on marketing consultants. <3ZA> Marketing strategies are vital to companies

which want to find <3IA> a successful way to increase their market share.

However, many examples you will find in academic essays relate to hypothetical non-specific

referents, as shown in the example below, in which it clear from the wider paragraph context

that the student is theorising about an imaginary company. In cases of ambiguity, you are asked

to select the same context type as the surrounding paragraph context.

If a company does not have <4IA> a clear marketing strategy, its managers will make

inconsistent decisions about pricing and promotion. <4ZA>Marketing strategies are

vital to companies which want to find <4IA> a successful way to increase their market

share.

Figure 40: Example 2 from annotator instruction materials

153

5.6.1 Dealing with disagreements in the sample checked

While there was general agreement about most annotation decisions, the three raters met to

discuss the tagging which was disagreed. Any tags that could not be quickly agreed

unanimously were then discarded from the dataset.

On decisions of accuracy (about which there were very few disagreements), the disputes mainly

arose in questions of overspecification and were most difficult to resolve when there was

ambiguity about the meaning due to ungrammaticality of the sentence. Errors [1] and [2] in

Figure 41 below show the challenge annotators faced in these situations (both annotations were

removed from the data).

[1] People prefer works in <$ overspecification?> the happiness environment.

[2] * After <$ overspecification?> the spending in the education of customers, if they

could not be maintain, that is a waste of money .

Figure 41: Oversuppiance issue disagreements

Disagreements about whether overspecification of definite articles caused inappropriate

informality for the genre were not as frequent as had been expected. Figure 42 shows two

cases of where raters could not agree about the ‘conversational tone’ created by definite

articles.

[1] *Trust is an important driver while considering how to retain $the existing

customers.

[2] *To sum up, under $the pressure from big supermarkets, it is necessary for retailers

to seek for a reasonable way to attract more loyal consumers .

Figure 42: Appropriacy for genre disagreements

154

In terms of the tagging system itself, the main cause of annotators’ disagreement was in issues

of uses of a/an and whether to tag them as Type 3 indefinites or Type 4 non-referentials. Figure

43 gives two examples of such issues.

[1] * In order to create <3IA or 4IA?> a close relationship with customers, many

companies differentiate their products from other competitors .

[2] *Every company has different kinds of organization structure, but the accounting

department is <3IA or 4IA?> an essential part of <3IA or 4IA?> a company.

Figure 43: Disagreements about whether Type 3 or Type 4

5.7 Results and discussion

After annotating the noun phrases with the 22 framework tags the TLU for each student’s

response in each essay Task was calculated. Following Crosthwaite (2016a), the median value

and median absolute deviations of these TLU values are reported in this thesis, rather than

mean and standard deviations. As can be seen, Figure 44 below summarises the median TLU

Figure 44: Box plot showing distributions of median TLU scores (n=24)

155

values of the three article types across all three Tasks. The distribution of a/an TLU values was

less normally distributed than the and Ø TLU values. Given the small sample sizes and the

slight skew of the distributions, the data was statistically analysed with non-parametric tests.

Table 14 and Table 15 summarise the median values and median absolute deviations (MAD)

for each article context in each Task across all groups. One issue confronted during the data

analysis was the treatment of zero counts. For example, 11 of the 24 participants made no use

of Type 1 generic a/an (1IA), so the TLU is shown as 0.00. The use of the median value of the

24 participants may therefore underrepresent the accuracy of those that used this article.

Table 14: TLU ratings for all groups summarised (n=24) with no adjustment

Task 1

IDA

11A

1ZA

2DA

3IA

3ZA

4IA

4ZA

5DA

5IA

5ZA

Median

0.25

0.00

0.75

0.85

0.75

0.50

0.83

0.92

0.80

0.00

1.00

MAD

0.25

0.00

0.25

0.10

0.25

0.50

0.17

0.05

0.20

0.00

Task 2

Median

0.73

0.71

0.87

0.88

0.89

0.80

0.88

0.96

0.84

0.00

0.97

MAD

0.27

0.29

0.13

0.05

0.11

0.09

0.13

0.04

0.16

0.00

0.03

Task 3

Median

0.75

1.00

0.86

0.83

0.94

0.85

0.71

0.94

0.78

0.25

1.00

MAD

0.25

0.00

0.13

0.09

0.06

0.15

0.21

0.04

0.22

0.25

0.00

On the one hand, the Target-Like Use measure is a robust equation and the literature shows

other researchers leaving these zero counts in the data. In fact, this may show avoidance of an

156

article (an issue that will be examined in the next chapter) and therefore a justifiable and

representative value. Moreover, this issue only affected the rarest article types (generic articles

and 5IA type contexts). On the other hand, with such small data sets and with rarer article uses,

a case could be made for the removal of these zero counts. The effect of this adjustment can

be quite dramatic, as can be seen in Table 15. This shows that removing all instances of zero

use from median calculations, the TLU for 1IA rises from 0.00 to 0.50. Even more dramatically

with the rarest form or article, the TLU for 5IA contexts (use of a/an with proper nouns and

fixed expressions) goes from 0.00 to 1.0.

Table 15: TLU ratings all groups (n=24) with adjustments* for zero counts

Task 1

IDA

11A

1ZA

2DA

3IA

3ZA

4IA

4ZA

5DA

5IA

5ZA

Median

*0.50

0.50

*0.86

0.85

0.75

0.50

0.83

0.92

*0.83

*1.00

1.00

MAD

*0.50

*0.14

0.10

0.25

0.50

0.17

0.05

*0.17

0.00

Task 2

Median

*0.86

*0.88

*0.89

0.88

0.89

0.80

0.88

0.96

0.84

*1.00

0.97

MAD

*0.14

*0.13

*0.14

0.05

0.11

0.09

0.13

0.04

0.16

0.00

0.03

Task 3

Median

0.75

1.00

0.86

0.83

0.94

0.85

0.71

0.94

0.78

*1.00

1.00

MAD

0.25

0.00

0.13

0.09

0.06

0.15

0.21

0.04

0.22

*0.00

0.00

Nevertheless, this mostly affects only these two rarer contexts (1IA and 5IA). Thus, in order to

enable comparisons with other studies, the remainder of this thesis makes no adjustment for

these zero counts. Using the unadjusted values shown in Table 14, the remainder of this chapter

157

examines the quantitative effect of the teaching/error correction of articles. Comparisons with

L1 writers and comparison of TLU scores in other studies will be examined later in Chapter

5.7.1 Improvements in accuracy of article use (pre-treatment to post-treatment)

Examining the summary data visually in Figure 45, Figure 46 and Figure 47 (before statistical

tests are presented in the next section), improvements can be seen in the TLU of all articles

(overall, all types) in the taught group and the error corrected group. Meanwhile, the control

group’s TLU figures started from a higher relative position but then fell for unknown reasons.

Figure 45: Target-Like Use of definite article (Type 1, 2 and 5 combined)

0.4

0.5

0.6

0.7

0.8

0.9

THE TLU essay 1 THE TLU essay 2 TLU The essay 3

Definite article TLU (all types)

Median taught group Error corrected Median control group

158

Figure 46: Target-Like Use of Ø (Types 1, 3, 4 and 5 combined)

Figure 47: Target-Like Use of a/an (Types 1, 3 and 5 combined)

The TLU in 2DA definite article contexts went up markedly between pre-treatment (Task 1)

and post-treatment (Task 2) in the taught and error-corrected groups, as shown in Figure 48

below.

0.4

0.5

0.6

0.7

0.8

0.9

1 2 3

Ø TLU (all types)

Median taught group Median error corrected Median control group

0.4

0.45

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

1 2 3

a/an TLU (all types)

Median taught group Median error corrected Median control group

159

Figure 48: TLU of 2DA definite article (n=24, all participants by group)

5.7.2 Statistical tests: Progress in accuracy made by the taught group in Task 2

The 8 students in the experimental group demonstrated significantly greater accuracy in their

use of definite articles in Task 2 (following extra teaching and correction) than they showed in

Task 1. A Wilcoxon Signed Ranks Test indicated that the median post-treatment TLU (Task

2) of all definite articles combined (see Figure 48) in the experimental group, Mdn=0.86, was

statistically significantly (P ≤ 0.05) higher than the pre-treatment TLU of all uses of definite

article (in Task 1), Mdn=0.79, (Z=33, p =.036). The improvement in accuracy of all Ø articles,

from Mdn TLU 0.84 to 0.91 was also statistically significant (P ≤ 0.05) according to a

Wilcoxon Signed Rank Test (Z=36, p=.012). The same test also found significant

improvements in the 8 students’ article accuracy in 1IA contexts (Mdn TLU rising from 0 to

0.94, Z=27, p=.028), in 1ZA contexts (Mdn TLU rising from 0 to 0.92 (Z=28, p=.018), and

3ZA contexts (Mdn TLU rising from 0.63 to 0.8, Z=26, p=.043).

Table 16: on the next page presents all p values (asterisks* shows significant values).

0.7

0.75

0.8

0.85

0.9

0.95

Taught group Error corrected group Control group

TLU 2DA definite article

Essay 1 Essay 2 Essay 3

160

Table 16 Wilcoxon Signed Rank Test *=statistically significant at 0.05

Framework

Taught group

Error corrected group

Essay 1 v Essay 2

Essay 1 v Essay 3

Essay 1 v Essay 2

Essay 1 v Essay 3

1DA

0.779

0.746

0.233

0.750

1IA

0.028*

0.281

0.610

0.890

1ZA

0.018*

0.058

0.237

0.484

2DA

0.012*

0.161

0.028*

0.779

3IA

0.779

0.500

0.078

0.225

3ZA

0.043*

1.000

0.025*

0.093

4IA

0.161

0.575

0.310

0.499

4ZA

0.2068

0.401

0.161

0.575

5DA

0.933

0.090

0.778

1.000

5IA

1.000

0.591

0.552

5ZA

0.176

0.345

0.684

All the

0.036*

0.889

0.050*

0.779

All a/an

0.262

0.263

0.069

0.025*

All Ø

0.012*

0.161

0.012*

0.401

161

Looking in detail at the Bickerton contexts of use (see Table 16 on preceding page), many

article contexts showed no significant improvements between Task 1 and Task 2 in the

experimental group. Taking a statistically rationalist position that Type 2 errors can occur

through Bonferroni adjustments (Feise, 2002), no statistical controls for Type 1 errors were

applied. Given that multiple testing occurred, the P values in Table 15, particularly those at ≤

0.05 levels of significance, must therefore be treated with a degree of caution. On the one

hand, the improvement in 2DA context accuracy in the experimental group are significant with

or without correction at the ≤0.01 level, rising from Mdn TLU (n=8) 0.82 to 0.89 (Z=36,

p=.012). On the other hand, even at the ≤ 0.05 level of significance no improvement was found

in 1DA or 5DA, 3IA, 4IA, 5IA, or 5ZA contexts.

Progress in accuracy made by the error corrected group

The 8 students in the error-corrected group also displayed significantly higher accuracy in their

use of definite articles in Task 2 (having received extra correction) than in Task 1. A Wilcoxon

Signed Ranks Test indicated that the Task 2 post correction TLU (Mdn=0.89)) of all definite

articles combined (see Figure 48), were statistically significantly (P ≤ 0.05) higher than the

pre-corrections TLU (Mdn=0.78) in Task 1 (Z=32, p =.050). Again, statistically significant

improvements were also suggested by the same test of the TLU of all Ø articles, from Mdn

TLU 0.80 to 0.90 (Z=36, p= .012) The same test (P ≤ 0.05) showed statistically significant

improvements in Type 2DA definite articles (Mdn TLU rising from 0.76 to 0.89 (Z=27,

p=.028), and 3ZA (Mdn TLU increasing from 0.25 to 0.84, Z=34, p=.025).

The effect of group on progress

Without a control group, the case could be made that all progress of two experimental groups

was due to the task effect. Indeed, Crosthwaite (2017: 22) found even greater significant

162

improvements in error-corrected student writing, but concluded that the improvements were as

likely to be caused by choice of prompt topic and the use of word processors than by the

pedagogy:

Despite the overall finding that the number of errors dropped overall between data point

1 and data point 3, closer inspection reveals that any significant drop is likely to be

related to either the topic or conditions under which the writing was produced, rather

than any impact of the online materials taken, or the corrective feedback given to

students at DP1 or DP2

However, the use of a control group in this study for Task 1 and 2 (with the same topic) suggests

that topic was not the main effect of article accuracy between these two Tasks. The control

group completed the same two Task. On the one hand, this group admittedly started from a

higher accuracy level. On the other hand, their decline in the TLU of all definite articles was

found (from Mdn TLU 0.93 to 0.86) contrasts to the relative Mdn TLU rise in the experimental

group (from 0.79 to 0.86). The control group’s decline was mainly accounted for by a decline

in 2DA definite article TLU accuracy (from .92 to .82). Analysing the progress made by the

experimental and control groups, an independent Mann-Whitney U Test showed that the

difference in progress in TLU between the two groups of all definite articles (U=52, p=.005),

2DA(U=54, p=.021), and all Zero articles combined (Z=57.000, p=.007) was significant (P ≤

0.05).

Comparing the progress of the error-corrected group and the control group, again several

improvements in TLU were shown to be significant (P ≤ 0.05). The Mann-Whitney U Test

demonstrated that the Mdn 0.11 improvement in TLU of all definite articles combined,

observed in the error-corrected group, was significant (U=53, p=.038) when compared to the

lack of relative progress in the control group. Compared to the control group’s decline in

progress, the same test also found significant (P ≤ 0.05) progress (U=55, p=.015) in the error

163

corrected group’s progress in 2DA articles (rising by Mdn 0.06 TLU) and the Mdn 0.08 TLU

improvement in all Ø articles (U=60, p=.002).

As visualised in Figure 49 overleaf, if it is assumed that the variable of topic was having a

downward effect on accuracy and that no other variable affected performance, this finding of

improvement between 1

and 2

Task would suggest that the two interventions (teaching and

error correction) impacted upon the observed increase in accuracy in some article contexts.

The progress made by all individuals in all three groups between Task 1 and Task 2 was

statistically analyzed by group and a Kruskal-Wallis Test showed highly significant (p ≤ 0.01)

differences in progress in Target-Like-Use of all Ø articles combined across all three groups

(H (2) =11.195, p= .004). Significant differences (p ≤ 0.05) in the progress made in definite

article 2DA contexts across groups were also suggested by the Kruskal-Wallis Test (H

(2)=7.595, p= .022). However, the differences between groups in overall definite article TLU

were not significant (H (2) =5.808, p=.055) at p ≤ 0.05. Moreover, the control group’s higher

initial accuracy is acknowledged as a limitation. In fact, the control group’s TLU remained

only marginally below those of other groups in Task 2. Caution must therefore be shown in

overinterpreting the progress made by the taught and corrected group.

Furthermore, it is also conceded that the impact of other variables, on the results reported here,

must be considered. The motivation with which students approached the tasks; the extent to

which they proofread their homework; the distractions of external factors; or even simple

tiredness could all have affected improvement (and lower accuracy in the control group).

Across the programme, perhaps the most likely extra unintended help students received before

Task 2 was the student exposure to a case study which had some examples of marketing (the

topic students addressed in the second prompt). On the one hand, the observed increases in

164

TLU scores in Task 2 could therefore have results from the fact that students were primed to

better understand the colligations and grammatical features of topic lexis. On the one hand, the

Figure 49: Progress by three groups (Mdn TLU of Ø) n=24

control group also studied this case study, yet did not demonstrate increased TLU scores. The

extent to which the control group teacher exploited this case study (as the researcher with the

taught class did) is not known. The likely combined effect of these multiple factors will be

further discussed in Chapter 6.

5.7.3 Analysis of differences between taught group and experimental group

While there were significant improvements in definite article TLU and Ø article TLU, no

significant differences were found between these improvements in the taught and error

corrected groups between Task 1 and Task 2 (P ≤ 0.05). A Mann-Whitney U Test

demonstrated that the Mdn 0.11 improvement in TLU of all definite articles combined,

165

observed in the error-corrected group, was not statistically significantly different to the Mdn

0.06 improvement in the taught group (U=36, p=.721). Equally, the 0.08 increase in target like

use of TLU of Ø in the control group between Task 1 and Task 2 was not significantly different

to the 0.07 increase observed in the taught group’s TLU (U= 42, p= .328).

5.7.4 Further analysis of definite article use

Further analysis shows that improvements in the TLU of Type 2 definite articles and all zero

articles could be accounted for by the lower counts of overspecification of definite articles in

Task 2. Table 17 compares the issue of omission and overspecification of the definite article

(Types 1, 2 and 5 combined) by presenting the median values of each group in all three tasks.

Table 17: Omission and Oversuppliance of definite articles (all types, Mdn coefficients*)

Task 1

Task 2

Task 3

Omission

Overspecification

Omission

Overspecification

Omission

Overspecification

Experimental

taught group

0.114

0.111

0.091

0.036

0.087

0.105

Error

correction

group

0.073

0.143

0.069

0.000

0.045

0.092

Control

group

0.045

0.049

0.099

0.081

*To adjust for response length and obligatory uses, the figures are provided by the following

coefficients:

Omission of THE coefficient = Omission in all contexts/accurate use in all contexts

Overspecification THE coefficient = oversupply in all contexts/accurate use + oversupply all contexts

166

The information from Table 17 is further visualised in Figure 50 and Figure 51. It can be clearly

seen from these bar charts that omissions of definite article persist in Task 2, but the taught

group (as a ratio of obligatory contexts) is oversupplying the definite article to a much lesser

extent in Task 2, while the median overspecification value of the error-corrected group was 0.

Figure 50: Bar chart – omission and oversuppliance of definite articles in Task 1

Figure 51: Bar chart – omission and oversuppliance of definite articles in Task 2

0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

0.160

Experimental taught group Error correction group Control group

Essay 1 Definite article (all types) omission and

overspecification (coefficients) medians per

group

Omission Oversuppliance

0.000

0.020

0.040

0.060

0.080

0.100

0.120

Experimental taught group Error correction group Control group

Essay 2 Definite article (all types) omission and

overspecification (coeffecients) medians per

group

Omission Oversuppliance

167

5.7.5 Article misuse errors

There were only 14 article misuse errors (a for an, a/an for the, the for a/an) in total in the

whole corpus. Indeed there was only one ‘a for an’ error. This result is in contrast with Díez-

Bedmar & Papp’s (2008) findings but confirms Master’s (1995) report that Chinese students

at this level make errors of omission (of any article) or overspecification of definite article, as

opposed to confusing the article (i.e. correctly judging nouns to need an ‘a/an/the’ specification

but choosing the wrong specifying article). This very small number of misuse errors is shown

in Table 18, but is ignored in the remaining discussion.

Table 18: Misuse of article errors

a/an confusion

a/an for the

the for a/an

Task 1

Task 2

Task 3

5.7.6 Accuracy at the 14-week stage

Unfortunately, in the taught group most of the improvements observed between Task 1 and

Task 2 were not sustained at the last data collection point. For example, with regard to Type 2

definite article use at the 14-week stage, six weeks after extra lessons on article use and article

error correction had ceased, the taught group’s TLU was in fact marginally lower (Mdn 0.81)

than it had been in Task 1 (Mdn 0.82) as illustrated in Figure 48 (p.173). Moreover, although

the error-corrected group’s Mdn TLU in Type 2 definite articles remained higher than it had

in Task 2, a Wilcoxon ranked test no longer showed a significant improvement (Z=19, p=.889).

168

TLUs of overall Ø articles maintained a fairly similar level, but a Wilcoxon Signed Rank test

suggested the improvement was no longer statistically significant (Z=28, p=.161). Even

though the TLU of a/an overall remained higher at the 14 week stage (Mdn 0.82) than when

students had began in week 1 (Mdn 0.66), this improvement was no longer significant (Z=26,

p= .263).

In the error-corrected group, one article category (a/an articles overall) showed a significantly

improved Mdn TLU (0.70) in week 14 to its equivalent pre-treatment TLU (Mdn=0.50) in

week 1 (Z=34, p=.025). However, this was the only article showing significant improvement.

As was presented in Table 15 (p.160), no individual Bickerton article context showed

significant Target-like accuracy improvement in week 14. For example, the Mdn TLU in

definite articles (0.88) in week 14 was not shown by the Wicoxon Signed Rank test to be

significantly higher than it had been in week 1 (Mdn= 0.76) TLU (Z=20, p=.779). Equally,

this inferential test suggested that TLU of overall Ø articles in week 14 (Mdn=0.91) was not

significantly different from pre-correction levels (Mdn=80) (Z=24, p=.401). The absence of

control group data for Task 3 unfortunately makes it impossible to evaluate the effect of the

task prompt on this disappointing drop in performance.

5.7.7 End-of-study anonymous survey

Twenty-three of the Twenty-four students in this study completed the online end-of-study

anonymous survey in week 15. This section reports on the most salient student responses. Full

responses to all questions can be found in Appendix 6.

With a view to gauging students’ reaction to written error correction, Question 3 asked

participant respondents:

169

Your teacher marks your writing homework and points out an error in an article, e.g. (Art. –

you forgot the definite article).

How do you feel? (which one emotion below best described your reaction?

As visualised in Figure 52, the majority of the respondents (20 out of 23) reported a positive

or very positive response to receiving error corrections. Two of the respondents expressed

negative feelings and one was not sure.

Figure 52: Participant responses to Survey Question 3 (n=23)

In order to investigate students’ affective reaction to being corrected orally in class time,

Question 4 asked:

In an academic writing class a teacher notices an error that you make with your article (a/an/the)

and tells you. While other students are reading/writing he quietly explains a rule to you (about

article use).

How do you feel? (which one emotion below best described your reaction?

Negative (I dislike receiving feedback like this)

Not sure

Positive ( I like receiving feedback like this)

Very positive ( I really like receiving feedback like this)

170

As shown in Figure 53, while 18 of the 23 participant respondents felt positive or very positive

about such correction, three reported they were ‘not sure’ and two disliked receiving such oral

feedback in class.

Figure 53: Participant responses to Survey Question 4 (n=23)

Although the negative affective responses to error correction (written or verbal) were low, there

is a possibility that students are under reporting their dislike of error correction or a focus on

the form of article. It is important (Wang, 2001; Shi, 2006; Leedham, 2015) to avoid the

cultural stereotypes that other authors have made in reference to Chinese students’ respect for

the teacher (Cotazzi & Jin, 1996; Wu, 2009). However, international students (regardless of

nationality) can feel uncomfortable criticising a researcher’s methods when that person is a

senior teacher on their programme, so the less positive reaction of five students could represent

a slightly larger opposition. Two responses in the open question at the end of the survey give

some insight into the students’ frustration with the topic (all five responses are reported in

Appendix 6).

Negative (I dislike receiving feedback like this)

Not sure

Positive ( I like receiving feedback like this)

Very positive ( I really like receiving feedback like this)

171

Anonymous student 1 (open question response):

I have learned about the article use since I was in a primary school. I can remember

some general rules for using article. It is helpful but not as much as I expect. I still want

to study more about this topic because it is very important for learning English.

Anonymous student 2 (open question response):

I like the correction from teacher, however, when I make some stupid error, I feel a

little bit embarrassed and lesser confidence in English.

5.7.8 Comparison with the literature and critical evaluation of results

In sum, this study showed that the taught group made significant improvements in their

accuracy of use of Type 1 generic a/an and Ø articles, Type 2 definite articles, and Type 3 Ø

articles in the period in which they received extra teaching/article correction. The use of the

error-corrected and control group enabled the study to show that students facing the same

prompt without any such extra consciousness raising activities (in the control group) made

significantly more errors in many article contexts. This evidence supports the hypothesis that

teaching can make a difference in students’ accuracy (Master, 1990; Master, 1995; Master,

1997; Bitchener & Knoch, 2008), at least in the short term. However, the finding that this

progress was not sustained significantly (in all but a/an articles in the error-corrected group)

six weeks after the extra teaching had ceased confirms Master’s (1997) conclusion that

students’ improvement at this level was mostly sustained only as long as their attention was

focussed explicitly on article use. Their improvement could thus be interpreted as a result of

the students’ extra attention to proofing their work for article errors that they were more

conscious of due to the teaching they had received.

172

Similarly, in the error corrected group, the finding that error correction (by simply underlining

the error) may have a significant impact in the short term supports the arguments, to some

extent, made by Ferris (1999) and Chandler (2003) about its role in writing classes and that it

is most effective when focussed on single language areas (Bitchener, Young & Cameron,

2005; Ferris & Kurzer, 2019). However, most significant improvements were lost six weeks

after the corrections were focussed so intensely on article accuracy, so it is not argued here that

the error-correction helped students across all article contexts in a sustained way. Moreover,

given lack of long-term impact, the negative affective impact on several students and the

significant time spent on such error correction, a clear argument can be made against such

‘scatter gun’ policies of correction.

5.8 Conclusion

The error-tagging methodology (using the adaptation of the Bickerton Framework for article

use) led to reliable results, as shown by the refined inter-rating reliability tagging. In contrast

to the initial issues encountered in the preliminary study, when the noun phrases were presented

to the annotation checkers in decontextualised forms, this text-based method showed that

properly trained researchers could apply the framework with greater reliability. To achieve

this consistency, an extremely thorough standardisation was required (particularly in

judgements about definite article redundancy/oversuppliance). Some borderline cases in which

some raters disagreed were presented, but these differences in opinion were not as frequent as

those in the preliminary study. While intensive training and the standardization process

achieved satisfactory reliability in this study, there could be difficulty in comparing studies that

examine the complex issues of register appropriacy with relation to article use if the same

polices and standardization was not repeated.

173

Several limitations to this study, not least the small sample sizes, mean that caution must be

shown in interpreting these findings. Furthermore, participation was limited to IELTS 5.5 and

6.0 level, slightly below the average 6.0 – 6.5 level of most students across the UK sector.

While efforts were made to control the teaching input of all three classes, outside laboratory

conditions it is acknowledged that certain variables cannot be scientifically controlled. Indeed,

in the following chapter some insights from student interviews will show the near impossibility

of designing three prompts on topics with equally familiar topic vocabulary for all students

(and this knowledge of the lexis will affect knowledge of countability and thus article

accuracy). Since these data were collected, several studies have emphasised the importance of

task prompt and the effect of topic and task and this may have been underestimated during the

design of this study and this variable has a huge effect on the conclusions that can be drawn

(Díez-Bedmar, 2015; Crosthwaite, 2017). Future studies need to find a way of presenting the

same topic to the students, perhaps requiring different sections of an extended piece of work to

be submitted. A further limitation is that there appears to have been a larger discrepancy

between the proficiency levels in the control group and other groups than had been initially

estimated, as shown in the initial control group’s higher TLU scores in Task 1. Comparisons

between the progress of the three groups must therefore be treated with a degree of caution.

Another limitation is the missed opportunity to reflect new modes of error correction and

feedback (Hyland & Hyland, 2006), since the experimental and control groups received

traditional paper-based correction methods rather than the latest innovative contemporary

feedback processes afforded by technology.

Notwithstanding the limitations, the use of a control group for Task 1 nevertheless suggests

that the improvement in accuracy shown by both the taught group and the error corrected group

may not have been due to this change of topic alone. At least in the time the students’

174

consciousness of article use was raised and in this limited proficiency range, both groups

showed statistically significant progress, particularly in Type 2 definite articles while a control

group studying the same topic and writing the same essay made more errors (relative to their

own Task 1 results) without these interventions. Future studies with larger sample sizes and

control group responses for both tasks are now required to make firm conclusions about the

effect of teaching/error correction, but this study offers some tentative evidence for the

argument that either of these interventions may help students improve their accuracy in the

short term.

Perhaps the most important finding in this study was that both teaching of article use and error

corrections had more effect on reducing overspecification than article omission. As redundant

definite articles have been shown to contribute to an inappropriate conversational tone and

since register awareness is a key objective of EAP courses, this insight requires further

investigation and future research is needed to more qualitatively investigate this effect. The

next chapter presents evidence from several student interviews, but a larger scale study is

needed to look into one possible explanation that students are simply careless with many article

redundancies while their omissions in article use may be the type of errors they make in true

misunderstanding of the noun’s countability or use in its lexical context.

The findings of this study have several implications for EAP teachers if the findings of such a

small scale study involving IELTS 5.5/6.0 levels were representative of the wider UK

population of Chinese L1 lower English L2 learners. The more practical recommendations for

the teaching of article use in EAP will be presented in the Chapter 8 conclusion in light of the

findings that will be reported in Chapter 6 and 7, but it can be immediately seen that the ‘one

off’ teaching session recapping the grammar of article rules may not necessarily have a

sustained effect on Chinese students’ written accuracy of articles. Equally, the occasional mass

175

correction of all of a students’ article errors can improve students’ consciousness and have a

short- term effect, but this extra attention to article accuracy may only last as long as students

have such regular corrections. Meanwhile, it should be remembered that a small number of

students expressed negative views towards such correction.

This study does not support the pessimistic assumption that teaching or error correction will

have no impact as, notwithstanding the fact that the improved accuracy only lasts as long as

the teaching/corrections, the study did observe significant improvement in accurate use in

certain contexts. Students generally embark upon presessional programmes knowing the

general rules of article use, so this was most likely a result of the students’ added consciousness

of articles during their essay proofing rather than learning anything new about articles during

teaching. On the other hand, as any consciousness raising activities are likely to have an effect

on short-term accuracy (which Chapter 4 showed as an end in itself for some teachers as they

prepare students for extended written assessments), this study has shown that a teaching

approach could be valid, however limited. As to which method of focussing attention is

preferable, there is a clear case for teaching and against overcorrection of articles because, even

if both are equally effective in the short term, teaching is less likely to cause a negative reaction

among learners.

As shall be developed in the conclusion, given the possible impact that article use is suggested

here to have had on the register (with many redundant errors argued to cause an informal tone

in the writing) this effect of inaccuracy may merit attention in EAP programmes. In parallel

with the positive effect that teaching/correction had on overspecification, a strong case shall be

made in the thesis recommendations for focussing in particular on definite articles and their

appropriacy for cohesive purposes but inappropriate effect when making generalisations or

176

discussing hypothetical or idealised situations in Type 4 contexts in their essays (when bare

noun plurals are usually preferred).

Before any conclusions or recommendations are made, there was a need to conduct research

with a larger population with a wider range of levels (as will be presented in Chapter Seven),

particularly since this chapter has only focussed on IELTS level 5.5 and 6.0 students. Before

extending the focus, however, the following chapter analyses these Chinese learners’ article

errors more deeply and compares the same learners’ use of articles in essays to their accuracy

in an alternative genre of academic writing. In addition, a comparison with responses from L1

tutors provides insights to underuse and overuse issues.

177

6 CONTRASTIVE ANALYSIS OF CORPORA

This chapter presents the further analysis of the data collected for Chapter 5 (which focussed

on the dynamic effects of article consciousness raising) together with further L1 data, data from

lower proficiency Chinese students, and data from a different genre of writing that some of the

students submitted (a case study) in order to gain a deeper understanding of the contexts in

which Chinese learners make article errors. In addition, through further qualitative and corpus-

based (deductive and inductive) methodologies, the thesis turns its attention to the possible

causes of errors made by Chinese L1 students in particular.

6.1 Study overview

Two of the four research questions as originally stated in the Literature Review are addressed

in this study:

Research Question 2: What are the linguistic contexts of obligatory article use in which

Chinese L1 students at Upper-Intermediate levels of English L2 are most likely to make

errors?

Research Question 3: To what extent does the Chinese students’ L1 background affect

(1) TLU and (2) their types of article error?

Research Question 2 refers to ‘contexts’ firstly in terms of the Bickerton article use Types as

presented in Section 2.2.1. By replicating the data analysis approach of recent studies (Díez-

Bedmar & Papp, 2008; Crosthwaite, 2016a) into Chinese students’ article errors, comparisons

of ‘Target-Like Use’ (TLU) can be made. This analysis will thus examine whether the

University of Birmingham’s Chinese L1 15-week B2 level students continue to demonstrate

the many ‘non-native features’ reported by Díez-Bedmar & Papp (2008) or whether, as found

178

by Crosthwaite (2016a: 94), they are generally ‘not struggling with article use in the majority

of article contexts’.

After analysing ‘context’ in terms of the Bickerton Framework, replicating the two

investigations mentioned above (Díez-Bedmar & Papp, 2008; Crosthwaite, 2016a), the study

then also turns to a more data-driven analysis through concordance tools. As discussed in

Chapter 2, academic writing presents a steep learning curve for L2 novice academic writers

not least due to the very frequent use of densely packed nominalisations in this register

(Halliday & Martin, 1993, Biber & Gray, 2010; Bennett, 2011). On the basis that novice

academic readers/writers have been shown to struggle with such compression of the noun

phrase (Fang, Schleppegrell & Cox, 2006) the effect of pre-modification of the head noun on

accuracy of article use is also examined.

The likely effect of L1 transfer (Research Question 3) will be addressed through a comparison

of the findings with secondary research findings. By comparing the findings with those of

Díez-Bedmar & Papp (2008) and Crosthwaite (2016a), their hypothesis that Chinese learners

omit articles due to L1 transfer can be tested. As reviewed in Chapter 2, Díez-Bedmar & Papp

(2008) hypothesised a greater omission of articles among article-less L1 students whose L1

language does not grammaticalise definiteness and specificity in the same way. They also

found Chinese learners oversupply articles due to a deficit in pragmatic awareness.

Crosthwaite’s (2016a) more nuanced hypothesis is that Chinese L1 learners show a lack of

positive transfer effect in their inaccuracy with generic article uses, particularly in indefinite

generic contexts as Chinese has no equivalent grammaticalization. Thus, evidence that Chinese

learners had higher accuracy in Type 2 and 3 contexts would support his hypothesis that

Chinese learners do show positive transfer effects, since unlike other article-less languages,

179

Chinese has the potential to show definiteness and indefiniteness with classifiers and

demonstratives.

Following the study presented in Chapter 5, it was also hypothesised that the Chinese students’

tendency to ‘oversupply’ definite articles may create an effect on the perceived register (Lee

& Chen, 2009). This effect could contribute to explaining claims in the literature that novice

L1 Chinese students’ L2 academic English writers can be perceived to be using an overly

informal or even conversational tone in their academic writing (Gilquin & Paquot, 2007;

Mayor, 2006; Lee & Chen 2009; Chen, 2014).

6.2 Methodology

L2 data

Four corpora will be used in this chapter: 1) a B2 level student corpus (n=24), 2) a B1/B2 level

corpus (n=30), 3) a micro case study corpus (n=16) and 4) a micro L1 corpus (n=5). As this

research project mainly pertains to B2 level learners most findings in this chapter relate to

Corpus 1 (three essay tasks written by 24 B2 learners) with the other corpora used for

comparative purposes.

Corpus 1: B2 level corpus.

This corpus contains data recycled from Chapter 5’s longitudinal study (35,716 words). As

suggested in the previous chapter, this data may have been affected by efforts to raise students’

consciousness about article accuracy. On the one hand, given that some of these essays (see

Table 19, p.199) were written while students were asked to pay extra attention to English

articles, it could therefore be unrepresentative of Chinese students’ grammatical accuracy or

at least only representative of Chinese students more consciously trying to proofread their work

180

for article mistakes. On the other hand, as the research is interested in investigating the average

Chinese international student in the middle of the process of writing their first academic texts

(with time to proofread their own work), the errors that such students continue to make –

despite these efforts – are argued here to represent a rich and valid source of data in their own

right.

Corpus 2: B1/B2 level corpus

In order to enable an analysis of the effect of proficiency presented in Section 6.4.5, 18 essays

(Tasks 1, 2 and 3) from lower level B1 students from one of the BME classes were added to

the corpus. The class teacher of these students had volunteered to ask students to take part in

the earlier study, but the lower level of the students and the added tuition given to the students

from several language support tutors during the programme (and the consequent difficulty in

controlling the pedagogic input) led to their disqualification from the investigation of Chapter

5. However, with lower scores in the BME writing placement test (which streamed these

students into a class that was provided with additional language support) their essays from each

of the three writing prompts were used to examine the effect of general writing proficiency on

article accuracy.

As can be seen from Table 19, the six lower level students had all achieved a maximum score

of IELTS 5.0 (overall) in the last two years. All had been studying on the university’s longer

presessional course prior to the time of the study (four students for five weeks and two for three

months). Their results on entry writing test (ranging from 48-51) showed that these six lower

level students continued to write at a borderline B1/B2 level. In sum Corpus 2 participants, in

addition to these six students at B1 level, included 14 students at the IELTS 5.5 low/middle

181

end of the upper=intermediate level and 10 students with stronger Upper-Intermediate IELTS

6.0 scores.

Table 19: Participant information (including borderline B1/B2 students).

IELTS 5.00

(n=6)

IELTS 5.5

(n-14)

IELTS 6

(n=10)

Total/Summary

Age

21-23

23-26

22-24

Median: 23

Male

Female

Male 10

Female 20

Chinese

Taiwan

21 Chinese

9 Taiwan

Entrance writing

test score

Range: 48-51

Median: 50

Median AD= 1

Range: 53 -59

Median: 5.5

Median AD= 1

Range: 56-61

Median: 57

Median AD= 2

Mean:

56.04

SD: 4.26

Length in the UK

or English L1

countries at entry

Four students

under two months,

two students under

five months

One student under

three months, one

student under two

months, 12

students under one

month.

All students under

one month

No students over

five months in the

Corpus 3: Micro case-study corpus

In order to investigate the effect of genre, a micro-corpus of case studies produced by 13 of

the students from Chapter 5 was also compiled (14,603 words). Details about the case study

prompt can be seen in Table 20 (p.183) where the case study is labelled ‘Task 4’ to avoid

confusion with previous labels (although this case study was actually submitted in week 7,

182

before Tasks 2 and 3). This case study submission was the first draft of an assessed piece of

work (students turned in a final submission after feedback in week 12).

While the essays reflected the Presessional’s approach to building the academic writing skills

that students would need for the literature reviews of their Masters’ level dissertations, the case

study assessment on the Presessional more authentically reflected the assignment tasks that

students would face in the Business School (i.e. report structured assignments based on a case

study text). Unlike the essay prompt tasks, students were given a background case study text

and asked to use examples and make references mainly to this text in their response (only

synthesising theories from the literature where necessary). The main difference in this case

study prompt, however, was the instruction to provide recommendations. According to

multidimensional genre analysis (Conrad & Biber, 2002; Gardner, 2008) case studies can be

differentiated from essays due to their purpose of recommending a course of action. In terms

of Gardner’s (2008) cline of persuasion, the case study represent a register in dimension 4 quite

distinct from more argumentative forms of academic writing.

Summary of Prompts for all corpora

For ease of reference, the four prompts are presented in Table 20 overleaf.

6.2.1 Corpus 4: L1 data for contrastive analysis

To facilitate a contrastive analysis (and identification of ‘overuse’ or ‘underuse’ of articles),

five L1 English speaking tutors agreed to write responses for Tasks 1 and 3. All five ‘expert

writers’ held Masters/DELTA teaching level qualifications and had taught on the BME for at

183

Table 20: Four Task prompts

Rubric

Texts

Words

total

Minimum and

Maximum

lengths and

Median word

length

Task 1

Essay

(week 1)

Does fear motivate employees to work harder?

Illustrate your ideas with reference to theories of

motivation studied on the course. (350 - 400 words)

11147

Minimum: 302

Maximum:440

Mean: 371.6

words

Task 2

Essay

(week 8)

An effective marketing strategy is crucial to the

success of a business. Critically assess the value of

segmenting, targeting and positioning as a

marketing tool. Illustrate your answer with suitable

examples. (400-600 words)

*option for the three accountancy students:

Summarise the responsibilities of accountants and

discuss how they can contribute to efficient

information flow both within and outside business

organizations.

14306

Minimum: 443

Maximum:605

Mean: 476.9

words

Task 3

Essay

(week 14)

Retailers compete for customer loyalty and retention

in many ways.

Examine the issues, illustrate with examples and

suggest ways in which retailers can promote loyalty

and retention. (400-600 words)

10263

Minimum:403

Maximum:667

Mean: 466.5

words

Task 4

Case Study

(week 7)

Students were given a ‘case study’ about a Scottish

soft fruit grower and instructed to:

1) identify the marketing, human resources and

financial constraints. 2) With reference to the

Product Life Cycle tool, give two alternative methods

of extending their products’ life cycle and 3) make

recommendations about the company’s payment

strategy. (900 words)

14603

Minimum: 853

Maximum 976

Mean: 912.7

words

184

least one summer previously. Three responses were also gathered for Task 2 and three

responses for the case study Task 4, all from tutors with L1 backgrounds and Masters/DELTA

English teaching qualifications.

6.2.2 Limited number interview sessions

Three students participated in post programme one-to-one unstructured interviews designed to

investigate students understanding of why they made article errors. During interviews, students

were presented with a number of correct and incorrect article uses from their Task responses

and asked 1) whether they believed them to be correct and 2) if incorrect, why they had made

the error in their Task response. One-to-one sessions that enabled such qualitative investigation

of error were planned for all participants. However, due to the high pressure/stress students

were under in the high stakes assessments at the end of the course, it was not possible to

schedule these sessions until six weeks after the Presessional had finished. By this point,

participation was lower than expected. Nevertheless, some qualitative findings were gleaned

from these interviews and these will be reported in Sections 6.4.7 to 6.4.15.

6.3 Data analysis

For ease of reference, the Framework adapted from Bickerton (1981), as critically evaluated in

Chapter 3, is reproduced overleaf in Table 21. Procedures for tagging, training two annotation

reviewers, and checking the reliability of annotation were presented in the previous chapter

(Section 5.6). As aforementioned, these procedures led to a Fleiss Kappa inter-rater correlation

of 0.87, showing strong reliability.

185

Table 21: Framework adapted from Bickerton’s (1981) semantic and pragmatic uses

Type

Features

Context

Tag/Form

of Article

Examples

[-SR, +HK]

Generic: The reader knows of

its existence, but not a specific

reference.

1IA a/an

1ZA Ø

1DA The

An elephant never forgets.

Elephants never forget.

The elephant never forgets.

[+SR, +HK]

Writer and reader both know

of the specific referent.

2DA The

Remember to feed the

elephant!

That’s the biggest elephant

I’ve ever seen.

[+SR, -HK]

The writer knows of a specific

referent the reader does not

know.

3IA a/an

3ZA Ø

The local zoo has an elephant.

The local zoo has elephants.

[-SR, -HK]

Neither writer or reader

believe the noun refers to a

specific thing

4IA a/an

4ZA Ø

The zoo does not have an

elephant.

The zoo does not have

elephants.

Fixed

expression

Conventional use (use in

idioms and proper nouns)

5ZA Ø

5IA a/an

5DA The

On the one hand, Nelly the

elephant enjoyed the 1950s.

On the other hand, she was

now having a ball.

Alternative

determiners

Quantifiers or other

alternatives to article

determination

Two elephants.

Another elephant

Examples 1-4 taken from Langendoen (1970), Cziko (1986), Heubner (1983), and Butler (2002). Examples 5

& 6 author’s own examples.

186

6.4 Results and discussion

Before reviewing ‘Target-like Use’ (TLU) data, this section gives an analysis of percentage of

error type findings for the most common type of article error made by the students.

6.4.1 The most frequent errors Chinese students made in short essays (percentages)

Restricting this part of the analysis to the B2 level students (n=24), as this Upper-Intermediate

level is the main focus of this research project, 3.2% of all noun phrases were judged to be

underspecifying the referent, omitting either an obligatory the or a/an article (oversuppliance

of Ø). Overspecification (overuppliance of a/an or the when Ø was judged obligatory) occurred

in 2.1% of all noun phrases. Focusing on the definite article alone, omission was a slightly

more frequent problem (2.2% of noun phrases) than overspecification (1.7% of noun phrases).

In contrast, such oversupply was an extremely rare problem for a/an article use (0.3% of noun

phrases) compared to omission of a/an which occurred in 1.6% of noun phrases. In contrast

with some of the claims made in the literature (Master, 1987; Chang, 2001) misuse of a/an for

the (or vice versa) was highly infrequent (0.1% of noun phrases),

This analysis of percentages allows for comparisons with studies that reported raw data alone

(without ratio measures) and confirms Chuang & Nesi’s (2006) findings that Chinese B2 level

students make both omission and oversuppliance definite article errors, but that omission is a

marginally more frequent error. These results also lend support to the hypothesis of Díez-

Bedmar & Papp (2008) that L1 Chinese learners omit articles as a result of their article-less L1

background and oversupply articles as a result of their deficit in pragmatic awareness.

Table 22 presents the ranked contexts in which students produced errors as a percentage of all

article errors. As can be seen, almost half of all errors in the three essay prompts were judged

to be in Type 2 and 4 contexts. As 95.2% of the 4 GAZA type errors were a result of the

187

overspecification, it can be reported that over 47% of the article errors therefore involved either

the omission or oversupply of definite articles.

Table 22: Proportion of errors by context

Grammatical description

Bickerton Framework

Context

% of Head Noun phrases

Referential definite articles

2GADA

24.89

Non-referential Ø article

4GAZA

23.23

Non-referential a/an article

4GAIA

12.97

Referential a/an article

3GAIA

10.41

Conventional definite article use

5GADA

6.64

Generic definite article use

1GADA

5.13

Indefinite Ø article

3GAZA

4.98

Generic a/an article

1GAIA

4.07

Conventional Ø article use

5GAZA

3.92

Generic Ø article

1GAZA

3.17

Conventional a/an article use

5GAIA

0.60

100%

6.4.2 Contrastive analysis of all uses of articles

Returning to the issue of both correct and incorrect uses of articles, the B2 essay corpus of

Chinese students (n=24) was next compared to the L1 writer corpus. The number of ‘other

determiners’ (6OD annotations) was compared and found to be very similar. L1 writers used a

range of pronouns, quantifiers, possessives and alternative determiners for 16.6% of all tagged

188

noun phrases while the Chinese students chose other such determiners in 15.3% of noun

phrases.

As shown in Table 23, the most striking contrast between the L1 tutor essay responses and the

Chinese students’ responses is in the two groups’ use of Type 3 indefinite/Ø articles.

Particularly in Task 1, Chinese students used Type 3 articles (6.4% of all articles used) far less

frequently than L1 tutors (20.2%). In contrast, Chinese learners more frequently wrote in Type

4 non-referential contexts (54.7% of articles) than the L1 tutors (43.1%). The Chinese learners’

use of Type 3 indefinite/ Ø articles more than doubled in the essay 2 and 3 responses, but it

remained constantly lower than the L1 tutors’ use.

Table 23: L1 writers (n=5) article uses compared to Chinese L1 essays (n=24)

Essay 1

Essay 2

Essay 3

Case study

Chinese

English

Chinese

English

Chinese

English

Chinese

English

n=24

n=5

n=24

n=3

n=24

n=5

n=24

n=3

Type 1

6.2

6.4

11.7

7.4

8.6

5.8

3.8

4.3

Type 2

22.2

20.8

26.7

24.2

21.1

28.6

55.0

51.4

Type 3

6.4

20.2

14.3

22.8

14.3

20.8

12.5

13.2

Type 4

54.7

43.1

35.1

32.1

46.2

36.6

22.8

25.7

Type 5

10.5

9.5

12.2

13.5

9.7

8.3

5.9

5.4

Total

100.0

Closer inspection of the reasons for the divergence in article use seen between Chinese L1 and

English L1 writers showed that the latter were more frequently weaving between

generalizations and specific examples in all parts of their essays. L1 writers were likely to use

189

abstractions and generics in their introductions and conclusions, but far more likely to cite

examples in their introductions. By contrast, Chinese students often introduced their essay with

reference to generic, prototypical companies, but rarely used specific examples from the real

world to develop these ideas. Throughout the body of the essay, L1 writers often used generics

or non-referentials in the topic sentences and ends of paragraphs in the more abstract critical

evaluation of the topic, but were far more likely to invite the reader to consider concrete real

world examples (see Figure 54). In contrast, as shown in Figure 55 below, through their use

of conditional sentences, modality and continued discussion of generalised referents of

‘employees’ and ‘managers’, Chinese students often maintained a non-referential unspecified

level of abstraction throughout the entire paragraph and some maintained this level of

abstraction throughout most of the essay, particularly in Task 1.

[Task 1, Expert writer #1] Indeed, there have been well publicised high awards in courtrooms

in recent years following accusations of worker intimidation.

[Task 1, Expert writer #2] Another issue is that a fear of failure has been shown to lead to

risk averse employees.

Figure 54: Tutor examples

[Task 1, Student #10] On the other hand, motivation by fear may reduce employees’

performance. If the employees are only motivated by fear in a long term, it will be hard to

offer them a sense of job satisfaction and security.

[Task 1, Student # 7] Few can deny that one single motivation theory cannot fit all particular

situations, by which I mean effective managers should focus on various of motivation theories

and give more incentives to employees.

Figure 55: Student examples

Students’ increased use of Type 3 contexts in Tasks 2 and 3, to some extent reflects their

development as novice academic writers. Indeed, the principle objective of the academic

190

writing component on the Presessional programme was to take students from the habits

developed for IELTS writing exams (use of learned generic answers and formulaic expressions

learned in IELTS preparation courses) to a more academic style and the use of evidence to

support all ideas and claims. Another key aspect of academic writing taught in all classes

through the academic writing component was the ‘general to specific’ textual pattern of

paragraphs and ‘general-specific-general’ shape in most EAP essay structures. Thus, the

increased use of the specific referents between Task 1 and later Tasks may reflect the students’

growing understanding of this aspect of academic writing. On the assumption that the L1 tutors

were appropriate models of academic writing, the outstanding gap that remained between the

two groups shows that the students were still failing to provide examples and specific

references for the abstractions in their generalised claims.

In regards to the addition of case study prompts, it is interesting to note the very similar patterns

of article type use by L1 tutors and the Chinese students, as was summarised in Table 23

(p.188). Despite being a new genre of academic writing for the students, it appears that the

students had little difficulty in applying the same discourse patterns as tutors: i.e. using

examples from the given case study text to support their answers. When asked to keep their

focus on one specific company, the students had little difficulty in keeping their references to

concrete things in the real world. Two factors therefore are likely to explain the students’

greater challenges in other genres of more discursive essays: 1) the habits learned in IELTS

preparation courses (which reportedly often encourage formulaic answers) and 2) the need in

more academic writing to more dexterously weave between the specific examples and abstract

claims to critically argue a position.

191

6.4.3 Order of accuracy

The medians and their distributions for accuracy of use of each article type are shown in Figure

56. Chinese students showed the highest TLU of Ø (0.89), followed by definite article

accuracy (0.85) and then a/an (0.81). These differences, while marginal, were shown to be

significant (P ≤ 0.05) by a related-sampled Friedman’s Two Way Analysis of Variance by

Ranks ( X² (2) =29.747, p=.000). Applying the Bonferroni correction for multiple tests, the

differences remained significant between a/an and the TLU (X² (2) =2.598, p=.028), and

between a/an and Ø TLU ( X² (2) =-4.980, p=.000), but not statistically significant between

the and Ø (X² (2) = -2.382, p=.052). These findings agree with Díez-Bedmar & Papp’s (2008)

finding that Chinese students are most accurate with Ø articles, but contrast with their finding

that the definite article was the most challenging article for students of this L1.

Figure 56: Median TLU scores in the three article types (3 prompts, n =24).

192

6.4.4 TLU of all types

As shown in Table 24, the data from the three essays shows that the B2 Chinese students

showed greatest target-like accuracy in 5ZA, 4ZA, 3IA and 2DA contexts. These four contexts

were among the five most frequent contexts for error, but when obligatory use was taken into

account (and the higher frequency of such contexts in academic essays), it can be seen that

Chinese students are generally more confident with definite articles and Ø than the raw error

frequencies suggested. However, the 4IA indefinite article, which was highly frequent in

essays, was one of the more inaccurate contexts for the Chinese learners (0.71 TLU). It should

be noted that the TLU for 5IA is greatly affected by its rarity and is therefore misleading – if

zero counts are removed from the analysis the students showed almost perfect accuracy.

Table 24: Overall TLU (all essays) in all contexts (n=24)

IDA 11A 1ZA 2DA 3IA 3ZA 4IA 4ZA 5DA 5IA 5ZA

Median 0.67 0.75 0.86 0.88 0.89 0.80 0.71 0.95 0.80 0.00 0.97

MAD 0.33 0.29 0.11 0.06 0.12 0.10 0.21 0.05 0.20 0.00 0.03

In sum, the essay data from this study would appear to suggest that B2 Chinese students are

most challenged by the following article contexts in order of difficulty:

IDA<4IA<1IA<5DA=3ZA<1ZA<2DA<3IA<4ZA<5ZA.

MAD = Median Absolute Deviation

193

6.4.5 The effect of proficiency on accuracy

By adding the six B1 students to B2 data to make Corpus 2, the effect of general writing could

be investigated with a slightly broader range of English proficiency levels. However, efforts to

find significant relationships between general writing proficiency and accuracy of article use

in the eleven contexts annotated met with only limited success. Students’ (n=30) individual

combined TLU coefficient for all three essays (in all eleven contexts) were analysed in relation

to their writing proficiency (as measured by the Presessional entrance writing test). No

significant relationship was found between the students’ general writing (as measured either

by writing test or by their IELTS writing bands) and their combined TLU in the three articles

(the/a/ Ø). Equally, no correlation was found between general proficiency level (measured by

either IELTS test band or writing entrance test score) and the students’

omissions/overspecification of either definite or indefinite articles in any of the contexts.

A Spearman’s rho test found a significant relationship (P ≤ 0.05) between all students’ (n=30)

placement writing score and their Target-Like Use with Type 1 definite articles (r

=.282,

p=.013). The same test (P ≤ 0.05) also found a significant relationship between all students’

(n=30) placement writing score and their Target-Like Use of Type 3 (3IA) indefinite articles

=.228, p=.046). However, no significant relationship was found in this data between writing

proficiency and accuracy in 1IA, 1ZA, 2DA, 3ZA, 4IA, 5DA, 5IA and 5ZA contexts of article

use. This finding of no significant effect from proficiency level on the overall article accuracy

is surprising, since Chapter 2 showed a widely reported positive relationship in the literature

(Crosthwaite, 2016a; Murakami & Alexopoulou, 2016). Here, the effect is somewhat

accounted for by the very small sample of B1 students in the study and the large number of

students with an IELTS 5.5 band score.

194

6.4.6 Comparison of student accuracy in essays and case studies

Underspecification was again a greater problem than overspecification in the case study

responses. In terms of 2DA context definite articles, omission was a far more frequent issue

(2.6% of all noun phrases) than oversupply (0.7%). Equally, there was only one single

oversuppliance of indefinite article, while omission was the only error related to a/an articles

(1.3% of all noun phrases). This reduction in overspecification of definite articles is partly due

to the register differences of the genre. A case study response naturally entails inter-textual

cross references to the actors/organisations described in the case study prompt so the uses of

specific references to ‘the company’, ‘the workers’, the competition’ was more permissible

than an essay in which the focus is upon generalisations rather than specific cases.

This improvement in accuracy was sustained in the figures when ratios were used to factor in

the obligatory contexts. In fact, the Target-Like Use (TLU) of articles among these Chinese

students using the Bickerton Framework outlined in Section 2.2, as shown in Table 25 was

Table 25: TLU is essay prompts compared to case study prompt

Tasks 1-3 (3 essay prompts,n=24)

IDA 11A 1ZA 2DA 3IA 3ZA 4IA 4ZA 5DA 5IA 5ZA

Median 0.67 0.75 0.86 0.88 0.89 0.80 0.71 0.95 0.80 0.00 0.97

AD 0.33 0.29 0.11 0.06 0.12 0.10 0.21 0.05 0.20 0.00 0.03

Task 4 (Case study prompt, n=16)

IDA 11A 1ZA 2DA 3IA 3ZA 4IA 4ZA 5DA 5IA 5ZA

Median 0.00 0.00 1.00 0.92 0.91 0.89 0.78 0.95 0.83 1.00 0.86

AD 0.00 0.00 0.00 0.02 0.09 0.89 0.06 0.03 0.17 0.00 0.14

195

generally higher in the case studies than the essay responses. Ignoring use of generic articles

(students very rarely attempted their use in the case studies) the students showed greater

accuracy, for example, in Type 2DA (0.92), Type 3IA (0.91) and 4IA (0.89).

It is very likely that both the genre and the prompt Task were responsible for this greater

accuracy. In case study Tasks, students were required to only use the one set case study text

provided to give examples to support their answers. In effect, students were mostly

paraphrasing from a given text and synthesising these examples with their own analysis. In

addition to the greater allowance of specific reference, the students could see many examples

of the new vocabulary and its colligations with article use throughout the text that they were

referring to when writing their answers, so their confidence of article use with the new

vocabulary is therefore likely to have been enhanced. Although this makes the comparison

with essay tasks potentially problematic, the phrases were not removed from the analysis

because this use of case study texts authentically represented Business School assessments.

Thus, while this to some extent explains the students’ greater accuracy in this task type, it was

considered more valid to include the uses of prompt vocabulary (‘the company’, ‘the workers’)

in the analysis because the summary of such texts (retaining key words) is a normal part of

academic writing for business school students.

Given that students would more usually be writing responses to case studies of this kind (with

set texts to contextualise the genre and familiarise students with the new topic lexis) once

entering the Business School, it is possible that these higher accuracy rates are more reflective

of the students’ true abilities in the academic writing of their chosen discipline. This

196

conclusion would seem to support Crosthwaite’s (2016a) conclusion that students at B2 level

show little inaccuracy in article use outside generic contexts.

The remainder of this chapter now focusses on the individual article use contexts in order of

the frequency at which each error was found in Corpus 1. The analysis is confined to essay

responses at B2 level for comparative purposes to the studies replicated. In each section, the

L1 Chinese students’ errors in the three essays (n=24), are put in context of other studies and

the qualitative research undertaken in this project based on insights from student interviews.

In each of the sections below, a brief summary of the quantitative analysis will be followed by

some examples from the students’ three essays together with qualitative insights gained from

their interviews and the inductive analyses to which the student interviews led. In each of the

figures below, correct obligatory use examples are provided in blue text and examples of

incorrect use are given in red.

6.4.7 Context 2DA

The most frequent article error was omission on the in Type 2 contexts (24.9% of all article

errors), replicating the results of previous studies (Chuang & Nesi, 2006; Díez-Bedmar & Papp,

2008). Examples of correct and incorrect use are shown in Figure 57. The high frequency of

the error reflects its frequency in academic writing. In fact, among both students and L1 English

tutors the Type 2DA definite article was the second most frequent article used in essays. As a

ratio that takes account of high frequency obligatory use and overspecification, TLU of 2DA

definite articles was more ‘mid-range’ (0.88). This TLU contrasts with the slightly lower

accuracy reported by Díez-Bedmar & Papp (2008), but contrasts with Crosthwaite’s (2016a)

finding that Chinese students had almost mastered 2DA articles by B2 level.

197

2DA

It is clear from

the

above analysis that as a market tool,

segmenting, targeting and positioning all

do not have absolute advantages or

absolute disadvantages.

2DA

For instance,

the

main colours of Sainsbury’s stores are

orange and red which create the warm and

relax atmosphere for customers when they

are shopping in the store,

Error

example

The other one is the build-up method,

which only subdivides the market by

a single factor, such as gender in

cosmetics market and age in ^dress

market.

Error 2

^Switch ratio was developed by

Crouchely. It calculates

number of times a customer switches

from a retail store to another during a long

term (Kirkup, 2002).

Figure 57: 2GADA errors compared to 2DA correct use

At interview, students were presented with a mixture of such correct and incorrect uses (see

Figure 57) and asked to reflect on accuracy and then explain errors made. The student who

made Error 1 spent over one minute considering the example (with all articles removed) and,

as shown in the transcript below, showed great hesitation before finally agreeing the need for

a definite article:

198

Hmmm…can I take my time? This is stressful now, you take me back to my high school…

{student reads and rereads sentence for 50 seconds]…I’m not sure but I would prefer to put

nothing there. Let me read it again…[10 second pause]… no I’m not sure. [Teacher gives

student correction]…but …oh…there should be the here? It’s confusing. It confuses me

because cosmetics is plural…ah …ok I understand it is singular now.

[Student 1, six weeks after Presessional)

The above example shows the effect a pre-qualifying noun (cosmetics) had in obscuring the

student’s perception of the head noun (market) that they should have been determining. Indeed,

a recurring feature seemed to be that a pre-modification of any head noun (with adverb,

adjective or noun) confused the student’s grammatical understanding of the head noun. Thus

many omissions, as also shown in the example ‘^dress market’ occurred with pre-modified

head nouns.

On further investigation if the corpus with the Antconc concordance tool, an analysis of the

word ‘market’ used as head noun (n=83) showed that the was omitted in 13 of the contexts in

which the word was qualified (e.g. ‘^consumer market; ^young child toy market, ^adult market;

^new business market) representing nearly three fifths of the 22 qualified ‘market’ noun

phrases provided. Meanwhile, used singularly (without pre-modification) the lexis resulted in

only 6 omissions (under one tenth of the 61 uses of ‘market’ without any pre-modification).

Further analysis of all the 2GADA errors in the corpus, as shown in Table 26, showed that the

overwhelming majority of definite article omissions in 2DA contexts (69.4%) occurred when

the head noun was pre-modified, while only 18.5% of simple nouns were underspecified. The

effect of post-modified ‘of’ noun phrases, in contrast, appeared to have the opposite effect,

increasing the accuracy rate. It is possible that the ‘…of…’ pattern is a useful cue that students

can use to consider the use of definite article (which frequently colligates with ‘of’ cataphoric

reference).

199

Table 26: The effect of pre-modification

2DA (n=1780)

2GADA (n=159)

Head noun with no pre-

modification /not in cataphoric

‘of’ phrase.

37.3%

18.5%

Head noun ‘of’ phrase

(cataphoric reference)

28.4%

12.1%

Head noun of pre-modified

noun phrase (qualified by

adverbial, adjective or noun

classifier)

34.4%

69.4%

6.4.8 Context 4ZA

The second most frequent context for article error (23.2% of all article errors), occurred through

the overspecification of articles in Type 4 Ø article contexts, as exemplified by Error 3 and

Error 4 in Figure 58 below. With full sight of the context of these errors, the researcher and

two L1 English checkers agreed that redundant the was inappropriate because the writers were

judged to be attempting to construe non-specific concepts (‘turnover’ and ‘managers’) in a

general sense (in the conclusions to their essays). Owing to the fact that Ø morpheme is the

most frequent choice of article (and given than Type 4 contexts were the most frequent for both

students and L1 tutors), Ø TLU (0.95) is relatively high, despite these errors. 93% of these

147 errors involved the overspecification of definite article. In terms of frequency, this result

suggests that oversupply of definite article is the second most common error among Chinese

200

4ZA

people are worried about something, they

will not give their best.

4ZA

Aural cues such as music also could build

up an atmosphere which would influence

the consuming behaviours of customers.

Error

… may mean the sustainable

development of the companies is

hard to achieve. For example,

fear of redundancy may lead to

the

high staff turnover, which might destroy

the structure of companies’ human

resources.

Error

it works actually. Furthermore, since

the economic crisis last year, everyone

is afraid of losing their job. So

the

managers may consider that high pressure

is a helpful way to lead employees to

achieve the company’s goal .

Figure 58: 4GAZA errors compared to 4ZA correct use

L1 students at B2 level, replicating the conclusions of both Chuang & Nesi (2006) and Lee &

Chen (2009).

Interviews with the three students confirmed that overspecification of the errors were usually

genuine errors, rather than mistakes. Students often reported feeling unsure of whether an

article was required but also unsure about why the use of the definite article was problematic.

The three interviewed students admitted to sometimes adding definite articles when unsure,

almost as if considering the ‘odds’ in gambling statistics and deciding that adding the was

always the less risky option.

Asked to review Error 3, the student who made the error did not understand the explanation of

redundancy and informal tone created by the (given by the researcher) when prompted:

201

OK I don’t know whether we can say the, a or nothing here with turnover. Is it countable?

[researcher explains variation…on prompt explains redundancy of the in this context]. It’s

wrong? [5 second pause]… But even uncountable, I have a choice?... if I put the there, I mean

certain people. Is it not possible?…[5 second pause]…Actually, I often put the there when I’m

not sure.

[Student 2, six weeks after Presessional)

There was also evidence of students very logically overgeneralising aspects of English that

they had learned. For example, one student (see Error 4) could not see why ‘the management’

created an informal tone to the writing when they had seen examples of creation of nouns from

adjectives (the elderly, the poor, the young) which was not argued to have this effect. Indeed,

the researcher could not answer beyond discussing the ‘feeling’ of an informal tone created by

‘the management’ that was not created by ‘the elderly’:

Yes I see now, I feel less comfortable now with ‘the’ here. I see there is no article here

[researcher confirms]. I think I wanted to talk about the managers as a group. You know I

hate articles. I can say ‘the elderly’, ‘the poor’, ‘the rich’. So I think I wanted to specify a group

in a similar way. Is that not possible?

[Student 3, six weeks after Presessional)

6.4.9 Context 4IA

Type 4 a/an indefinite article was used by students with less accuracy across all three essay

Tasks. Examples of correct and incorrect use are shown in Figure 59. With a ratio TLU of

0.71, this was one of the most challenging contexts for students while also being the third most

frequently made article error in students’ essays (13% of all article errors). In addition, non-

referential noun phrases were the most common context of noun phrase use in the essays of

both students and L1 English tutors. These accuracy results contrast with previous research

202

4IA

firm does not differentiate, it will be like all

the rest and be forced to compete on price.

4IA

In conclusion, retailers should find

suitable strategy to keep their competitive

Error

If a retailer can provide high quality

products and service and establish

close relationship with its customer, it is

highly likely that they will gain competitive

advantages.

Error

hourly rate as a means of achieving high

quality output would make the pickers pay

more attention to pick the fruits without

damaging it.

Figure 59: 4GAIA errors compared to 4IA correct use

which has reported higher accuracy in this context (Bedmar & Papp, 2008; Crosthwaite,

2016a). The finding that Chinese students were less accurate with indefinite articles in non-

referential contexts than referential also contrasts with that of Xu, Shi & Snape (2016) who

reported the opposite effect, i.e. that non-referential types were more accurately used.

In interviews, Error 5 continued to confuse the student who reported that ‘…close

relationship…it sounds pretty good to me’ [Student 1]. Tentative anecdotal evidence was also

found that students were most confused when hypothetical recommendations were being

discussed in sentences where the head noun was within a densely packed nominal clause acting

as subject for a ‘would’ sentence. For example, the student who made Error 6 as the transcript

of her interview below shows, took over a minute to correct her error:

203

[45 second pause while student reads]. How?..., I’m struggling…[10 second pause to reread].

OK it is definitely with the or an. Definitely it needs an article I see. But to be honest I prefer

an. [Tutor confirms]. Because..yeah…I put nothing here…to be honest it’s challenging. I had

to read it several times to remember what it’s talking about. It’s the subject of the ‘would’…it

is the thing we are talking about. And the …it’s all a long noun. In fact it’s kind of

complicated…I find it hard when it’s so long and confusing.

[Student 2, six weeks after Presessional)

The insight shown above and several similar comments from students adds support to the

hypothesis that the density of nominalisations (through pre-modification) may have an effect

on this sort of error. Indeed, further analyses of the corpus with the AntConc concordance

programme showed that 51 of the 82 instances of 4GAIA errors were made determining pre-

modified noun phrases.

6.4.10 Context 3IA

There was great variation in the frequency with which students made reference to Type 3

contexts across prompts, ranging from as low as 6.4% to 14.3% (as mentioned in Section 6.4.2).

This analysis suggests underuse on behalf of the students, since the ‘expert writers’ never used

Type 3 indefinites in less than 20% of the noun phrases in the three essays. In terms of

accuracy of indefinite Type 3 a/an, 10.4% of all article errors related to omission of 3IA

indefinite articles. There was some range in the TLU accuracy of students across the three

prompts, but the median average across three essays was 0.89. Examples of correct and

incorrect use are shown below in Figure 60.

These results are between Diez Bedmar & Papp’s lower accuracy level findings and those of

Crosthwaite’s higher accuracy reports (2016a). Crosthwaite found that this article context is

204

3IA

report from YouGov SixthSense found that 93%

of consumers would continue to shop

somewhere, even if the retailers have their

loyalty scheme.

3IA

HP now has

functional structure which can be seen in figure

Error

Nevertheless, the manager used

very strange way to incentivize its employees.

Error

According to their website (Google, 2010),

their matrix structure creates

more relaxed and free working environment.

Figure 60: 3GAIA errors compared to 3IA correct use

inaccurately used at lower levels but is generally mastered by Upper-Intermediate levels. As

discussed in Chapters 3 and 5, a strong variable affecting this article type’s accuracy is likely

to have been the extent to which students were familiar with the topic. It is possible that

students’ accuracy improved in Task 2 due to both teaching/error correction and the case study

that they had previously been given on a similar topic to the essay 2 prompt. After showing

students examples of their correct and incorrect 3IA article use (Figure 60) student interviews

again provided tentative evidence that article omissions can be caused by a lack of knowledge

about the grammatical properties of the head noun. For example, Student 3 in response to Error

7 stated ‘ I think that’s a problem …is It’s about whether way is countable of not? Is ‘way’

countable?’

Again, the location of the head noun at the end of long pre-modified noun phrases challenged

students. For example, reviewing her Error 8 Student 1 reported:

205

Yes I understand it needs a. Let me read it again (20 seconds). I think I was confused by

‘environment’. I’ve never heard people say ‘a environment’. Also, it was at the end of the

sentence. To be honest this is challenging. At my high school they gave us lots of tests like

this and it was difficult when the word was in a…what do you call this? In a long group.

[Student 1, six weeks after Presessional)

Further analysis of the corpus with Antconc showed that many 3GAIA errors occurred with

count nouns requiring determiners acting as head nouns in more complex nominal groups.

These were often pre-qualified with adverbials or adjectives such as ‘very’. In fact, only 16

errors in the corpus were unmodified single nouns and 57 of the 73 Type 3GAIA errors were

pre-modified. In addition to lexis familiarity, the pre-modification of noun phrases appears to

be a variable affecting the accuracy of use in this article context. The two variables may also

naturally be linked, in that students need knowledge not only of the nouns they find in their

discipline, but of how these nouns are typically compacted and condensed into longer noun

phrases.

6.4.11 Context 5DA

In terms of frequency, the omission of definite articles with proper nouns, idiomatic academic

expressions and cultural uses (5GADA) represented a mid-level percentage of all article errors

(6.6%). With a TLU of 0.80, the three interviewed students corrected the mistakes (shown in

Figure 61) quite promptly when asked to do so, suggesting that some academic fixed

expressions were careless slips while other errors may have been related to terms (e.g. the UK,

the EU, the Middle East, the Caribbean) that the student had not known before but had since

learned.

206

5DA

the

other hand, in ASML, if an employee

overtime is too frequently, the manager

might be required to report to the HR

manager.

5DA

For example, E-series mobile phones

of Nokia features to attract business

people with the special design on

^buttons which are easily accessible to

connect to

the

Internet to check and send e-mail by one

click.

Error

one hand, staffs can take part in the

management of the company. On the other

hand, staffs can be given some equity

capitals.

Error

For example, the United Kingdom and

Poland both belong to

EU.

Figure 61: 5GADA errors compared to 5DA correct use

6.4.12 Context 1DA

With the second lowest level of accuracy (TLU=0.67), this study supports previous findings

that generic context definite articles are very challenging for Chinese learners (Diez Bedmar &

Papp, 2008; Crosthwaite, 2016a). Accounting for 5.13% of article errors in a context that

accounted for 5-9% of article use by either L1 writers or Chinese students, the errors were

found, as shown in Figure 62, most frequently in the introductions, background statements and

definitions provided by students.

207

1DA

According to Mintzberg (1979),

the

structure of an organisation can be defined

simply as the….

1DA

An accountant is a practitioner of

accountancy, which is

the

measurement, disclosure or provision of

assurance about financial information…

Error

In the 21st century,

PC or laptop has changed our life.

Error

Segmentation refers to the process of

dividing

whole market into smaller segments.

Figure 62: 1GADA errors compared to 1DA correct use

At interview, students demonstrated their confusion with generic contexts and Errors 11 and

12 were clearly not simple mistakes. The student who made Error 11 spent 50 seconds at

interview hesitating between Ø, a and the, before deciding the was necessary. The student who

committed Error 4 suggested a after some hesitation between Ø and a (not the definite article,

as three L1 English tutors had instead agreed was more appropriate).

6.4.13 Context 3ZA

With a TLU of 0.8 across all essays, errors of over specifying articles in Type 3 Ø contexts

were infrequent, but a contributing factor to the aforementioned ‘informal’ tone that was judged

to make its use inappropriate. Most overspecification errors occurred in Type 4, not Type 3

contexts. Only 5% of article errors were within this Type 3 obligatory Ø article context. The

majority of the 26 omitted Ø articles (overspecification of a/the), resulted from over

208

specification of the definite article (n=22), as shown in errors 13 and 14 in Figure 63 below,

but there was also occasional overspecification of a/an (n=5).

3ZA

For example, Coca-cola divided its

differentiated marketing into

soft drinks which include Coca-cola, Diet

coke, Sprite and Fanta.

3ZA

For example,

customers will not go to Sainsbury’s if it is

located in the area where the transportation

is not convenient.

Error

In the 21st century,

the

retailing has increasingly become intensely

competitive.

Error

The market has switched from

the

transaction marketing to the relationship

marketing since the 1980s

Figure 63: 3GAZA errors compared to 3ZA correct use

In 3GAZA contexts students again shared that, just as they had in 4GAZA contexts, they were

not confident about whether an article was required but often supplied one on the balance of

probabilities. As shown in the following two examples, the students saw this ‘gamble’ as more

prudent than a less confident omission. The first student certainly remained confused by the

researcher’s explanation about Error 13:

OK, so retailing is not countable. I can see that. But it’s not wrong to put the there is it?

[researcher explained informal tone it added]. To be honest, if I know there’s a choice and I

know adding the is not wrong, I always add it because it is not usually wrong. What is the

answer here? Is it grammatically wrong?

[Student 1, six weeks after Presessional)

209

Asked about her Error 14. Student 2 explained:

I was wondering… I feel uncomfortable with the transaction marketing now. There is no

article here? [Researcher confirms overspecification and explains how its use was overly

informal for the register]. [Laughing] ? You know, sometimes in my mind I’m not sure. I often

just add an article if I’m not sure!

After some minutes looking at the next set of errors, the same student returned the researcher’s

focus to error 14 to explain:

I think I also like balance sometimes. it was because of ‘The market’…so I wrote ‘.the

marketing’. I wanted balance. Chinese people ... I think logically we need the in one part of a

sentence if we see it in others…why not the here? It’s more about balance. You know, Chinese

people like balance. We like to match everything.

Analysing further the 22 instances of oversuppliance of definite article in Type 3 contexts

within the corpus with Antconc, it was found that 17 occurrences had qualified head nouns.

6.4.14 Context 1IA

The finding that Chinese students’ median TLU of indefinite articles in generic contexts was

0.75 confirms Crosthwaite’s (2016a) hypothesis that L1 transfer effects may be more

prominent for indefinite article uses in generic contexts than in other contexts due to some

equivalent uses of proximal demonstratives in other generic contexts. Indeed, Crosthwaite

(2016a) reported a TLU of 0 in one essay prompt by B2 learners. Nevertheless, the analysis of

such rare uses of articles is not without problems. Crosthwaite’s finding that students showed

a TLU of 1.00 in another prompt shows the difficulty of analysing this rare generic context of

use in small samples with non-parametric data since the median value can be skewed by zero

counts. Indeed, were all zero counts to be removed from the calculation, the median TLU value

would have also been 1.00 across all Tasks in this study. Yet in the rare occasions of use, such

as in definitions, students do find this context challenging, as shown by Errors 15 and 16 in

210

Figure 64 below, perhaps due to lack of L1 positive transfer effects. Unfortunately, these

students were not interviewed and participating students had not made any 1GAIA errors.

1IA

A market segment has been described as

homogeneous group of consumers with similar

needs, wants, values and buying behaviour

(Harrell, 2002:166).

1IA

While, another definition from Cole (1995)

is a little bit different from Stacey, which is

intangible web of relationships between people,

their shared purposes, and the tasks they set

themselves to achieve those purposes.

Error

Buchanan and Huczynski (1997) claim that

tall organization structure is the structure with a

great deal of levels of hierarchy, such as the

armed forces, the police and the civil.

Error

Schein(1984) define corporate culture as a

model that …

organization has found or improved in studying

to solve their own problems, which involved in

outer change and inner integration.

Figure 64: 1GAIA errors compared to 1IA correct use

6.4.15 Context 5IA

Chinese students showed the lowest accuracy in 5IA contexts (Mdn TLU = 0) and examples of

both correct use and incorrect use are shown below in Figure 65. This evidence that students

had low accuracy with fixed academic expressions colligated with indefinite articles supports

earlier research by Liu and Gleason (2002) which suggested that inaccuracy with such

idiomatic and cultural uses of articles remain at higher levels of proficiency. However, errors

in this context accounted for only 0.6% of article errors and none of the three students

interviewed had made such errors in their writing.

211

5IA

Take gender as

example. The buying behaviours of female…

5IA

result, the company encourage staff improving

themselves

Error

For example, 13 employees committed

suicide…

few months ago.

Error

short time, it could increase customer turnover,

but it is useless to attract high-income

customers.

Figure 65: 5GAIA errors compared to 5IA correct use

6.5 L1 transfer effects

These findings presented above suggest that the B2 level Chinese students had far greater

accuracy with Type 2 definite articles and Type 3 indefinite articles than Type 1 generic

articles. These results appear to lend more support to Crosthwaite’s (2016a) hypothesis that

Mandarin does have equivalent classifiers and demonstratives that somehow provide positive

L1 transfer to 2DA, 3IA and 3ZA. In contrast, the use of bare nominals for generic contexts in

Mandarin may have contributed to the Chinese students’ poorer accuracy with generic context

articles in this study. Indeed, Crosthwaite’s (2016a) suggestion that indefinite generic articles

are the least likely to have positive transfer effects have been confirmed by this study as this

context showed the least accurate use apart from the 5IA context.

On the one hand, the deficit of pragmatic awareness that the Chinese students displayed in their

overspecification of redundant definite articles lends weight to Díez-Bedmar & Papp’s (2008)

212

hypothesis. Equally, the fact that classifiers and demonstratives are reportedly more ‘optional’

in Chinese (Li & Thompson,1981; Robertson, 2000; Xu, Shi & Snape, 2016) may also provide

insights regarding some of the confusion the interviewed students showed about the

redundancy of some of their definite articles. It is certain that the complexity of the English

article system and the variance in how different disciplines generalise about actors in their field

(e.g. ‘The elderly’ versus ‘managers’) affects all L1 students. However, article-less languages

are likely to have fewer positive transfer effects given the use of bare nominals in generic

reference.

6.6 Linguistic context of noun phrases

Given the suggested effect of pre-modification highlighted in the students interviews, a

summary analysis was made of the linguistic contexts of the noun phrases in the entire corpus

(including the case studies), in particular with respect to the effect of noun phrase density.

As the comparisons shown in Table 27 and

Table 28 demonstrate, the introduction of qualifying adjectives and nouns had a dramatic effect

on accuracy of use: 64 percent of all article errors were found within noun phrases in which

the head noun had at least one qualifying word.

Table 27: Density of noun phrases (in 610 article errors)

*Some Percentages were rounded up for ease of reading

Table 28: Analysis of 9186 correct noun phrases

Article errors: head noun

not pre-qualified

Article errors: one word

qualifying head noun

Article errors: two or more

words qualifying head noun

221

343

36%

56%

8%*

213

*Some Percentages were rounded up for ease of reading

This error effect can be best exemplified by some examples (see Figure 66 below) of proper

^Aldi store

^KFC company

^Tesco Club card

^Intel Core chip

Figure 66: Omissions of article with nouns pre-qualified with proper nouns

nouns which qualified singular common count nouns. It may be that students were confused

about which nouns were setting the ‘rules’ of determination, i.e. overgeneralising their learned

knowledge that these proper nouns required no article.

This issue of noun phrase density also affected common nouns, as shown in Figure 67 below.

This preponderance of premodifiers in nominal groups confirms the literature’s

characterisation of academic writing as making extensive use of complex nominal groups

(Halliday & Martin, 1993; Biber & Gray, 2010; Bennett, 2011). In some ways this shows the

Correct use: head noun not

pre-qualified

Correct use: one word

qualifying head noun

Correct use: two or more

words qualifying head noun

6491

2510

185

71%*

27%

214

^business environment

^management team

^rural market

^economic crisis

^loyal customer

^retention scheme

^concentration strategy

^higher needs

^employment relationship

$the consumer information

$the traditional marketing

$the customer loyalty

$the brand loyalty

$the unpaid overtime

$the foreign nationals

$the population growth

$the purchasing behaviour

Errors with two or more words qualifying head noun

^undifferentiated marketing strategy

^information technology industry

^high speed calculation

^very strange way

^online retail store

$ the popular and profitable production

$ the internet search industries

$ the three important parts

$ the efficient accountancy information

$ the traditional pricing strategy

Figure 67: Typical omission and oversuppliance errors (with pre-modified head nouns)

215

success of the Presessional course in developing students’ ability to engage in what Halliday

& Martin (1993) term ‘grammatical metaphor’: objectifying actions and events in nominal

groups rather than verbs in order to compress information (Biber & Gray, 2010) and make

chains of arguments in the juxtaposition of these nominalisations (Halliday & Martin, 1993).

Students’ problems determining these more densely packed phrases is likely affected by the

ambiguity created in compressed noun phrases in addition to the many other variables always

affecting L2 students’ accuracy.

This finding lends support to Fang, Schleppegrell and Cox's (2006) conclusion that novice

academic readers/writers face challenges using densely packed nominalisations. The

hypothesis that pre-modification could affect Chinese learners’ confidence in article use would

to some extent go against the L1 transfer assumption, since the head noun is always the last

noun in Chinese. This hypothesis was tested further in the study presented in the next chapter.

6.7 Conclusion

The findings in this study have confirmed that one common perception among some teachers

that Chinese students are extremely inaccurate with articles is a fallacy even at this Upper-

Intermediate (IELTS 5.5/6.0) level. In fact, article use by Chinese L1 students is much more

accurate than it is inaccurate. The two most frequent errors (omission of definite articles in

referential definite article contexts and oversupply of the definite article in non-referential Ø

contexts were marginal compared to students’ near target-like accuracy in both areas (2DA

TLU of 0.88 and 4ZA TLU of 0.95, respectively). The reason these levels of accuracy are

higher than those reported by Díez-Bedmar & Papp (2008) but lower than those of Crosthwaite

(2016a) may be attributable to the narrower range of proficiency level than the former study

and the slightly more challenging task demands than the latter.

216

The academic writing genre and the familiarity of topic lexis have been shown to greatly impact

upon students’ accuracy of article use. However, while Crosthwaite’s (2016a) claims that B2

level Chinese students face few challenges with referential article use seems convincing when

the students have a simple prompt with known vocabulary/genre, at B2 level they may continue

to be less accurate with new topics that require more abstract levels of reasoning (e.g.

discussion of possible scenarios, unreal conditions, and possibility). Since, most Postgraduate

students will ultimately be required to write a dissertation, which will requires dexterity in

moving between generalised abstractions and specific examples in a way that is more similar

to the essay Tasks than the case-study assignment, it is concluded that Chinese students at B2

level entry to the UK Higher education still face challenges with the article system in certain

contexts.

In terms of L1 transfer effects, the strongest claims in the literature of a ubiquitous error effect

that article-less background students suffer arbitrarily (fluctuation hypothesis) were not

supported, as there were very few ‘a for the’ or ‘the for a’ misuses of articles. However, as

discussed in Section 8.5, these findings provide some limited support for the L1 positive

transfer hypothesis, since the Chinese students at this level showed greater accuracy in the

article contexts that have been shown to have optional equivalents in Chinese, and lower

accuracy in generic and unspecific contexts that have no equivalent (Crosthwaite, 2016a). In

addition to low accuracy in definite and indefinite generic contexts, this study has shown

Chinese students’ use of non-referential indefinite articles to be less accurate than other

contexts, which could result from negative transfer because Chinese never takes any sort of

classifier phrases in such noun phrases (Li &Thompson, 1989).

In its finding that oversupply of the definite article is frequent, even if not quite as frequent as

that of definite article omission, this study confirms the findings of previous research (Chuang

217

& Nesi, 2006; Lee & Chen, 2009). Many uses of the definite article in 4ZA contexts were

judged not only to be ‘marked’ but so inappropriate as to be considered incorrect by the three

annotators. Moreover, when asked, three students who were interviewed conceded a lack of

confidence and reported using articles whenever in doubt, on the basis that they were not

grammatically impossible. The three annotators judged that many definite article

overspecifications had a marked impact on the register of some of the students’ essays, to the

extent that they were judged incorrect. This impact of article errors on register seems to echo

Lee & Chen’s (2009) report of ‘a tendency to use the even when talking about general rather

than definite subjects.’ (p.287).

On the one hand, there is no reason to suspect this problem to be limited to Chinese learners.

On the other hand, it does show that excessive use of redundant definite articles may be one of

the contributors to the informal tone that novice Chinese students are sometimes argued to

display in their academic writing (Gilquin & Paquot, 2007, 2008; Hinkel, 2003). A detailed

discussion of the implications of these findings for teaching will be provided in Chapter 8, but

it should be highlighted here that students are challenged by such register problems. Indeed,

they acknowledged their confusion. Rather than vague comments about ‘informal tone’ and

‘register’ in their error corrections, it is more likely that students would be helped by more

focus in EAP writing classes on the conventions for generalising the most common terms in

their discipline (plurals, mass nouns or generic the). Moreover, to increase their register

awareness in their disciplines, students need to learn how their disciplines may or may not use

adjectives as nouns (e.g. the elderly, the poor, the young), or may prefer to use mass nouns

(e.g. government) or plural nouns (managers, students, teachers) when generalising about the

main actors within a given disciplinary field. This thesis therefore frames article accuracy, in

so far as this involves reducing overspecification of definite articles, as an issue of register

218

awareness. The argument that register is affected by article oversuppliance, may convince

more EAP teachers and materials writers that article use cannot be ignored in the EAP syllabus.

The student interviews also provide support for the hypothesis that students often make article

errors through omission or overspecification due to a lack of grammatical knowledge (e.g.

countability), as has been suggested in the literature (Lu, 2001; Milton, 2001; Díez-Bedmar &

Papp, 2008; Xu, Shi & Snape, 2016). This confusion about countability is also likely to be

related to the link other researchers have highlighted between article accuracy and the

familiarity students have with a topic and its lexis (Díez-Bedmar, 2015; Leedham, 2015;

Crosthwaite, 2016a, 2016b). This study has also proposed the novel hypothesis that the degree

of noun phrase complexity can affect whether students omit or remember a definite or

indefinite article when obligatory. This hypothesis is confirmed in the next chapter, which

lends support to the voices of those who call for greater focus in EAP on this aspect of academic

writing (Biber and Gray, 2010; Bennett 2011).

In the previous Chapter it was concluded that error correction and teaching may have

contributed to a temporary improvement in accuracy. Equally, however, as shown by the

improved accuracy levels in the case studies, the effect of reading and topic familiarity should

not be underestimated (Crosthwaite, 2016a). The finding that Chinese students’ accuracy in

article use was higher in the case study Task confirms that article use is greatly affected by

students’ knowledge of the topic-related lexis, suggesting that article accuracy is linked most

with developmental factors and students’ knowledge of the topic vocabulary (Díez-Bedmar,

2015; Leedham, 2015; Crosthwaite, 2016a). The use of a set text (a case study) in this task may

have allowed students to familiarise themselves with new vocabulary and its grammatical

characteristics. In contrast, asked to use new topic vocabulary in essays, without the

contextualisation of the new vocabulary in a case study text, students made more errors in

219

articles. This may also be related to the needs for more abstract (less specific) levels of

reasoning.

At the very least, it would seem that more familiar vocabulary can reduce the likelihood of

article omission, which would appear to support Crosthwaite’s (2016a) conclusion that Chinese

students, by Upper-Intermediate levels can in the right circumstances present near target-like

accuracy levels. While the responses to essay prompts 2 and 3 were more authentic (in

demanding the use of referenced evidence), the lower accuracy of article use in prompt 1

responses may also reflect the students’ unfamiliarity with the topic. Although the students had

reviewed texts on motivation theories before being given the task, the keyword in the rubric

(‘fear’) had not been contextualised within texts for the students and their inaccuracy in

responses to prompt 1 may have been greatly affected by the lack of opportunities to see

contextualised examples. The more authentic case study prompts may therefore allow more

valid insights into what students are capable of once they are fully embedded in their

disciplines’ readings and aware of the genre’s demands. In her criticism of the short essay

tasks in many learner corpora studies, very different from the background reading and research-

facilitated topic familiarisation in authentic academic writing, Leedham (2015) argues that this

contact with the discipline specific texts allows students to show greater accuracy.

The hypothesis that proficiency would be the main determiner of article accuracy was not

supported by the analysis in this study. However, the extremely small sample and the

particularly small sample of B1 level students are likely to have greatly impacted on this

analysis, so the current findings must be treated with caution and cannot be regarded as

conclusive. As shall be shown in the next chapter, the important effect of proficiency required

further investigation with a far larger number of students. The next study also allowed further

investigation of the novel hypothesis proposed in this study that the degree of pre-qualification

220

in a noun phrase may also be an important factor in students’ accuracy of articles in more

academic genres of writing.

221

7 TESTING OF HYPOTHESES USING ONLINE MULTIPLE-

CHOICE GRAMMATICALITY JUDGEMENT TEST

This chapter presents the final study of this thesis in which the methodology switches from a

smaller to a larger scale, with the use of grammaticality judgement multiple-choice questions

embedded within texts. A 100-item article use multiple-choice test was inserted within the

review quizzes on the Virtual Learning Environment (VLE) of an English for Academic

Purposes Presessional programme at the University of Birmingham between 2015 and 2017

and an analysis of the answers was used to investigate some of the hypotheses developed in

previous chapters. Chapter 6 presented some of the linguistic contexts that may challenge B2

level Chinese students, but three aspects of Chinese L1 university students’ English article use

required further investigation. Firstly, the corpus data was collected from only 30 students.

Thus, efforts to examine the effect of proficiency level had been previously constrained by the

fact that only 10 of Chapter 6’s participants had IELTS 6.0 levels (the minimum English

proficiency requirements on the majority of university programmes in the UK).

A larger sample of Chinese students from a higher range of proficiency levels was thus required

to investigate further the critically important issue of how students’ overall language

development related to their article use. The previous chapters reported two other findings that

needed further attention, namely that students struggled in the interchanging focus in academic

writing of discussing generalisations in one part of a text to providing specific examples in

another part (and the interweaving between general and specific focus). Chinese students often

appeared to overspecify definite articles when returning to general statements after giving

examples. Finally, their omission of articles often appeared related to noun phrase pre-

qualification.

222

The critical discussion of Methodology in Chapter 3 concluded that the balanced use of testing

and corpus-based approaches can be beneficial and that the testing approach can complement

the corpus-based paradigm. The advantages of multiple-choice tests (control over linguistic

context, avoidance of ambiguous error, focus on rarer language forms) were compared to their

disadvantages (the association of testing with a ‘deficit’ approach, a lack of

authenticity/validity and the exaggeration of error frequency) for investigating student

accuracy with grammatical forms. Although a corpus-based approach has generally superior

validity (Sinclair, 1991; Johns, 1991; Granger, 1996; Chuang & Nesi, 2006), a testing approach

can be used to address certain limitations imposed by real world constraints. The use of the

testing methodology in this chapter is certainly not proposed to be a necessary step in verifying

the findings of corpus linguistics. The assumption made in this mixed corpus-based and testing

approach methodology is that only the corpus-based methodology can provide insights into the

actual challenges students face in academic writing, but that a testing approach can complement

research by addressing several issues that real world constraints make more difficult for corpus-

based studies to examine. In this study these included the effect of proficiency on error and the

impact of pre-qualification of noun phrases.

Thus, while it might be tempting to see the use of a far larger number of participants as a

method of testing hypotheses developed regarding the order of use or article contexts according

to the Bickerton Framework, this was not the intention. This testing approach is instead

proposed as a method of overcoming the issue of broad IELTS bands and low participant

numbers of the preceding study which made statistical inferences so challenging to measure.

Where a corpus-based approach has identified an error type as a potential problem, the testing

approach offers a practical way of testing the hypothesis with different proficiency levels. The

scaled-up investigations into overspecification and noun pre-qualification and their relations to

223

general writing proficiency would have been problematic with a corpus-based study. For

example, the aforementioned challenge of differentiating between acceptable/marked/incorrect

use with regard to definite article redundancy meant that scaling up a corpus-based approach

would have entailed parallel reliability testing beyond the limited means of the research project.

A testing approach allowed for scaling up the participant numbers (far beyond the 30 students

previously researched) which was necessary to make claims about complex and nuanced issues

(pre-qualification/definite article overspecification) that could be in any way generalisable to

most Chinese L1 Higher Education students at B2 level in the UK.

7.1 Study overview

This study maintains focus on the second Research Question:

Research Question 2: What are the linguistic contexts of obligatory article use in which

Chinese L1 students at Upper-Intermediate levels of English L2 are most likely to make

errors?

Firstly, this investigation goes deeper into the issue of oversuppliance of definite articles. The

findings reported in Chapter 6 supported the observation of others (Chuang & Nesi, 2006; Lee

& Chen, 2009) that Chinese L1 university students frequently oversupply the. Building on Lee

& Chen’s (2009) evidence of a marked effect on academic style, the evidence collected for

Chapter 6 was used to speculate that Chinese students might be oversupplying the when

moving pragmatically from specific examples to more generalised statements. In particular, it

was noticed that this overspecification occurred in essay introductions (frequently Type 1

generic contexts) and critical evaluation of conditional/hypothetical abstract concepts (Type 4

non-referential contexts). Thus, when students were making generalised statements and claims,

they frequently overuse the definite article when L1 English tutors would have used Ø to make

224

it clear they were generalising. Moreover, in the opinion of three EAP tutors (the researcher

and two annotation checkers), this redundancy produced an informal tone in the writing.

In the present study, it was hypothesised that the observed oversupply would persist at higher

6.0 and 6.5 IELTS proficiency levels. Moreover, given the importance of familiarity with a

topic and its lexis, it was predicted that proficiency level would have a negligible effect on

accuracy among Chinese students who appear (when unfamiliar with the grammatical and

pragmatic norms of new vocabulary) to strategically ‘gamble’ on definite articles being

acceptable (even if marked).

Secondly, this study investigates the tentative evidence offered in Chapter 6 that pre-

modification of noun phrases appears to affect article accuracy. In particular, as discussed in

Section 6.6, more frequent omissions occurred in densely packed (pre-modified) noun phrases.

The testing approach allowed for students to be presented with identical nouns in the two

different conditions (simple noun phrases and pre-modified ‘complex’ noun phrase contexts).

It was hypothesised that article use in complex noun phrases would be significantly lower

among the students as a whole, but that proficiency level might have an important effect on

accuracy of use of this kind.

More generally, this study examines the effect of general writing proficiency level on the

students’ accuracy of article use. In the previous two chapters, the research focussed mainly

on IELTS 5.0-6.0 proficiency levels, and the small sample size was shown to have further

constrained the ability to review the important issue of students’ overall language development.

It was hypothesised that students’ general writing ability would significantly affect their

accuracy of article use.

225

7.1.1 Summary of three hypotheses

The three hypotheses can be summarised as follows:

1. Overspecification of definite articles in Type 1 and Type 4 generalisations (in which

L1 writers would use Ø) is a frequent error and will observed at IELTS 6.0 and 6.5

levels of proficiency among Presessional programme students.

2. Omission errors in Type 2DA, 3IA and 4IA contexts will be more frequent in pre-

modified noun phrases than simple noun phrases.

3. A student’s proficiency level on entry to the Presessional (as measured by IELTS

writing tests) will be related to their accuracy of article choice at the beginning of the

programme.

7.2 Methodology

The objective of the final study was thus to develop an online quiz that could ethically and

reliably present students with a range of questions in order to test the above hypotheses. The

issue of article use overspecification, however, is rarely covered in grammars or text books so

most items in the final test were created specifically for this research or adapted from published

grammar tests. The-100 item quiz is published in full in Appendix 9 (p.334).

7.2.1 Test development

Multiple drop-down question types are cloze type items embedded within texts (but with fixed

responses), allowing for the students to be presented with typical article use choices. With the

aim of presenting students with typical pragmatic choices they face when writing a text, it was

particularly important to contextualise the questions within authentic texts.

The creation of a 100-item multiple-choice (multiple dropdown) article use quiz required five

distinct stages of development. These comprised (1) the sourcing of possible questions; (2)

226

creating questions that could test the hypotheses; (3) preliminary checking of questions with

L1 English tutors; (4) piloting of the quiz interface with students; (5) final review and

development and (6) final reliability testing with L1 English tutors.

1. The sourcing of potential questions

The first stage in the process of creating the test items was to collect examples of students’

errors collected in previous chapters. Ten words that students had overspecified with definite

articles in the learner corpus were: industry, crime, pain, fear, analysis, policy, effects,

unemployment and demographics. To contextualise these words in extracts from authentic

texts, keyword searches were made for academic texts (mainly journal article abstracts) or

published grammar tests that contained this vocabulary. Although these texts were then

heavily adapted and paraphrased to incorporate other questions, the sources of these texts are

also shown in Appendix 9.

Given the importance of topic and the predicted importance of noun phrase pre-qualification,

all texts were analysed for lexical density (Ure, 1971) using the measure of the number of

lexical words divided by the total number of words. In addition, the Analyzemywriting.com

website was used to produce Flesch-Kincaid readability score for each text. This measure

provides an equivalent US educational grade level required for text comprehension. As shown

in Appendix 9 the final test was shown to include five registers of text: high density academic

English, mid-density academic English, mid-density news English, mid-density general

English and low-density general English.

In order to make the texts representative of academic English, efforts were made to

approximate the obligatory context proportions found in the essays in Chapter 6: 6 % Type 1,

30% Type 2, 20% Type 3, 30% Type 4 and 15% Type 5. To achieve this, the questions were

227

further supplemented with a number of questions adapted from two popular grammar books

that cover article use in some detail: Murphy’s (2012) ‘English Grammar in Use’ and Hewings’

(2013) ‘Advanced English Grammar in Use’.

2. Creation of questions to test hypotheses

Illustrating the developmental process with reference to Question 10 shown in Figure 68, to

investigate the observed oversuppliance error associated with ‘crime’ and ‘policy’ used in

generalised statements, an abstract from a paper about ‘Long term unemployment and violent

crimes’ (Almén & Nordin, 2011) was adapted. Analysis showed the lexical density to be

63.64% and its readability score on the Flesch-Kincaid scale to be 13.66 (university level),

suggesting that it was a good context in which to test article use in an EAP setting. The Type

1 generic phrase ‘property crime’ and the Type 4 phrase ‘future crime and policing policy’, as

underlined in Figure 68, were predicted to tempt students to use a redundant definite article if

they had lower confidence in article use.

Question 10.

Abstract.

The positive association between unemployment and [Ø] property crime is well known (Mustard

2010; Lin 2008; Nordin 2009). This article discusses [a] study that evaluated the association of [Ø]

long-term unemployment and [Ø] crime. This involved analysis of annual data from 288

municipalities in Sweden from 1998 to 2012. Results showed that unemployed adolescents are more

prone to violent crimes than property crimes. Several recommendations are made that could advise

[Ø] future crime and policing policy.

Figure 68: Example of question development (Question 10)

228

Multiple dropdown quiz questions were created and then uploaded to the university’s Canvas

VLE quiz tool. As shown in Figure 69, all dropdown menus gave a choice of four answers

(one correct answer, three incorrect).

Figure 69: Example of multiple dropdown question

In order to test the hypothesis that pre-qualification would affect article omission, students were

presented with ten countable nouns in twenty different contexts, firstly in simple noun phrases

and then as head nouns in pre-qualified nominal groups. Having identified the ten nouns that

229

were either found twice in the same text or could be added to other texts of similar difficulty,

six of the nouns were found in juxtaposed forms while four required a few simple

manipulations. Figure 70 shows how ‘system’ was presented twice in one text (Question 4)

while ‘scale’ occurred in two texts (Questions 4 and 6).

Question 4

Abstract

[ ] pain is a part of childbirth. However, there had been very few studies into relative

perceived pain levels at childbirth before 2015. Thus little is known about [ ] scale of pain

that different women feel. Even in recent years, although there have now been [a] few

empirical studies, there remains little research with the rigor and scale required to

scientifically examine this crucially important area. Moreover, without [ ] system to

reliably compare findings, researchers struggled to measure subjective self-reported

perceptions and Teale's (2015) study was the first project to propose [ ] severity of

perceived pain classification system. While there are still today only [a] small number

of research projects investigating this area, interest is growing in this promising research

field.

Question 6

Abstract

This article seeks to acknowledge [_] fear as a strong presence in [_] workforce by

identifying from whence it comes. It further seeks to offer clarity on how individuals

respond to [_] fear. For this research, [a] sample of 776 employees were posed 19

questions which were measured on [_] Likert-type scale. [_] questionnaire was

developed by [the] authors in the absence of any such measurement tool which blatantly

asks employees what makes them fearful in their place of work. Distributions of [the]

sample covered entry level, supervisor level, manager level, and senior manager level.

Results indicate that concerns over the opinion of their direct supervisor and the fear of

being fired or laid off are the primary fears of employees. [_] analysis indicates additional

categories or clusters of fear. This research recommends that manager-subordinate

relationships and communication must improve to foster an open environment which

reduces the instances of [_] fear.

Figure 70: Paired noun phrase examples

230

3. Preliminary checking of questions with L1 tutors

Five L1 English tutors agreed to review the questions and provide feedback about any answers

they disagreed with. The aim was to present students with dichotomous choices with only one

correct answer. In fact, in many academic writing contexts the/a/an/ Ø can be used

interchangeably with only a nuanced change in meaning (rather than unambiguous error), but

the quiz questions aimed to present students with unambiguous choices. Questions that were

answered incorrectly by two of the tutors were immediately discarded (on the assumption they

were ambiguous). Any questions with one incorrect response were taken back to the tutor for

discussion (and only included if they agreed with the answers given by the researcher and four

other tutors retrospectively).

4. Pilot of quiz with L2 students

20 volunteer students took the quiz and gave feedback on the preliminary quiz interface which

included 30 texts and 150 cloze choices.

5. Final review and development

Based on feedback that the original quiz was taking students too long to complete, the quiz was

reduced in size to 20 texts with 100 cloze article choices in order to encourage maximum

participation. As a result, the proportion of article error types was amended: 6% Type 1 generic

contexts, 33% Type 2, 32% Type 3, 25% Type 4, 15% Type 5.

6. Reliability checking of quiz with L1 tutors

In the final development stage, three further L1 English tutors agreed to take the completed

test. Scores ranged from 97-99%, with a 96.67 pairwise agreement and a Fleiss kappa of 0.93.

Two quiz items that were not answered correctly by all tutors (signifying ambiguity of choice)

were further modified.

231

7.2.2 Ethical considerations

As requested by the University of Birmingham Ethics Board (ERN_14-1455, ERN_16-1493)

it was made clear to students that they did not have to consent to the use of their review test

data by the researcher. First, a brief 10-minute overview of the research project was presented

to the students at the end of their VLE induction on their first or second day on the Presessional

course. Students were given some background information (see Appendix 4) about the

research and several assurances about how their data would be used. Rather than ask students

to sign a form under pressure, the form was only presented to the students, by the class teachers,

at the end of the second week of the presessional programme without the researcher present.

7.2.3 The participants

The participants in this study were all enrolled on the postgraduate level EAP PG 10-week

Presesional programme at the University of Birmingham between 2015 and 2017. Each year

this programme, which runs from June to September, teaches general academic English four

days a week and an ‘English in My Subject’ module about various disciplines one day of the

week to 500 - 550 postgraduate students. The IELTS entry scores on entry to the Presessional

typically range from 5.0 (overall) to 6.5 (overall). On successful completion, students progress

to all schools across the university (apart from the Business School which is served by the

aforementioned BME programme).

This programme was chosen, rather than the BME or undergraduate programme, because 1)

the researcher was then on the management team of this course and had responsibility for the

VLE and 2) the course was an early adopter of an online blended approach to learning. This

blended approach involved a weekly online review quiz on the university VLE which tutors

used to monitor learning and progress. Moreover, Unit 1 of the programme covered

232

nominalisations and article use, thus adding questions about article use to their week 1 review

quiz was a justifiable addition to the materials.

Final participant numbers

Of the 1423 students who took the review quiz between 2014 and 2016, 466 L1 Chinese

students consented for the use of their data in this study. Eleven students were seen to spend

less than 10 minutes in the quiz interface (i.e. had not engaged meaningfully with the quiz) so

their results were removed, leaving the quiz results of 455 participants who engaged fully with

the quiz. These quiz results were added to the a spreadsheet with information about the

students’ IELTS scores.

7.2.4 L2 English proficiency levels

By 2015, due to the massive expansion of numbers, the Presessional programme was no longer

using a placement test on entry to identify students requiring further language support (now

offered on a teacher referral basis). Therefore, this study used the students’ IELTS results as

the most appropriate measure of English L2 proficiency. With overall IELTS bands ranging

from 5.0 to 6.5 (four levels), statistical tests could be applied with significance that could not

be achieved in previous studies (with only two IELTS band levels).

As shown in Figure 71 below, the overall IELTS results of the participating students (n=455)

included IELTS level 5.0 (34 students), 5.5 (205 students), 6.0 (201 students) and 6.5 (15

students) IELTS bands.

233

Figure 71: Histogram showing overall IELTs scores

However, as this thesis is focussed upon written accuracy rather than listening/speaking/

reading comprehension, the main measure of writing proficiency level used in this study was

the IELTS writing band achieved by these students prior to arrival. As presented in the

histogram in Figure 72, the IELTS results of the 455 students who took the article use quiz

included IELTS writing scores of 4.5 (4 students), 5.0 (105 students), 5.5 (239 students), 6.0

(90 students), 6.5 (11 students) and 7.0 (6 students). As shall be acknowledged in the findings,

the extremely small number of participants in the IELTS 4.5, 6.5 and 7.0 levels means that the

results are more representative of IELTS levels 5.0, 5.5 and 6.0 levels than others.

234

Figure 72: Histogram showing IELTS writing test scores

7.2.5 Procedures

The 100-item article use multiple-choice test was inserted into Unit 1 of the Virtual Learning

Environment (VLE) between the end of the first week and the end of the second week on the

10-week EAP PG programme. Generally speaking, the review quizzes on the VLE do not form

any part of the course assessment and there is variation in the extent to which individual tutors

make the quizzes optional or mandatory for their individual classes (e.g. some tutors use the

analytics to inform discussions in tutorials with students while others do not). Students could

not see their feedback about their quiz responses/the quiz answers until the quiz was closed at

the end of Week 2.

7.2.6 Data analysis

No other attempt was made to remove ‘outliers’ or ‘clean the data’ in any way other than

removing 11 students who has spent fewer than 10 minutes on the quiz. After removal of these

235

11 respondents the distribution of the percentage of correct responses (mean 64.63, sd 8.67) of

the remaining 455 participants was reviewed, as illustrated in Figure 73. Although the

histogram appears normally distributed at first sight, a Shapiro-Wilk Test of Normality (see

Appendix 10, Figure 96) showed that the distribution was not statistically normal (p value of

0.12). Non-parametric statistic tests were therefore used for this study.

Figure 73: Histogram of test scores

After downloading students’ responses from the Virtual Learning Environment to a

spreadsheet, various question groups were further summarised using Microsoft Excel formulas.

The researcher then checked the spreadsheet for data-entry errors and a colleague did a small

audit to check for formula error.

7.3 Findings and discussion

7.3.1 Hypothesis 1

Overspecification of definite articles in Type 1 and Type 4 generalisations (in which L1 writers

would use Ø) is a frequent error and will observed at IELTS 6.0 and 6.5 levels of proficiency

among Presessional programme students.

236

As hypothesised, overspecification of definite articles was an extremely frequent error

produced in the multiple-choice test. Indeed, as Table 29 shows, over half the respondents

chose a redundant definite article (an error as judged by five L1 English tutors) in three of the

Table 29: Results of ten questions investigating oversuppliance (n=455)

Question extract from fuller text shown to students

The



✓

Other



Abstract

This article seeks to acknowledge [_] fear as a strong presence

in [_] workforce by identifying from whence it comes.

Abstract

The article discusses how [_] latent semantic analysis (LSA)

can be used to analyze electronic health records to show…

Compared to the service sector, the UK has witnessed a decline

in [_] industry since the 1950s.

Abstract

[_] pain is a part of childbirth. However, there had been very

few studies into relative perceived pain levels at childbirth

before 2015.

(end of paragraph) Several recommendations are made that

could advise [_] future crime and policing policy.

(end of paragraph). [__] results support the notion of [_] direct

effects of the performance appraisal characteristics on the

consequence variables.

(Start of abstract) [_]Cluster analysis is often used for [_]market

demographics segmentation.

[_] Cluster analysis is often used for [_]market demographics

segmentation.

(Abstract). This article discusses [_] study that evaluated the

association of [_]long-term unemployment and [_] crime.

This article discusses [_] study that evaluated the association of

[_]long-term unemployment and [_] crime.

NR = no response

237

ten questions testing overspecification in Type 1 and Type 4 contexts. Of equal importance, in

all but one of the nine prompts, fewer than 50% of respondents correctly opted for Ø article

(the article judged as most appropriate by five L1 English tutors).

Students’ writing proficiency, as measured by their IELTS writing grades, had no significant

effect on their scores in the ten questions testing overspecification. A marginal effect can be

observed between IELTS level 4.5 and IELTS level 6, as presented in Figure 74 below. This

shows the median number of oversupplied definite articles dropped from 10/10 to 6/10 between

IELTs level 4.5 and 6.0. However, the scores show great variance and students with IELTs

6.0/6.5 and 7.0 made an equal median number of overspecifications with the definite article.

Figure 74: Oversuppliance (ten questions) box-plot (n=455)

238

In fact, Figure 74 shows the Mdn score (7/10) remained constant between IELTS 6, 6.5 and

7.0 writing levels. As a result, analysing the results of all six groups with an Independent-

Samples Kruskal-Wallis test, no significant differences (P ≤ 0.05) were observed in the

frequency with which students at different IELTS levels oversupplied definite articles (H

(5)=9.330, n=455, p=.097). However, while these findings may be representative of the

Presessional cohort of students, reliable generalisation to the wider population in HE would

need more representative samples of students at the lower IELTS 4.5 and higher IELTS 7.0

levels.

Notwithstanding the likely effect of proficiency at lower levels, the fact that overspecification

remained an issue for students with IELTS writing scores of 6, 6.5 and 7.0 provides tentative

evidence to support the view that, at least at the start of their programme, many B2 Chinese

students may oversupply definite articles with certain words in Type 1 and Type 4 contexts.

7.3.2 Hypothesis 2

Omission errors will be more frequent in contexts of pre-modified noun phrases than simple

noun phrases.

Table 30 presents the percentages of omitted articles in ten paired nominal groups (simple noun

phrases and pre-modified noun phrases). Pre-qualification has a negative impact on accuracy

in eight out of the ten noun phases, particularly with regards to the nouns algorithm, scale,

industry, and system with omission rates of 23%, 20%, 23% and 28%, respectively. A

Wilcoxon Signed Rank test showed that the differences between students’ accuracy of articles

with simple noun phrases and the same students’ accuracy with pre-qualified noun phrases

were highly significant (P ≤ 0.01, Z=12055, n=455, p=.000).

239

Table 30: Articles omitted in twenty nominal groups (with and without pre-modification)

Nominal Group

Type

Omission

Correct

Other

% No

Response

[an] algorithm

Simple

[the] clustering algorithm

Complex

[a] pilot

Simple

[a] long haul passenger airline pilot

Complex

[a] month

Simple

for [a] third consecutive month

Complex

[a] scale

Simple

[a] Likert-type scale

Complex

[the] industry

Simple

in [the] car manufacturing industry

Complex

[the] system

Simple

[a] severity of perceived pain classification

system

Complex

[the] law

Simple

[a/the] particular law

Complex

[the] method

Simple

[the] active learning method

Complex

240

(Table continued) Nominal Group

Type

Omission

Correct

Other

% No

Response

[the] market

Simple

[the] housing market

Complex

[a] change

Simple

[a] rapid change

Complex

The effect of English writing proficiency level on the issue of omissions in article use was more

complex, as illustrated in the boxplot shown in Figure 75. In terms of simple nominal groups,

Figure 75: Boxplot of omission rates (grouped by proficiency) in complex noun phrases

241

students with higher IELTS writing scores scored better in the ten questions involving simple

noun phrases (a median of 8/10 correct) than the students with the lowest IELTS writing scores

(median 6.5/10). A Kruskal-Wallis test showed significant differences (P ≤ 0.05) between the

students’ scores with the ten words within simple nominal groups when grouped by IELTS

writing proficiency scores (H (5)=13.195, n=455, p=.022). However, when these same ten

words were placed within pre-qualified nominal groups, writing proficiency level no longer

had a significant effect. As shown in Figure 75, the students scored a median of 7/10 in all

groups and a Kruskal-Wallis test showed no significant (P ≤ 0.05) differences in the variances

(H (5)=3.346, n=455, p=.647).

7.3.3 Hypothesis 3

A student’s proficiency level on entry to the Presessional (as measured by IELTS writing

tests) will be related to their accuracy of article choice at the beginning of the programme

The Mean score in the 100-item test overall was 64.67% (with a Standard Deviation of 8.66).

As presented in Figure 76, IELTS writing scores appeared to be positively associated with test

scores. A Spearman's rho two-tailed test showed there was a moderate positive correlation

(highly significant at P ≤ 0.01) between IELTS writing band and overall scores in the 100

question quiz (r

=.306, n=455, p=.000), with students at higher proficiency levels showing

greater accuracy in the quiz. Furthermore, a Kruskal-Wallis test (P.= 0.000) showed that the

differences between the median values of the students grouped by IELTS writing test score

were significant (H (5)=25.222, P ≤ 0.01, n=455, p=.000).

242

Figure 76: The relationship between general writing ability and overall test scores (n=455,

100 question items)

7.4 Conclusion

As predicted, a testing approach has enabled focussed attention on the ‘problems’ students have

with articles rather than revealing their generally correct use as highlighted by the corpus-based

studies in earlier chapters. The mean score of 64.67% should not be interpreted as a measure

of students’ accuracy since the students faced many more complex choices than they would

normally find in academic writing. Moreover, the methodology could also be criticised as also

testing students with items of vocabulary (e.g. ‘scale’, ‘system’, ‘industry’) that the corpus-

based study had identified as challenging – in effect setting the students up for failure.

However, results of the testing approach have given some insights into the extent to which

243

different types of article errors are affected by proficiency level and some of the possible causes

of error.

As hypothesised, on entry to the university Presessional L1 Chinese students frequently over

supplied definite articles in Type 1 and Type 4 generalisations (in which L1 writers would use

Ø) at all levels of entry proficiency. The statistical tests presented in this study provide some

evidence that this tendency to oversupply definite articles in general statements and claims

about a topic were not significantly linked to their general writing ability. Further replication

of this study with greater numbers of students with IELTS grades 6.5 – 7.0 is necessary, as the

vast majority of the students in this study had IELTS 5.5 - 6.0 overall levels of English

proficiency and it is likely that the small number of higher level students skewed the results.

Nevertheless, as most students arrive on UK presesssionals with IELTS 6.0 levels of English

(and many students start UK academic programmes each September with this score), this

finding is noteworthy. Faced with unfamiliar new topics and vocabulary students appear

unaware of the nuances of register caused by overspecification. Their likely strategy of adding

a determiner ‘if in doubt’ may explain the impression reported by other researchers of an overly

oral tone in Chinese students’ L2 academic English writing (Cobb, 2003; Hinkel, 2003;

Granger, 2004; Mayor, 2006; Gilquin & Paquot, 2007; Lee & Chen 2009).

On the one hand, the quiz items used in this study were ‘cherry picked’ from previous errors

and the low scores certainly exaggerate the extent to which students make article errors in

normal writing. Moreover, it is not suggested that the omission errors in dense noun phrases or

overspecification errors affect Chinese L1 students uniquely. Indeed, other L1 background

students are likely to make such errors, as may L1 English home students albeit to a lesser

extent as they are also novice academic writers who may present a conversational tone on entry

to university but will learn to show progressively greater register awareness throughout their

244

academic programme. However, article-less L1 background students may find the issue of

using bare noun phrases in general statements (after specifying the noun phrases in detailed

examples earlier in their writing) particularly confusing. Thus, the finding does provide some

support for the argument that register awareness, which is generally agreed to be a central tenet

of EAP teaching, should be extended on Presessional programmes to include the effect of

article use. Students with article-less L1 backgrounds may arguably need extra help in building

their confidence to use Ø articles. Interestingly, these students appear to lack the confidence to

apply the identical Ø + bare noun structure in L2 English that is required in non-referential

contexts in their own L1. By making greater reference to how expert writers move from specific

examples (with a/the) to generalised statements (with bare uncountable nouns and plural count

nouns) in academic texts, students could be helped to notice in parallel both the general-

specific text structure and the pragmatic choices of article use which these moves of focus

entail.

With regard to the second hypothesis, this study provided some evidence that pre-modified

noun phrases are more strongly associated with omission errors than simple noun phrases. The

Chinese students in this study made more omission errors when the head noun was obscured

by the use of different pre-modifiers. This supports Biber & Gray’s (2010) view of academic

language posing unique challenges to learners because of its ‘compressed’ nature. Moreover,

while students’ accuracy with articles in simple noun phrases was associated with proficiency

level, no significant relationship was shown by proficiency to their use of articles in more

complex nominal groups. This finding suggests that Chinese learners have no problem with the

grammatical rules in Type 2 and Type 3 contexts when applied to simplified language (or the

prototypical examples in grammar books), but have less accuracy in authentic academic

English. In short, it appears that it is not the rules of grammar but the application of these rules

245

in densely packed noun phrases that students find challenging, particularly when academic

texts obscure grammatical choice. This would lend weight to the argument that L2 students of

English for academic purposes need help noticing these dense nominal groups in academic

English (Fang, Schleppegrell & Cox, 2006). It would appear that academic writing and its

nominalised abstractions can obscure grammar mappings for L2 English learners in the same

way as they have been shown to obscure meaning (Halliday, 1985; Halliday and Martin, 1993).

Replication studies with a far greater number of paired noun phrases are now required. It is

also acknowledged that the effect of sentence structure and transitivity was not controlled in

article quiz questions. Nevertheless, these findings taken in parallel with those of the corpus

study findings show one possible cause of article omission for Chinese students that could

usefully be drawn attention to in EAP textbooks.

Finally, this study has investigated the effect of general writing proficiency on Chinese

students’ confidence in selecting the correct article across all types of article. In terms of article

use in all contexts (including idiomatic use and proper nouns), students’ scores in the test were

highly correlated with their IELTS writing scores. On the one hand, it therefore appears that

Chinese students show greater accuracy in their decisions in tandem with their greater general

confidence in written English. However, this effect was not significant in relation to their

overspecification of articles in Type 1 and 4 contexts or omission of articles in complex noun

phrases. Given the very small number of the lowest IELTS 4.5 level and advanced IELTS 7.0

level students, these results should be treated with a degree of caution. There is a clear need for

future studies that can recruit higher numbers of the lowest and highest proficiency level

students (and thus represent the overall population of Chinese L1 students in HE).

246

The theoretical implications of these three main findings are developed in the next chapter

along with some recommendations. One final methodological conclusion suggested here is

that a testing approach has been shown to complement the findings of a corpus-based study.

Particularly with regard to oversuppliance and the challenges involved in reliably

distinguishing between redundancy (dispreferred forms) and inappropriacy (error), this

methodology has helped to investigate two hypotheses developed from the earlier corpus-based

studies. It has been argued that while only corpus-based studies can identify actual features of

student writing and demonstrate how accurate students are in different contexts, such a testing

approach can complement and provide more nuanced understanding to the hypotheses made in

studies using learner corpus methodologies.

247

8 CONCLUSIONS

8.1 Restatement of aims

This research project had four main aims. First, it aimed to investigate how EAP tutors

currently teach or correct English article errors in academic writing. Second, it sought to

examine whether pedagogical interventions have any effect on article accuracy. Third, it

focussed on the sorts of English article errors made by Chinese L1 students (with Upper-

Intermediate English L2 levels) on entry to UK universities. Finally, it aimed to identify the

reasons why some article uses challenge these learners. As stated in the introduction, a greater

knowledge of the challenges created by the English article system for Chinese learners and a

clearer understanding of whether and how accuracy can be accelerated would help EAP tutors

to integrate a focus on articles within a student centred text-based/data-driven approach and

avoid less effective practices such as overcorrection of article uses in a manner that does not

lead to greater accuracy. This concluding chapter will first summarise the findings related to

each research question, then evaluate some of the limitations and needs for future research

before ending with some practical recommendations for the EAP tutor. As the final comments

will explain, these recommendations will support the argument made in the introduction that

an occasional lexico-grammatical focus on the article system (fully integrated within a genre-

based approach) can be consistent with the emphasis on text structure, functional moves, genre

and register awareness that is typical of the text-based/data-driven approaches most advocated

in the literature.

8.2 Summary of main findings

Contrary to some EAP tutors’ assumptions demonstrated in interviews, it is necessary to stress

the extremely high accuracy that B2 level Chinese L1 students demonstrated in the majority of

authentic academic writing tasks. While this research has investigated ‘problems’ of article

248

use for Chinese learners and suggested that certain challenges faced by Chinese learners may

be influenced by their L1, the majority of the errors and variables examined in this research

have identical causes to those made by all L1 groups of L2 English learners. In fact this

research project has confirmed that most uses of articles in academic writing are affected by

variables that concern all students equally: general level of proficiency, familiarity with the

topic lexis, genre of the task and the extent to which their attention is consciously focussed on

article accuracy.

However, the findings in this research equally do not support a naturalistic assumption that

Postgraduate Chinese learners’ article errors will ‘sort themselves out’ without intervention by

the time the students need to submit their dissertation or thesis at the end of their academic

programme. The corpus-based studies and the testing methodology in this research showed that

Chinese L1 students with IELTS 6 and 6.5 levels of academic writing continue to make article

choice errors in L2 English academic writing and that these errors sometimes contribute to an

informal tone in their writing. While the evidence in Chapter 6 supports the view that it is well

within these students’ ability to use articles more accurately in simpler descriptive tasks (such

as the case study genre) when they are fully familiar with new vocabulary, Postgraduate level

students will be required to use sources and levels of abstraction beyond that of case study

reports later in their studies and it is likely that their article issues will surface again, as they

return to unfamiliar topics with less specific referents.

On the one hand, students’ more in depth research and background reading affords greater lexis

familiarisation and knowledge of colligations, which will contribute to growing article

accuracy during their studies. On the other hand, students could be helped by leaving the

presessional programme with the autonomous learning skills that enable them to acquire,

record and learn new vocabulary in their discipline with attention to register awareness and

249

how a topic’s terms are generalised. Through EAP reading paradigms, students might also

usefully be taught how to notice the pragmatic effect of articles. If students can leave

presessional programmes with the independent confidence in proofreading their work with

attention to head nouns in dense nominal groups, perhaps distant from their determiners, their

EAP tutors will have contributed greatly to accelerating students’ accuracy in academic

writing.

8.2.1 Research Question 1

To what extent do EAP teachers in Higher Education programmes in the UK explicitly teach

or correct English article use?

The answer to this question that emerges from the research reported in this thesis is that most

EAP tutors teach and correct article use in L2 academic writing while only a small minority of

tutors totally ignore the issue. While explicit teaching in a decontextualised ‘rules’ based

approach is less common, many tutors use consciousness- raising techniques in reading classes

or frame attention to articles within proofreading skills sessions fully compliant with the

authentic text-based approach in EAP. Moreover, many tutors use student-centred approaches

to error correction (guided self-correction and peer marking) that again appear informed by

EAP best practices. Unfortunately, however, many teachers also overcorrect article use in a

more traditional manner that is arguably less consistent with contemporary recommended EAP

approaches, either because they perceive students expect such corrections or because they

assume that corrections lead to learning.

Chapter 4 characterised five types of strategic response to article errors among EAP tutors by

the presentation of five ‘teacher types’: 1) tutors who teach English articles within EAP text-

based best practices, 2) the optimistic ‘rule giving’ teacher, 3) the ‘proofreader’ (error hunter)

250

4) the reluctant corrector and 5) the principled ignorer. Most teachers surveyed in this study

reported data-driven approaches, noticing techniques and text-based approaches within the first

‘EAP paradigm’ category. However, a smaller number of tutors continue to use prototypical

grammar exercises that are most consistent with decontextualised approaches within an EFL

paradigm. These tutors appear to believe that Chinese students do not understand the concept

of article use and, assumedly, that they can succeed (where others have been unsuccessful) by

providing what they believe to be a better/simpler explanation of article rules. ‘Reluctant’ error

correctors do not wish to focus on article use, due to their doubts about its usefulness or their

own lack of confidence in the article system, but feel obliged to do so in response to student

expectations. A small group of ‘principled ignorers’ appear to believe that any focus on article

accuracy is inconsistent with the best EAP practices or progressive World English arguments.

However, the survey in Chapter 4 suggests that only a small minority of EAP tutors have such

principled objections.

The findings in Chapter 4 pertaining to error correction showed that EAP tutors’ stated beliefs

about EAP principles and their reported best practices in teaching are sometimes inconsistent

with their indiscriminate correction of article errors in academic writing. Some EAP tutors see

themselves as proofreaders whose main role is to correct every error they see on the assumption

that students learn from frequent correction. In fact, the most common ‘teacher type’ identified

in the survey was a combination of Type 1 and Type 3 (teaching within the ‘EAP paradigm’,

but correcting errors with less regard to learner autonomy). That is to say, while many tutors

report many modern data-driven methods for whole class teaching and more student-centred

approaches to error correction (peer marking), many appear to supplement best practices with

the more traditional and questionable ‘mark all errors we see’ approach in their own written

feedback. This confirms earlier research that reports a tension between EAP tutors stated

251

positions/beliefs and their actual practices (Etheringon & Burgess, 2002; Barnard & Scampton,

2008).

The extent to which the different approaches taken by EAP teachers are validated in light of

this research is very much arguable. Although the studies presented here showed that teaching

and error-correction had a significant effect on article accuracy of a few types of article use,

this improvement was not sustained six weeks after receiving the corrections. On the one hand,

this may be less relevant to Presessional EAP teachers and their students who are facing high

stakes assessments at the end of the course (for which temporary improvement in accuracy is

arguably a valid focus). Yet some tutors may be persuaded to abandon the implicit assumption

that correction leads to sustained learning and devote the considerable time required for such

corrections by both teachers and students to more constructive parts of the feedback process.

In terms of the assumptions made by ‘ignoring’ teachers (that articles are eventually acquired

more accurately without intervention), this research has confirmed that most article uses are

strongly correlated with general writing proficiency. Nevertheless, certain errors

(oversuppliance in generalised statements and omission in pre-qualified noun phrases) were

less connected to proficiency. Thus, these assumptions of students improving naturalistically

without help are also not supported.

8.2.2 Research Question 2

What are the linguistic contexts of obligatory article use in which Chinese L1 students at

Upper-Intermediate levels of English L2 are most likely to make errors?

This question was addressed both in terms of raw frequency and in terms of the ratio of

correct/incorrect occurrences of articles in different contexts. The results of all investigations

252

presented here have shown that Chinese students achieved varying degrees of accuracy in

different contexts of English article use.

In terms of raw frequency, the errors in article use by Chinese L1 students were frequently

made by omitting definite articles or (next most frequently) by oversupplying definite articles

when inappropriate, confirming previous research in this area (Chuang & Nesi, 2006; Lee &

Chen, 2009). In terms of a/an articles, the corpus-based evidence in this research suggested

that omission was the more likely error, not redundancy. Finally, in contrast to the claim made

by Chang (2001), this research has shown misuse of ‘the for a/an’ or vice versa to be extremely

rare. However, when frequency of use is taken into account (and errors are expressed as a ratio

of correct to incorrect use), this research showed that Chinese students were marginally least

accurate with a/an articles, followed by the and lastly (most accurate) in Ø article contexts.

In terms of the Bickerton Framework used throughout this thesis, when writing short essays,

the corpus study presented in Chapter 6 shows evidence that Chinese students are most

challenged by the following article contexts (in order of difficulty): IDA<4IA<1IA<5DA=3ZA

<1ZA<2DA<3IA<4ZA<5ZA (see Sections 2.2.1 and 6.3 for explanation of Framework.

Interestingly, with the exception of a/an use in non-referential contexts (4IA), an inverse

relationship between accuracy and frequency of occurrence (Type 4>2>3>5>) can be seen in

this pattern, reinforcing previous research that has found the most frequently occurring

language items are the most quickly acquired (Larsen‐Freeman, 1976). A strong case could

therefore be made for grammar and EAP materials writers to pay greater attention to such rare

uses, to help learners notice them in their reading and less attention to the anaphoric references

so overrepresented in grammar (assuming simpler uses of article will be acquired naturally by

learners).

253

In Type 1 (generic) contexts Chinese learners at Upper-Intermediate English L2 level

frequently omit and oversupply articles and display lower accuracy in generic definite and

indefinite contexts than when discussing more specific Type 2 or Type 3 referents. This

broadly confirms previous research findings (Master, 1995; Snape, Leung & Ting, 2006; Ionin

et al., 2011; Crosthwaite, 2016a). While generic use (as defined in this thesis) is far less

frequent than other uses, in both Chinese L1 students’ and L1 tutors’ essays (6.4-11.7% of

articles) their function in EAP is pertinent, in as much as contribute to defining terms and

providing the background for topics. Given the complexity and difficulty of teaching generic

article use, some recommendations are made in Section 8.4.

It was found that indefinite articles in non-referential (4IA) contexts were the second most

inaccurately used context (a TLU of 0.71). More generally, Chapter 6 reported that indefinite

articles overall were the least accurately used article by the participants (in ratio terms). This

finding is contrary to that of Díez-Bedmar & Papp (2008) who found that Chinese students

were more accurate with indefinite than definite articles, but supports Master’s (1987) finding

that a/an mastery is acquired only at the latest stages of L2 development. The divergence in

this finding underlines the difficulty of comparing studies that define proficiency levels

differently (Díez-Bedmar, 2015). The relevance of this finding is that Type 4 contexts were

shown to be the single most frequent occurring article use context in the essays of both Chinese

L1 students and English L2 tutors. Chinese students’ relative inaccuracy with indefinite

articles, in addition to their frequent overspecification of definite articles in 4ZA contexts is

thus found in many of the most common language functions in academic essays: generalising,

discussing future scenarios and conditional clauses in non-referential contexts.

Leaving the Bickerton Framework analysis, one main finding of this research was that the

dense nominalisation that often characterises academic English (Halliday & Martin, 1993,

254

Biber & Gray, 2010, Bennett, 2011) was also associated with lower article accuracy among

Chinese L1 learners. Students made significantly more article omission errors in pre-qualified

nominal groups than simple noun phrases. Moreover, while there was a significant effect of

proficiency upon the number of article omission in simple nouns phrases, there was no

significant effect on the omission rate in complex noun phrases. As shall be explained in

Section 8.4, this finding has implications for the way students are taught proofreading skills.

8.2.3 Research Question 3

To what extent does the Chinese students’ L1 background affect (1) TLU and (2) their types

of article error?

It is often assumed that Chinese students (and students from other Asian L1 backgrounds) will

find articles particularly difficult to master because their first language does not have an article

system at all. However, conflicting findings in this research respectively support and challenge

the L1 transfer effect hypothesis.

There is some support for the assumption of L1 transfer from the students’ omissions of articles

in many generic article contexts. Students’ lower accuracy with generic definite articles (0.67)

than Ø generic context articles (0.86) in the three essays (Chapter 6) provides evidence for the

L1 transfer effect hypothesis. The corpus-based results in this research demonstrated that the

B2 level Chinese students had far greater accuracy with Type 2 definite and Type 3 indefinite

articles than Type 1 generic definite articles. Mandarin does have equivalent classifiers and

demonstratives – generic references are normally achieved through bare nominal singular

forms without measure words (Li & Thompson, 1981). Therefore, the hypothesis that this

difference leads to a lack of positive transfer effects for definite generics while the equivalents

in 2DA, 3IA and 3ZA forms can create positive transfer effects (Snape, Leung & Ting, 2006;

255

Díez-Bedmar & Papp, 2008; Crosthwaite, 2016a) appears convincing. As Chinese has certain

demonstratives and classifiers that can be optionally used to pseudo grammatically determine

specific referents, this hypothesis claims that the use of English articles with specific referents

is less completely new to the students’ L1 parameters.

Evidence of negative transfer can also be inferred from the students’ relative inaccuracy in non-

referential Type 4 indefinite contexts requiring ‘a/an’. Chinese does not allow for the use of

any classifiers in such contexts (Li & Thompson, 1981) and their greater issues in non-

referential contexts could be explained by L1 interference.

On the other hand, the students examined here also oversupplied the definite article in many

contexts in which Chinese generally uses a bare nominal, somewhat contradicting the idea that

Chinese students’ errors closely follow the article-less features of their L1. Contrary to using

the same bare nominal form in 4ZA contexts in a similar way to their L1, students were shown

to add redundant definite articles. Chinese students’ evidenced lack of confidence in Ø article

Type 4 contexts thus shows the complexity of L1 transfer effects. Interviews suggested that

Chinese students believed they have a choice of Ø or the in such generalised statements and

that the definite article has an emphasising function, somewhat reflecting the fact that

classifier/demonstrative use in Chinese is a choice rather than an obligation (Li & Thompson,

1981). It could be argued that article-less L1 background students lack pragmatic awareness

of article use (Díez-Bedmar & Papp, 2008). However, a simpler conclusion could be that

Chinese students frequently face challenging choices of whether to treat new vocabulary terms

as countable or uncountable and, when in doubt, often ‘gamble’ on the word being a countable

noun requiring a determiner. The hypothesis that students gamble on the language form being

the opposite to their L1 would be an interesting area for future research.

256

In terms of gravity of error, Chinese students’ overspecification of definite articles in Ø (4ZA

and 3ZA) context has been argued in this thesis to occasionally cause a spoken tone

inappropriate for academic essays. In interviews, Chinese students showed that they

sometimes failed to choose Ø + uncountable noun structures when they mistook the noun for

a single countable noun and Ø + plural structures with a misconception that they could add

emphasis through the use of definite articles. On the one hand, a strong argument can be made

that students of all L1 backgrounds face such errors given the rather arbitrary nature in which

different articles are permissible. Indeed, all novice writers (including L1 speakers) are likely

to develop greater confidence in register nuances as they become familiar with the discipline

specific lexis. However, a convincing argument can be made that article-less L1 background

students need greater help in building confidence of using Ø when generalising. The students

interviewed in this study reported a “nothing to lose” gambling strategy that had not been

expected at students at B2 levels of writing. By inference, EAP materials that juxtapose how

new lexis in their disciplines is determined in specific contexts but unspecified in general

statements (as outlined in Section 8.4) could be valuable.

In sum, as shown by the analysis based on the Bickerton Framework and again evidenced in

students’ greater accuracy in case studies than essays, the more specific the referent (in equal

contexts of lexical density), the more confidence Chinese students seemed to show in article

choice. This greater confidence in specific contexts may be somewhat influenced by positive

L1 transfer effects. However, this thesis concludes that L1 is a minor contributor to most

errors and, counter-intuitively, students do not use positive transfer effects (bare nominal

forms) in many expected contexts. Questioned about errors made in their essays, most

interviewed students appeared to make errors due to unfamiliarity with the lexis, confusion

caused by the dense nature of academic language and the grammatical characteristics of the

257

head noun. As such challenges are shared by all L1 background students, this thesis supports

only a weaker form of the contrastive language hypothesis (L1 contributes to some forms of

error but is not the main effect).

Research Question 4

Does the explicit teaching or correcting of English article usage lead to more target-like use

of articles by Chinese L1 university students whose L2 English displays inaccuracy?

On the one hand, by the use of a control group that had no extra attention focussed upon articles,

Chapter 5’s findings showed that teaching article use or correcting article use can have a short

term significant effect of increasing students’ accuracy of article use in some contexts. In

particular, improvements in definite article accuracy (and the reduction of overspecification

errors) were significant when the teachers focussed students’ attention on article use. These

findings therefore to some extent support the claim that explicit teaching can improve students’

accuracy (Master, 1990; Master, 1995; Master, 1997; Bitchener & Knoch, 2008) and that error

correction reduces inaccuracy (Ferris, 1999; Chandler, 2003; Bitchener, Young & Cameron,

2005; Ferris & Kurzer, 2019). However, the improvements were not sustained six weeks after

the extra attention had ceased. This adds weight to the idea that teaching and article correction

improve article accuracy only for as long as consciousness is raised (Master, 1997).

For teachers who employ the ‘scatter gun’ approach to article error correction, especially on an

assumption of sustained learner improvement, rather than a temporary impact on surface errors,

this finding may have some relevance. In fact, many reasons can be given for discouraging

teachers from the excessive written correction of all errors. Over and above the temporary

effect of error correction, Chapters 4 and 5 have highlighted the challenge of the reliable

identification of article errors and there is likely to be a strong risk of unhelpful miss-

258

corrections if tutors try to mark work hastily since accurate error correction is extremely labour

-intensive. Several students also reported that they felt discouraged and embarrassed by article

corrections and this attitude is likely to have been under-reported. Occasional focussed error

correction, especially in a peer feedback context as highlighted in the recommendations, may

be justifiable. However, given the equal impact upon article accuracy of teaching, a strong

argument is made here for the preference of simple attention focussing activities (e.g. teaching

of common patterns or highlighting of examples of frequent error types on the board).

8.3 Limitations and future research

The variable most difficult to control throughout this research has been the important influence

of topic and vocabulary familiarity upon article use. The effect of task prompt was admittedly

underestimated at the study’s design stage and has since been shown to massively impact on

measures of article accuracy (Díez-Bedmar, 2015; Crosthwaite, 2016a, 2016b, 2017). It is

necessary to stress that accuracy in article use will be mainly decided by students’ familiarity

with the topic, related lexis and the nature of the task. Thus particularly Chapter 6’s findings

must be treated with a degree of caution because it is possible that different TLU rates would

have been found with different essay prompts. This effect of task prompt was again highlighted

by the higher accuracy shown by the students in the case study (discussed in Section 6.4.6 ).

Indeed, the choice of text topic was also acknowledged to have affected the accuracy levels

reported in the testing approach (which is why the TLUs for different contexts were not

compared with the corpus results). For this reason, these findings are concluded to have their

greatest merit in showing the relative accuracy levels between different contexts in the same

topic prompt. Future research on the longitudinal progress of students’ accuracy of articles

might best maintain focus on the same topic, perhaps by analysing students’ drafting of an

extended piece of work over time.

259

The strength of the corpus-led research was argued to be its use of authentic academic task

types. Nevertheless, the essay assignments required far fewer words than the assignments faced

by most Postgraduate students in their main programmes. Future research could usefully look

at student accuracy in authentic Postgraduate dissertations since comparisons of different

chapters could provide insights into relative accuracy in different article contexts (i.e. moving

from general to specific focus).

A final limitation of the corpus-based research studies was also acknowledged to be the small

number of participants. On the one hand, these limitations were somewhat addressed by the

testing approach with 455 Chinese students, with the two novel findings of the effect of pre-

qualification and overspecification being tested with a representative sample of students within

which the effect of proficiency could be framed. On the other hand, the very small number of

C1 level students (IELTS 7.0) in this study means that there is a need for future research to

confirm the findings regarding development between IELTS 6.00 – 7.00 levels and investigate

later IELTS 7.5+ levels of proficiency.

8.4 Recommendations for EAP instructors

The first advice to be offered to EAP tutors is to more clearly understand the possible strengths

and weaknesses of their Chinese L1 students at 5.5, 6.0 and 6.5 levels of English proficiency,

as examined in this thesis. These students were found to be more accurate than inaccurate and

were rarely observed to have any conceptual deficit with regards to the English article system.

To the extent that these students do make errors, it has been shown in this thesis that most of

these are developmental: students improve accuracy as their vocabulary and proficiency levels

increase. They do however face three further challenges that do not seem associated with

general English development. Firstly, in non-specific contexts (such as generic uses or

generalisations, unreal or conditional clauses) when using less familiar lexis many Chinese

260

students display a ‘nothing to lose’ approach to definite articles, adding them when

inappropriate and occasionally causing a conversational tone in their writing. This oversupply

was observed even among students with IELTS 6.5 writing ability (and in the very small

sample of IELTS 7.0 students participating in this study). Secondly, in densely packed noun

phrases students are more likely to become confused with the head noun and its grammatical

properties and omit both definite and indefinite articles when required. Thirdly, Chinese L1

students show the most inaccuracy in some of the rarest uses of articles, for example within

generic uses within definitions.

With regard to the issue of overspecification of definite articles, two teaching approaches are

recommended which, figuratively speaking, focus students’ attention on articles as a ‘side dish’

rather than ‘the main course’ of a lesson. Firstly, effective noticing of article use can regularly

be integrated into the EAP teaching of text structure and register awareness. Previous research

has shown that register awareness can be significantly improved in EAP courses (Crosthwaite,

2016b) and occasional focus on article use could form part of such genre and register

awareness. For example, the structural shape of research projects is often visualised in EAP

for learners with the hourglass shape (Swales and Feak, 2004), as illustrated in Figure 77. In

sessions dealing with this aspect of structuring academic writing, the way in which most count

nouns (e.g. system) are used in a Ø +plural noun construction in an Introduction/Conclusion

can be juxtaposed with the more specific references in the main Methods or Results. Of equal

importance, tutors could present lexis along with any divergent patterns of use related to the

general-specific cline (e.g. ‘The elderly’ or ‘policy’) as shown when presenting or revising new

lexis. By inference, article use can be seen to be most teachable in Presessionals taught in

departmental disciplines.

261

GENERAL

SPECIFIC

GENERAL

Research project

Introduction

Literature Review

Methodology

Results

Discussion

Conclusion

Figure 77: Mapping specific-general functions of lexis to research project structure

By preference, it is recommended that students be asked to notice article form juxtapositions

through teaching, rather than via correction for reasons already discussed (the dangers of

inaccurate corrections, time demands and the temporary effect). Moreover, whenever

possible, students should be focussed on authentic texts from their own discipline or their own

academic writing. An example is shown in Figure 78 of the results that one Chinese student

(in the researcher’s insessional English class) found by analysing his own PhD thesis.

Policy

A policy

Policy

The elderly

A 75 year old

woman

Systems

A system

The elderly

262

When the AC/DC converter was invented in 1905…. (Introduction, generic use)

New generation [XY] AC-DC converters have reduced… (Background chapter, generic use)

The experiment first tested the [XY] converter in stressed conditions. The converters were then

checked for damage…(Methodology, referential use)

The results show that [XY] converter (2) was … (Findings, referential use)

In conclusion, further research is needed into [XY] converters’ reaction to.…(Conclusion,

generic use)

Figure 78: Deduced uses by Chinese L1 Engineering student [anonymised]

A second approach that also suits the aforementioned issue of article omissions in densely

packed noun phrases lies in the use of activities that develop students’ careful consideration of

head nouns within densely packed nominal groups. Given the need to develop students’ own

autonomous proofreading skills, asking students to check each other’s work (after a brief

presentation of typical errors related to omissions with pre-qualification or overspecification)

is far more likely to raise their future attention towards article errors (with the proviso that peer

corrections are checked by the tutor). While improvements may be temporary in the first

instance, once this focus has become an independent learning habit, with practice students can

use such a focus throughout their academic careers.

8.5 Final comments

This research has shown that L1 Chinese students entering university at Upper-Intermediate

levels of English (the B2 level with which most students arrive) show near target-like accuracy

when writing about specific referents when they are familiar with the language in simpler

genres of writing. While article errors are the most frequent grammatical error made by Chinese

263

learners, this is mainly due to the fact that they are the most commonly occurring word in

academic English. However, regardless of level, Chinese B2 level students are, as seen in the

results presented in this thesis, less accurate with less familiar language in dense pre-qualified

noun phrases. Moreover, these learners are less accurate in generic uses of articles or in texts

requiring more critical discussion about unspecific referents. At B2 levels, this thesis has

shown that Chinese L1 students frequently overspecify referents when they should instead be

generalising and speculating about more abstract concepts with bare noun phrases in plural or

mass noun forms. This can add an unintended informal register to their writing which is

inappropriate for some academic genres. The two most novel findings in this research are that

neither omission of errors in dense nominalisations nor overspecificaton in unspecific contexts

are mastered by higher B2 level Chinese L1 learners and that proficiency level may have little

effect until much higher levels. While errors may ‘look after themselves’ over a lifetime, they

appear not to be developmental errors that disappear by IELTS level 6.5.

Some of the anecdotal observations that motivated this research were confirmed. Many EAP

tutors do appear to be overcorrecting article misuse indiscriminately. Occasional error

correction of articles may be appropriate for justifiable proofreading purposes, to quality

control peer correction, to model the proofreading process for students or highlight a function

of articles related to lesson aims (for example, pointing out overspecification in

generalisations). However, this thesis has presented EAP tutors with several arguments against

the overcorrection of article errors. Firstly, it is extremely time-consuming for tutors and

learners. Secondly, it is very easy to miss-correct articles if a text is read hastily. Lastly, the

evidence offered in this thesis suggests that constant correction has little sustained effect.

EAP tutors’ continued correction practices, even against their better judgement, is likely to

arise from the tensions they feel between theory and students’ expectations. It is clear that many

264

EAP tutors feel under some pressure to provide corrections regardless of received wisdom.

EAP Presessionals have been marketed to students as a means not only to learn about academic

genres, study skills and the process of academic writing, but also as programmes which develop

their language proficiency and accuracy. While most errors have a minor effect on

comprehension, articles can negatively affect the register of English and perceptions of a

writer’s meaning. In turn, as speculated by one Chinese tutor in Chapter 4, it is likely that

Chinese L1 students, who come from a background in which a focus on grammar and accuracy

takes up the greater part of college English preparation classes, are confused by the reduced

focus on form in EAP. In the knowledge that they will have to write in challenging academic

genres within months of leaving a presessional programme Chinese students either need more

information about how EAP pedagogy makes assumptions of naturalistic acquisition or, as

this thesis argues, greater occasional focus on form within their EAP syllabus.

This thesis has not argued for the appeasement of student expectations through a return to

decontextualised grammar type activities. Students frequently ask for such activities, but this

thesis has shown that they have little sustained impact on written accuracy. The added ‘English

language content’ bolted on to the EAP syllabus on many UK Presessional programmes in

recent years as a solution to students’ perceived accuracy problems is equally not an ideal

solution. Instead, a strong case is made for fully integrating an occasional focus on lexico-

grammatical forms within EAP text-based best practices. It has been shown that such a focus

can be contextualised within a genre-approach to EAP. While no evidence has been offered in

this thesis to support the efficacy of such pedagogy or for self-correction and peer correction,

such practices appear more consistent with the EAP literature, pose less risks of discouraging

learners, are more time-efficient and appear to give students the skills they will need in their

future studies without the presence of the tutor, thereby promoting learner autonomy.

265

In terms of error gravity, although most article errors are trivial surface errors that have little

impact on academic writing, this thesis has shown that article omission errors on occasion can

affect communication. Furthermore, some students appear to be oversupplying definite articles

in essay conclusions and generalised claims to the extent that the reader perceives an informal

tone to the register of writing. While all writers make mistakes in writing, the Chinese students

interviewed for this research showed they were often confused about article use in unspecific

contexts. Further research is required to investigate why Chinese students are sometimes

hesitant about applying the Ø determiner to bare noun plurals, even though their L1 similarly

requires bare noun phrases. It was speculated in this thesis that some Chinese students have

decided to add articles whenever in doubt, and future research would usefully examine whether

this strategy remains ‘fossilised’ at IELTS 7.0 - 8.0 levels of proficiency. The fact that Chinese

learners made fewer oversuppliance errors when extra attention was focussed on this problem

in addition to the finding that this error persisted at IELTS 6.5 and 7.0 writing levels lead to

the recommendation that EAP tutors should teach students the proofreading skills to check for

this informal style they may unconsciously add to their writing. Rather than overcorrecting

these errors themselves, tutors should be training students to have learner autonomy and self-

correct for these problems when they leave the EAP course.

The findings in this study have a number of further practical implications. It is important to

train new entrants in the EAP field in such student-centred approaches to teaching proofreading

and article accuracy rather than insisting that EAP never involves a focus on form. The

argument for such an integration of a lexico-grammatical focus upon determiners within the

‘EAP paradigm’ of text structure, functional moves and register awareness is also an argument

for teaching EAP students in discipline-related classes (looking at the article system mappings

unique to various topics and genres). English article teaching would therefore ideally sit within

266

a syllabus that intertwined discipline-specific registers and other lexico-grammatical aspects

(e.g. verb tense, transitivity, hedging language). Fundamentally, given that most EAP teachers

continue to focus on language form and correct errors regardless of advice from the doyens of

EAP, a strong case can be made for greater discussion about how EAP tutors can and should

focus on the grammar of such small words if they do not feel they can ignore them.

267

REFERENCES

Alexander, O., Argent, S. and Spencer, J. (2008) EAP essentials: A teacher’s guide to principles and

practice. Oxford: Garnet.

Alexander, O., Bell, D., Cardew, S., King, J., Pallant, A., Scott, M., Thomas, D., Ward and Goodbody,

M. (2008) Competency framework for teachers of English for academic purposes. Available

at:ahttps://www.baleap.org/wp-content/uploads/2016/04/teap-competency-framework.pdf

(Accessed: 1 April 2016).

Almén, D. and Nordin, M. (2011) Long term unemployment and violent crimes-using post-2000 data

to reinvestigate the relationship between unemployment and crime. IDEAS Working Paper

Series from REPEC. Department of Economics, Lund University. Available at:

https://project.nek.lu.se/publications/workpap/papers/WP11_34.pdf (Accessed: 4 January

2015).

Barnard, R. and Scampton, D. (2008) ‘Teaching grammar: A survey of EAP teachers in New Zealand’,

New Zealand Studies in Applied Linguistics, 14(2), pp. 59–82.

Bennett, G. R. (2011) A staircase model for teaching grammar for EAP writing in the IEP: freshman

composition and the noun phrase. Unpublished PhD Thesis. University of Birmingham.

Available at: http://search.proquest.com/docview/1033192284?accountid=14542 (Accessed 2

December 2015).

Berry, R. (1991) ‘Re-articulating the articles’, ELT Journal, 45(3), pp. 252–259.

Biber, D. and Gray, B. (2010) ‘Challenging stereotypes about academic writing: Complexity,

elaboration, explicitness’, Journal of English for Academic Purposes, 9(1), pp. 2–20.

Biber, D., Johansson, S., Leech, G., Conrad, S., Finegan, E. (1999) ‘Longman grammar of spoken and

written English’. Tesol Quarterly, 34, p. 1203.

Biber, D., Leech, G. N. and Conrad, S. (2002) Longman student grammar of spoken and written

English. Harlow: Longman.

Bickerton, D. (1981) Roots of language. Ann Arbor, MI: Karoma Press.

Bitchener, J. and Knoch, U. (2008) ‘The value of written corrective feedback for migrant and

international students’, Language Teaching Research, 12(3), pp. 409–431.

Bitchener, J. and Knoch, U. (2009) ‘The contribution of written corrective feedback to language

development: a ten month investigation.’ Applied Linguistics, 31(2), pp. 193–214.

Bitchener, J., Young, S. and Cameron, D. (2005) ‘The effect of different types of corrective feedback

on ESL student writing’, Journal of Second Language Writing, 14(3), pp. 191–205.

Borg, S. (2003) ‘Teacher cognition in language teaching: A review of research on what language

teachers think, know, believe, and do’, Language Teaching, 36(2), pp. 81–109.

268

Brinton, D. M. and Holten, C. A. (2001) ‘Does the emperor have no clothes? A re-examination of

grammar in content-based instruction’. In J. Flowerdew and M. Peacock, (Eds.) Research

Perspectives on English for Academic Purposes. Cambridge: Cambridge University Press, pp.

239–251.

British Council (2012) IELTS guide for teachers. Available at: https://www.britishcouncil.it/sites/

default/files/ielts_guide_for_teachers_italy.pdf (Accessed: 1 September 2016).

Brown, J. D. (2014) ‘The future of world Englishes in language testing’, Language Assessment

Quarterly, 11(1), pp. 5–26.

Brown, R. (1973) A first Language: The early stages. Cambridge MA: Harvard University Press.

Burt, M. K. and Kiparsky, C. (1974) ‘Global and local mistakes’. In Schumann, J. and Stenson, N.

(Eds) New frontiers in second language learning. Massachusetts: Newbury House Publishers,

pp. 71–80.

Butler, Y. G. (2002) ‘Second language learners’ theories on the use of English articles: An analysis of

the metalinguistic’, Studies in Second Language Acquisition, 24(3), pp. 451–480.

Bybee, J., Perkins, R. and Pagliuca, W. (1998) The evolution of grammar. Darwinian perspectives on

the origins of language. Philadelphia: Chicago University Press.

Byrd, P. and Coxhead, A. (2010) ‘On the other hand: Lexical bundles in academic writing and in the

teaching of EAP’, University of Sydney Papers in TESOL, 5(5), pp. 31–64.

Carrell, P. L. and Carson, J. G. (1997) ‘Extensive and intensive reading in an EAP setting’, English

for Specific Purposes, 16(1), pp. 47–60.

Chandler, J. (2003) ‘The efficacy of various kinds of error feedback for improvement in the accuracy

and fluency of L2 student writing’, Journal of Second Language Writing, 12(3), pp. 267–296.

Chang, J. (2001) ‘Chinese speakers’. In B. Smith and M. Swan (Eds.) Learner English: A teacher’s

guide to interference and other problems. Cambridge: Cambridge Univeristy Press, pp. 310–

316.

Chen, P. (2004) ‘Identifiability and definiteness in Chinese’, Linguistics, 42(6), pp. 1129–1184.

Chen, P. (2014) ‘The comparison of intermediate and advanced Chinese learners’ use of English

adverbial connectors in academic writing’, International Journal on Studies in English

Language and Literature, 2(8), pp. 85–92.

Chesterman, A. (1991) On definiteness: a study with special reference to English and Finnish.

Cambridge: Cambridge University Press.

Chuang, F. and Nesi, H. (2006) ‘An analysis of formal errors in a corpus of L2 English produced by

Chinese students’, Corpora, 1(2), pp. 251–271.

Cobb, T. (2003) ‘Analyzing late interlanguage with learner corpora: Quebec replications of three

European studies’, Canadian Modern Language Review, 59(3), pp. 393–424.

269

Conrad, S. and Biber, D. (2001) 'Multi-dimensional methodology and the dimensions of register

variation in English'. In Conrad and Biber (Eds) pp. 13-42. Variation in English: multi-

dimensional studies. Harlow: Pearson.

Corder, S. P. (1967) ‘The significance of learners’ errors’, International Review of Applied Linguistics,

5(4), pp. 161–169.

Cortazzi, M. and Jin, L. (1996) 'Cultures of learning: Language classrooms in China'. In N.H. Coleman

(Ed.) Society and the language classroom. Cambridge: Cambridge University Press, pp. 169–

206.

Coxhead, A. (2016) ‘Acquiring academic and disciplinary vocabulary’. In K. Hyland, P. Shaw (Eds.)

The Routledge handbook of English for academic purposes. London: Routledge, pp. 201–214.

Crosthwaite, P. (2012) ‘An error analysis of L2 English discourse reference through learner corpora

analysis’, Linguistic Research, 30(2), pp. 163–193.

Crosthwaite, P. (2016a) ‘L2 English article use by L1 speakers of article-less languages’, International

Journal of Learner Corpus Research, 2(1), pp. 69–101.

Crosthwaite, P. (2016b) ‘A longitudinal multidimensional analysis of EAP writing: Determining EAP

course effectiveness’, Journal of English for Academic Purposes, Volume 22, pp. 166–178.

Crosthwaite, P. (2017) ‘Does EAP writing instruction reduce L2 errors? Evidence from a longitudinal

corpus of L2 EAP essays and reports’, International Review of Applied Linguistics in Language

Teaching, 56(3), pp. 315–343.

Cziko, A. (1986) ‘Testing the language bioprogram hypothesis: a review of children’s acquisition of

articles’, Language, 62(4), pp. 878–898.

Dale, R., Anisimoff, I., & Narroway, G. (2012) 'Hoo 2012: A report on the preposition and determiner

error correction shared task'. In Proceedings of the seventh workshop on innovative use of NLP

for building educational applications, pp. 54–62. Available at: https://dl.acm.org/

ft_gateway.cfm?ftid=1303967&id=2390390 (Accessed 14 November 2015).

Davies, A. (2003) The native speaker: Myth and reality. Bristol: Multilingual Matters.

Díez-Bedmar, M. (2015) ‘Article use and criterial features in Spanish EFL writing’. In M. Callies &

S. Götz (Eds.), Learner corpora in language testing and assessment. Amsterdam: John

Benjamins, pp.163-190.

Díez-Bedmar, M. and Papp, S. (2008) ‘The use of the English article system by Chinese and Spanish

learners’. In G. Gilquin, M.B. Díez-Bedmar and S. Papp (Eds.) Linking up contrastive and

learner corpus research. Amsterdam: Rodopi, pp. 147-175.

Dudley-Evans, T. and St John, M. J. (1998) Developments in English for specific purposes.

Cambridge: Cambridge University Press.

Dulay, H., Burt, M. and Krashen, S. (1982) Language two. New York: Oxford University Press.

270

Ede, L. and Lunsford, A. (1984) ‘Audience addressed/audience invoked: the role of audience in

composition theory and pedagogy’, College Composition and Communication, 35(2), p. 155.

Ekiert, M. (2004) ‘Acquisition of the English article system by speakers of Polish in ESL and EFL

settings’, Teachers College Columbia University Working Papers in TESOL & Applied

Linguistics, 4(1), pp. 1–23.

Ellis, R. (1994) The study of second language acquisition. Oxford: Oxford University Press.

Ellis, R. and Barkhuizen, G. (2005) Analysing learner language. Oxford: Oxford University Press.

Ellis, R. (2006) ‘Current issues in the teaching of grammar: an SLA perspective’, TESOL Quarterly,

40(1), pp. 83–107.

Ellis, N. C. and Larsen-Freeman, D. (2009) ‘Constructing a second language: Analyses and

computational simulations of the emergence of linguistic constructions from usage’, Language

Learning, 59 (1), pp. 90–125.

Etherington, S. and Burgess, J. (2002) ‘Focus on grammatical form: explicit or implicit?’, System,

30(4), pp. 433–458.

Fang, Z., Schleppegrell, M. J. and Cox, B. E. (2006) ‘Understanding the language demands of

schooling: Nouns in academic registers’, Journal of Literacy Research, 38(3), pp. 247–273.

Feise, R. J. (2002) 'Do multiple outcome measures require p-value adjustment?'. BMC Medical

Research Methodology 2(1), p.8.

Ferris, D. (1999) ‘The case for grammar correction in L2 writing classes: A response to Truscott

(1996)’, Journal of Second Language Writing, 8(1), pp. 1–11.

Ferris, D. (2006) ‘Does error feedback help student writers? New evidence on the short- and long-term

effects of written error correction’. In K. Hyland and F. Hyland (Eds.) Feedback in second

language writing: Contexts and issues. Cambridge: Cambridge University Press, pp. 81–104.

Ferris, D. and Kurzer, K. (2019) ‘Does error feedback help L2 writers?’. In K. Hyland, K. and F.

Hyland, (Eds.) Feedback in second language writing: Contexts and issues (2

Ed.).

Cambridge: Cambridge University Press., pp. 106–124.

Fleiss, J.L. (1981) Statisical methods for rates and proportions. 2nd ed. New York: Wiley.

Flowerdew, J. (1993) ‘An educational, or process approach to the teaching of professional genres’,

ELT Journal, 47(4), pp. 305–306.

Flowerdew, J. (2016) ‘English for specific academic purposes (ESAP) writing: Making the case’,

Writing and Pedagogy, 8, pp. 5–32.

Flowerdew, J. and Peacock, M. (2001) ‘Issues in EAP: a preliminary perspective’. In J. Flowerdew

and M. Peacock (Eds.) Research perspectives on English for academic purposes. Cambridge:

Cambridge University Press.

271

Gardner, S. (2008) ‘Integrating ethnographic, multidimensional, corpus linguistic and systemic

functional approaches to genre description: An illustration through university history and

engineering assignments. In E. Steiner and S. Neumann (Eds), Proceedings of the 19th

European systemic functional linguistics conference and workshop. 23- 25 July 2007,

Universität des Saarlandes, Saarbrücken, Germany, pp 1–34. Available at:

http://scidok.sulb.uni-saarland.de/ volltexte/2008/1688/ (Accessed 10 September 2018).

Gardner, S. and Holmes, J. (2009) ‘Can I use headings in my essay? Section headings, macrostructures

and genre families in the BAWE corpus of student writing’. In M. Charles, D. Pecorari and S.

Hunston (Eds.) Academic writing: At the interface of corpus and discourse. London:

Continuum, pp. 251-271.

Gass, S. M. and Selinker, L. (2010) ‘Second language acquisition’, Annual Review of Applied

Linguistics (Issues in Second Language Research), 25(3), pp. 46–73.

Gilquin, G. (2008) ‘Combining contrastive and interlanguage analysis to apprehend transfer: detection,

explanation, evaluation’, Language and Computers, 66(1), pp. 3–33.

Gilquin, G. and Paquot, M. (2007) ‘Spoken features in learner academic writing: Identification,

explanation and solution’, in Proceedings of the 4th corpus linguistics conference, University

of Birmingham, pp. 27–30. Available at: https://dial.uclouvain.be/downloader/downloader

.php?pid=boreal:75861&datastream=PDF_01 (Accessed: 1 December 2015).

Gilquin, G. and Paquot, M. (2008) ‘Too chatty: learner academic writing and register variation’,

English Text Construction, 1(1), pp. 41–61.

Granger, S. (1996) ‘From CA to CIA and back: an integrated approach to computerized bilingual and

learner corpora’. In Aijmer, K., Altenberg, B., and Johansson, M. (Eds) Languages in contrast.

Papers from a symposium on text-based cross-linguistic studies. Lunk: Lund University Press,

pp. 37-51.

Granger, S. (2003) ‘Error-tagged learner corpora and CALL: a promising synergy’, CALICO, 20(3),

p. pp.465-480.

Granger, S. (2004) ‘Computer learner corpus research: current status and future prospects’, in Applied

Corpus Linguistics. Amsterdam: Brill Rodopi, pp. 123–145.

Granger, S. (2011) ‘How to use foreign and second language learner corpora’. In A. Mackey and S.M.

Gass (Eds) Research methods in second language acquisition: A practical guide. Hoboken,

New Jersey: John Wiley & Sons.

Groom, N. and Littlemore, J. (2012) Doing applied linguistics. London: Routledge.

Halliday, M. A.K. (1985) Spoken and written language. Victoria: Deakin University.

Halliday, M. A. K. and Martin, J. R. (1993) Writing science: Literacy and discursive power. Pittsburgh:

University of Pittsburgh Press.

Hewings, M. (2013) Advanced grammar in use book with answers. Cambridge: Cambridge University

Press.

272

Hinkel, E. (2003) ‘Simplicity without elegance: Features of sentences in L1 and L2 academic texts’,

TESOL Quarterly, pp. 275–301.

Hinkel, E. (2013) ‘Research findings on teaching grammar for academic writing’, English Teaching,

68(4), pp. 3–21.

Hirvela, A. (2016) ‘Academic reading into writing’. In K. Hyland, P. Shaw (Eds.) The Routledge

handbook of English for academic purposes. London: Routledge, pp. 151–162.

Huebner, T. (1983) A longitudinal analysis of the acquisition of English. Ann Arbor: Karoma.

Hutchinson, T. and Waters, A., (1987) English for specific purposes. Cambridge: Cambridge

university press.

Hyland, K. and Milton, J. (1997) ‘Qualification and certainty in L1 and L2 students’ writing’, Journal

of Second Language Writing, 6(2), pp. 183–205.

Hyland, K. (2002a) ‘Genre: language, context, and literacy’, Annual Review of Applied Linguistics,

22, pp. 113–135.

Hyland, K. (2002b) ‘Specificity revisited: How far should we go now?’, English for Specific Purposes,

21(4), pp. 385–395.

Hyland, K. and Hamp-Lyons, L. (2002) ‘EAP: Issues and directions’, Journal of English for Academic

Purposes, 1(1), pp. 1–12.

Hyland, K. (2004) Disciplinary discourses: Social interactions in academic writing. Ann Arbor, MI:

University of Michigan Press.

Hyland, K. and Hyland, F. (2006) ‘Feedback on second language students’ writing’, Language

Teaching, 39(2), pp. 83–101.

Hyland, K. (2007a) ‘Applying a gloss: Exemplifying and reformulating in academic discourse’,

Applied Linguistics, 28(2), pp. 266–285.

Hyland, K. (2007b) ‘Genre pedagogy: language, literacy and L2 writing instruction’, Journal of

Second Language Writing, 16(3), pp. 148–164.

Hyland, K. (2008) ‘As can be seen: Lexical bundles and disciplinary variation’, English for Specific

Purposes, 27(1), pp. 4–21.

Hyland, K. (2012) ‘Disciplinary differences: language variation in academic discourses’. In K. Hyland

and M. Bondi (Eds.) Academic Discourse Across Disciplines. Bern: Peter Lang GmbH, pp. 17–

40.

Ionin, T., Montrul, S., Kim, J., Philippov, V. (2011) ‘Genericity distinctions and the interpretation of

determiners in second language acquisition’, Language Acquisition, 18(4), pp. 242–280.

Ionin, T. (2012) ‘Formal theory-based methodologies’. In A. Mackey (Ed.) Research methods in

second language acquisition: A practical guide. London: Blackwell, pp. 30–52.

273

Ionin, T., Ko, H. and Wexler, K. (2004) ‘Article semantics in L2 acquisition: The role of specificity’,

Language Acquisition, 12(1), pp. 3–69.

Jarvis, S. (2000) ‘Methodological rigor in the study of transfer: Identifying L1 influence in the

interlanguage lexicon’, Language learning, 50(2), pp. 245–309.

Jarvis, S. (2002) ‘Topic continuity in L2 English article use’, Second Language Acquisition, 24(03),

pp. 387-418.

Johns, T. (1991) ‘Should you be persuaded: two examples of data-driven learning’, ELR Journal, 4,

pp. 1–17.

Krashen, S. D. (1992) ‘Comprehensible input and some competing hypotheses’. In R. Courchene, J.

Glidden, J. St. John & C. Therien (Eds.) Comprehension-based second language teaching.

Ottawa: University of Ottawa Press, pp. 19–38.

Krifka, M. and Gerstner, C. (1987) ‘An outline of genericity’, in Conference for SNS-Bericht,

University of Tübingen (25 October). Available at:http://amor.cms.hu-berlin.de/~h2816i3x/

Publications /Krifka 1987 Genericity.PDF (Accessed 2 December 2015).

Lado, R. (1957) Linguistics across cultures. Ann Arbor, MI: University of Michigan Press.

Langacker, R. W. (1986) ‘An introduction to cognitive grammar’, Cognitive Science, 10(1), pp. 1–40.

Langendoen, D. T. (1970) Essentials of English grammar. New York: Holt Rinehart and Winston.

Larsen-Freeman, D. E. (1976) ‘An explanation for the morpheme acquisition order of second language

learners’, Language Learning, 26(1), pp. 125–134.

Larsen-Freeman, D. and Long, M. H. (1991) An introduction to second language acquisition research.

London: Longman.

Lee, D. Y. W. and Chen, S. X. (2009) ‘Making a bigger deal of the smaller words: function words and

other key items in research writing by Chinese learners’, Journal of Second Language Writing,

18(4), pp. 281–296.

Lee, I. (2005) ‘Error correction in the L2 writing classroom: What do students think?’, TESL Canada

Journal, 22(2), pp. 1–16.

Leedham, M. (2015) Chinese students’ writing in English. Implications from a corpus-driven study.

London & New York: Routledge.

Li, C. N. and Thompson, S. A. (1981) Mandarin Chinese: A functional reference grammar. Berkeley:

University of California Press.

Li, J. and Schmitt, N. (2009) ‘The acquisition of lexical phrases in academic writing: a longitudinal

case study’, Journal of Second Language Writing, 18(2), pp. 85–102.

Liu, D. and Gleason, J. L. (2002) ‘Acquisition of the article the by nonnative speakers of English: an

analysis of four nongeneric uses’, Studies in Second Language Acquisition, 24(1), pp. 1–26.

274

Long, M. H. (1983) ‘Does second language instruction make a difference? A review of research’,

TESOL Quarterly, 17(3), pp. 359–382.

Long, M. H. (1990) ‘The least a second language acquisition theory needs to explain’, TESOL

Quarterly, 24(4), pp. 649–666.

Lu, F. C. (2001) ‘The acquisition of English articles by Chinese learners’, Second Langauge Studies,

20(1), pp. 43–78.

Luk, Z. P. and Shirai, Y. (2009) ‘Is the acquisition order of grammatical morphemes impervious to L1

knowledge? Evidence from the acquisition of plural -s, articles, and possessive ’s’, Language

Learning, pp. 721–754.

Master, P (1986) 'Teaching the English article system to foreign technical writing students'. The

Technical Writing Teacher, 13(3), pp.203-210.

Master, P. (1987) ‘Generic the in scientific American’, English for Specific Purposes, 6(3), pp. 165–

186.

Master, P. (1990) ‘Teaching the English articles as a binary system’, TESOL Quarterly, 24(3), pp.

461–478.

Master, P. (1995) 'Consciousness raising and article pedagogy'. In D. Belcher, & G. Braine (Eds.),

Academic writing in a second language: Essays on research and pedagogy. Norwood, NJ:

Ablex Publishing Corporation, pp. 183-204.

Master, P. (1997) ‘The English article system: Acquisition, function and pedagogy’, System, 25(2), pp.

215–232.

Master, P. (2002) ‘Information structure and English article pedagogy’, System, 30(3), pp. 331–348.

Mayor, B. M. (2006) ‘Dialogic and hortatory features in the writing of Chinese candidates for the

IELTS test’, Language, Culture and Curriculum, 19(1), pp. 104–121.

Milton, J. (2001) ‘Elements of a written interlanguage: a computational and corpus-based study of

institutional influences on the acquisition of English by Hong Kong Chinese students.’ In G.

James (Ed.) Research reports, volume two. Hong Kong: University of Science and Technology.

Murakami, A. and Alexopoulou, T. (2016) ‘L1 influence on the acquisition order of English

grammatical morphemes’, Studies in Second Language Acquisition, 38(3), pp. 365–401.

Murphy, R. (2012) English grammar in use with answers: A self-study reference and practice book

for intermediate students of English. Cambridge: Cambridge University Press.

Nation, P. (2007) ‘The four strands’, Innovation in Language Learning and Teaching, 1(1), pp. 2–13.

Nesi, H. and Gardner, S. (2012) ‘Families of genres of assessed writing’. In H. Nesi and S. Gardner

(Eds.) Genres across the disciplines: Student writing in higher education. Cambridge:

Cambridge University Press, pp. 21–56.

275

Norris, J. M. and Ortega, L. (2000) ‘Effectiveness of L2 instruction: a research synthesis and

quantitative meta-analysis’, Language Learning, 50(3), pp. 417–528.

Nunan, D. (1991) ‘Communicative tasks and the language curriculum’, TESOL Quarterly, 25(2), p.

279.

Odlin, T. (1989) Language transfer: Cross-linguistic influence in language learning. Cambridge

University Press.

OED Online (2010) ‘Oxford English dictionary online’, Oxford English Dictionary. Available at:

http://dictionary.oed.com. (Accessed 4 June 2010).

Paltridge, B. (2001) ‘Linguistic research and EAP pedagogy’. In Flowerdew, J. and Peacock, M. (Eds)

Research perspectives on English for academic purposes. Cambridge: Cambridge University

Press, pp. 55–70.

Parkinson, J. and Musgrave, J. (2014) ‘Development of noun phrase complexity in the writing of

English for academic purposes students’, Journal of English for Academic Purposes, 14, pp.

48–59.

Phakiti, A. (2014) ‘Ethical considerations in experimental research'. In A. Phakiti. (Ed.) Experimental

research methods in language learning: Research methods in linguistics. London: Bloomsbury

Academic.

Phipps, S. and Borg, S. (2009) ‘Exploring tensions between teachers’ grammar teaching beliefs and

practices’, System, 37(3), pp. 380–390.

Pica, T. (1983) ‘Methods of morpheme quantification: Their effect on the interpretation of second

language data’, Studies in Second Language Acquisition, 6(1), pp. 69–78.

Polio, C., Fleck, C. and Leder, N. (1998) ‘“If I only had more time”: ESL learners’ changes in linguistic

accuracy on essay revisions’, Journal of Second Language Writing, 7(1), pp. 43–68.

Quirk, R. and Crystal, D. (1985) A comprehensive grammar of the English language. London:

Longman.

Rahimpour, M., Salimi, A. and Farrokhi, F. (2012) ‘The effect of intensive and extensive focus on

form on EFL learners’ written accuracy’, Theory and Practice in Language Studies, 2(11), pp.

2277–2283.

Richards, J. C. and Rodgers, T. S. (1986) Approaches and methods in language teaching: A description

and analysis. Cambridge: Cambridge University Press.

Robertson, D. (2000) ‘Variability in the use of the English article system by Chinese learners of

English’, Second language Research, 16(2), pp. 135–172.

Russell, V. (2009) ‘Corrective feedback, over a decade of research since Lyster and Ranta

(1997).Where do we stand today?’, Electronic Journal of Foreign Language Teaching, 6(1),

pp. 21–41.

276

Savignon, S. J. (1991) ‘Communicative language teaching: State of the art’, TESOL Quarterly, 25(2),

pp. 261–277.

Schultz, R. (2001) ‘Cultural differences in student and teacher perceptions concerning the role of

grammar instruction and corrective feedback: USA—Colombi’, The Modern Language

Journal, 85(ii), pp. 244–258.

Shi, Dingxu (2016) ‘Nouns and nominal phrases’. In C.R. Huang, C. R. and D. Shi (Eds.) A reference

grammar of Chinese. Cambridge: Cambridge University Press, pp. 199–255.

Shi, L. (2006) ‘The successors to confucianism or a new generation? A questionnaire study on Chinese

students’ culture of learning English’, Language, Culture and Curriculum, 19(1), pp. 122–140.

Sinclair, J. (1991) Corpus, concordance, collocation. Oxford: Oxford University Press.

Snape, N., Leung, I.K. and Ting, H.C. (2006) 'Comparing Chinese, Japanese and Spanish speakers in

L2 English article acquisition: Evidence against the fluctuation hypothesis?'. In M. Grantham

O’Brien, C. Shea. and J. Archibald (Eds) Proceedings of the 8th generative approaches to

second language acquisition conference (GASLA 2006). Somerville, MA: Cascadilla

Proceedings Project, 132–39.

Sowton, C. (2012) 50 Steps to improving your academic writing: Study book. Reading: Garnet.

Spack, R., (1988) 'Initiating ESL students into the academic discourse community: How far should we

go?' TESOL quarterly, 22(1), pp.29-51.

Swales, J. M. (1990) Genre analysis: English in academic and research settings. Cambridge:

Cambridge University Press.

Swales, J. M. (2004) Research genres: Explorations and applications. Cambridge, UK: Cambridge

University Press.

Swales, J. M. and Feak, C. B. (2004) Academic writing for graduate students. Ann Arbor, Michigan:

University of Michigan Press.

Swan, M. and Smith, B., (2001) Learner English: A teacher's guide to interference and other

problems. Cambridge: Cambridge University Press.

Taylor, J. (1989) Linguistic categorisation: An essay in cognitive linguistics. Oxford: Oxford

University Press.

The Higher Education Statistics Agency (2019) Higher education student statistics: UK, 2017/18 -

Where students come from and go to study, 17 January 2019. Available at:

https://www.hesa.ac.uk/news/17-01-2019/sb252-higher-education-student-statistics/location

(Accessed: 1 June 2019).

Thewissen, J. (2015) Accuracy across proficiency levels. A learner corpus approach. Louvain: Presses

universitaires de Louvain.

Thomas, M. (1989) ‘The acquisition of English articles by first-and second-language learners’, Applied

Psycholinguistics, 10(03), pp. 335–355.

277

Truscott, J. (1996) ‘Correction, the case against grammar classes, in L2 writing’, Language Learning,

46(2), pp. 327–369.

Truscott, J. (2004) ‘Evidence and conjecture on the effects of correction: A response to Chandler’,

Journal of Second Language Writing, 13(4), pp. 337–343.

Universities UK (2019) International Facts and Figures 2019, 30 July 2019. Available at: https://

www.universitiesuk.ac.uk/policy-and-analysis/reports/Pages/Intl-facts-figs-19.aspx

(Accessed 1 September 2019).

University of Birmingham (2016) 'Ethical Review of Research' Available at: University of

Birmingham Intranet https://intranet.birmingham.ac.uk/finance/documents/public/ethical

reviewpgstudent.pdf (Accessed: 4 December 2016).

Ure, J. (1971) 'Lexical density and register differentiation.' In G. Trim and J Perren (Eds.)

Applications of Linguistics. Cambridge: University Press.

Verspoor, M. H. (2008) ‘Cognitive linguistics and its applications to second language teaching’. In

Encyclopedia of Language and Education. Boston, MA: SpringerLink Books, pp. 1843–1854.

Wang, T. (2001) ‘Understanding Chinese culture and learning’. In Proceedings of the international

research conference of the Australian association for research in education. Adelaide,

Australia, 26-30 November: Australian Association for Research in Education. Available at:

http://www.aare.edu.au/06pap/ (Accessed 4 March 2018).

Weale, S. (2019) ‘Chinese students’ applications to UK universities up by 30%’, The Guardian

Newspaper, 11 July. Available at: https://www.theguardian.com/ education/2019/jul

/11/chinese-students-applications-to-uk-universities-up-by-30 (Accessed 1 September 2019).

Wu, X. (2009) 'The dynamics of Chinese face mechanisms and classroom behaviour: a case study',

Evaluation & Research in Education, 22 (2-4): 87–105.

Xu, Q., Shi, Y. and Snape, N. (2016) ‘A study on Chinese students’ acquisition of English articles and

interlanguage syntactic impairment’, Chinese Journal of Applied Linguistics, 39(4).

Yamada, J. and Matsuura, N. (1982) ‘The use of the English article among Japanese students’, RELC

Journal, 13(1), pp. 50–63.

Yang, M. and Ionin, T. (2008) ‘L2 English articles and the computation of uniqueness’, in Proceedings

of the 3rd conference on generative approaches to language acquisition. North America,

University of Connecticut, pp. 325–335. Available at: https://pdfs.

semanticscholar.org/ee98/4f75c2bb1172738c8ad4fbe4e44eb6618988.pdf (Accessed:

1 December 2015).

Yoo, I. W. (2009) ‘The English definite article: What ESL/EFL grammars say and what corpus

findings show’, Journal of English for Academic Purposes, 8(4), pp. 267–278.

Young, R. (1996) ‘Form-function relations in articles in English interlanguage’. In R.B. Preston and

R. Bayley (Eds.) Second language acquisition and linguistic variation. Amsterdam:

Benjamins., pp. 135–175.

278

Zdorenko, T. and Paradis, J. (2008) ‘The acquisition of articles in child second language. English:

fluctuation, transfer or both?’, Second Language Research, 24(2), pp. 227–250.

Zhang, S. (1995) 'Re-examining the affective advantages of peer feedback in the ESL writing class',

Journal of Second Language Writing, 4 (3): 209–222.

279

APPENDICES

APPENDIX 1

Teacher Survey [Online] Questionnaire (Chapter 4)

280

281

APPENDIX 2

BALEAP Mailing List Responses to Survey Question 4 Open-Response (Chapter 4)

Supported self study

I believe that by drawing their attention to focused examples and errors, we can them get the students

to extrapolate 'rules' and formulate hypotheses about when (not) to use articles.

it is something they need to 'notice' and get feedback on - usually through text analysis of related work

or through dictogloss.

There seems little point in teaching "rules", apart from eg abstract nouns, and advise to use plurals to

avoid error, but would, if had time, occasionally distribute a short piece of student writing and error

correct/analyse in class. I would consider this mostly to be covered in giving individual student feedback

on their writing both orally and in writing.

Students correct an anonymous piece of work and explain the reasons for any changes they make.

Possibly working in pairs rather than individually as it would increase discussion of the task. The piece

of work could be actual or a fictional piece created by the tutor and containing typical errors.

Alternatively, after a few weeks, amalgamation of the group's own typical errors would be interesting if

time-consuming to produce.

This is something students need to 'acquire' rather than learn. So perhaps the best approach is promote

noticing: to draw attention to / ask questions about the use of articles as they occur incidentally in a

lesson, rather than explicit lesson segments.

Yes by consistently pointing out instances of misuse when it would impact on meaning or the impression

they give of themselves. Asking them to record their own speech and identifying article use/misuse

works well.

Not sure. I think articles are one of 'final errors', if you like, NNS that have been in an English country

for many years can still make errors with articles. It may be improved with greater exposure to language

and language environment than direct teaching in class.

Through awareness raising exercises, regular "small doses", texts (from students' own academic fields)

with articles removed, proofreading samples of student writing.

There is no easy answer, sometimes it is because their first language does not have articles. Getting

them to notice and a bit of translation (even though it is unfashionable) can get them to recognise the

difference.

Providing guided opportunities to notice. As much exposure to authentic material as possible.

282

Yes we can Firstly draw attention to the errors. Then isolate them and explain some simple rules of

thumb, explaining that it is a common area of error. Then recommend a grammar text book (eg

Murphy/Hewings) Then recommend a series of journal articles in their subject- tell them to look at the

noun phrases

Learn about articles ourselves. Know more about how they work, while still acknowledging that even

the best theorists (e.g. Hawkins) do not themselves have a definitive answer on this. Teach heuristic

rules explicitly. Highlight errors in S work and give them the opportunity to correct. Praise improvement

in use.

Simplify article use by showing that the definite article is an audience oriented particle which signals to

the reader - you know what I'm talking about.

It seems me that the biggest area our students have difficulty with is generalising. After that, they need

to be explicitly reminded that 'a' = 'one &/or any' while 'the' answers the question 'which'. My students

have also found it helpful to be reminded that prepositional phrases, especially the construction 'the x

of y', very often signal a need for 'the'. I've found a good way to raise their awareness (and hopefully

increase their accuracy, but I've never tested this) is to have them regularly mark a short passage (from

a journal, a text book, or news article usually): they highlight all the nouns, then they examine which

take articles and explain why (not for all, but for a sampling, usually one each of null, a/an, and the).

We can certainly help them improve their accuracy in English article use. I often use comparison with

their first languages, which helps them clarify similarities and differences and may serve as memory

hooks of when to use or not to use articles in English. Analysing texts and asking the students to explain

why a particular article is used or why there is no need for an article is also very useful. Perseverance

and constant repetition of rules in various forms does, eventually, bring an improvement.

I think it is partly a developmental stage which improves over time (like the 3rd person s), which also

encourages students as some can get very down about the difficulties they have with articles. I haven't

found generic exercises such as 'correct the errors' in random sentences help really. However I have

found that activities which help students notice their own specific errors in their own writing does seem

to make some difference.

If time allows, sit down with the student on a one-to-one basis as they are writing (or correcting

something written) and ask them to explain the reasoning behind their article use or non-use, and put

them on the right path if necessary. Direct them towards some reinforcement activities which they can

do on their own. Monitor future written work to see if article-related errors persist.

In science subjects it can be important to use articles accurately, and so close marking can be helpful.

These questions are not helpful however as they are too regimented. Academic writing correction

focuses on different aspects of a student's writing each time, depending on many many other factors,

articles are only one element of this. Sometimes the mistakes made change the meaning that is

intended, then it may be important to focus on that at that moment. So much to add. But a terrible choice

for a PhD.

plenty of tutor-facilitated targeted self-study (i.e. rather than classroom teaching)

By teaching them with noun phrases and discussing specificity

I think it's very difficult, especially if a student's first language has a different or no article system.

Perhaps trying to raise their awareness of the importance or significance of articles in English and how

meaning changes ("I have cake/a cake", "he has time/the time"). Otherwise, particularly as higher level

283

students are sometimes quite able to self-correct if an error is pointed out, it seems that they just don't

see it as being that important!

For example: Introduce and revise the rules of use; look at examples correct use in texts and identify

reasons for the presence/absence of articles. However, improvement is likely to be slow and a non-

native speaker will probably still make occasional errors even if they achieve a very high level of

proficiency.

Yes, teach, practice with paragraphs of error-filled text, etc.

Not sure how much we can. [Is it even desirable? I think that depends on context. For a subject tutor

who struggles with NNS writing, it could help (at least reducing their emtional recoil from "otherness").

For a more confident tutor, it's much less important. In a less parochial sense, English is becoming

much more global (cf English Next et al), with washback into "Centre" countries from other users.] Is it

practical? I suspect not in most cases, as habits will have formed long ago and actual use is very hard

to understand and emulate. Something like Rosetta Stone might do much more good than teachers in

classrooms.

This is the most complex area of English grammar with so many rules (all of which have exceptions)

that I know there is a theory that students cannot improve in terms of accuracy through conscious

application of rules. However, as with all areas of grammar, I think occassionally revising the rules,

together with subconscious acquisition gained through immersion (reading most of all) might lead to a

gradual reduction in the number of mistakes.

I am not sure we can 'teach' the use of articles beyond a very elementary stage, I always tell students

that the grammar rules are likely to hinder as much as help them. I recommend a 'language awareness'

approach - to pay attention to the use of articles in anything they read/hear/etc. For me, how articles

are used needs to be learned almost instinctively ...

This is a difficult one; I think reading and noting collocations (eg consumer loyalty) helps. I often

compose a paragraph of student errors and put it on the whiteboard and elicit feedback from groups. I

do occasionally devote time specifically to article use but early on in a 40 week pre-sessional course.

I've been teaching EAP for 9 years, and although I think high accuracy is important in article use, it is

quite minor compared to development of content, paragraph support and analytical skills.

Collect the mistakes we find in their work and periodically present them back to the group along with

error free examples. Ask them to identify whether correct or incorrect and to rewrite the incorrect ones.

students look at a text. mark all articles and absences of articles. try to extrapolate some rules. discuss.

repeat periodically.

Yes, we can help. Constantly flagging it up in corrective feedback may help over time. Access to a

grammar reference source such as McCarthy and Carter's Cambridge Grammar. Dictagloss is a handy

activity to focus learners on accuracy at the end of a task cycle.

As an awareness-raising exercise, I sometimes give a handout with article use explained, and ask them

to go through the rules deciding which ones also apply in their language and which are different (of

course this doesn't work with students whose 1st language has no articles! it does work with eg L1

Spanish speakers...). Otherwise, I tend to focus on articles when giving feedback rather than through a

dedicated lesson on the topic - eg by collecting some article errors, especially ones which could be

misleading in terms of meaning, putting them up on a slide, adding some phrases/sentences where

articles ARE used correctly, and asking pairs/groups to locate (and of course try and explain) the errors.

284

I used to think article errors were relatively unimportant, and in many cases (e.g. "in UK") I still think

they are. But then I realised that if I was reading a student text in a less familiar (to me) area I often

genuinely didn't know whether, for example, they were speaking generically about something or

whether they were actually referring to something specific and had simply omitted the article (or vice

versa, were they referring to something specific, or had they added an unwanted article when actually

the intention was to refer generically) (hope this makes sense!). Otherwise, I ask them, when they are

reading articles in their field, to stop sometimes and read one paragraph in detail, focusing on all the

nouns and explaining to themselves why the author has chosen to use an article or not, in each case.

And sometimes I might take a paragraph myself, remove all articles, and ask them to put back the

articles where appropriate. Hope this helps!

For me it is very bound up with nouns and countability, so we need to focus on those areas in our

teaching-but getting students to 'notice' the gap between expert users usage and their own work is also

key.

Awareness-raising will be useful for some students, allowing them to improve their understanding/use

over time. Short focused activities (e.g. correction exercises, use of corpora lines to note patterns) may

also help (some) students appreciate when/where/why the use/omission of an article may to blur

meaning.

I find it difficult to explain and believe that extensive reading on part of the student is the best solution

Students can only improve their accuracy in this if they themselves believe it is important i.e. will be

better for the reader and/or get them a better grade. Most students 'know' the rules, but don't notice

their application. Recognising correct use (error correction of modified texts), using this in editing their

own work or that of peers, and responding to feedback from tutors - these are crucial.

Articles can be persistently problematic for students even at advanced levels of proficiency. At lower

levels, raising students' awareness of their use of articles can make a difference, with peer correction

being useful in this regard.

Yes, but it's very difficult. Awareness raising, examination of rules, practice exercises.

I think it's about making them aware of the differences between their L1 grammar structure and English.

In some languages, articles are not used so it's making them aware of the differences and allowing

them to see that if they're not used sometimes it can impede meaning. With countries' names such as

the UK, the USA 'the' is part of the name so is needed. However, at the same time, if meaning can be

inferred then sometimes it's not worth bothering with too much as students can get mixed up and there

are bigger errors that can be focused on.

Encouraging them to read (ideally, extensively) relevant academic articles and papers (or in this case,

just any old text!) and NOTICING / concentrating specifically on how articles are used.

I'm unconvinced that beyond pre-academic study it's really worth explicitly teaching this point at all, for

the little return which may come from doing so, not least in terms of its impact on meaning. By the time

students reach undergraduate or postgraduate level, I doubt if it's more than a minor irritation to their

subject lecturers (with perhaps a few exceptions in language-related subjects). With more and more

English-medium research published (and edited) by non-native speakers of English, occurrences of

article 'errors' may well become increasingly frequent. On the other hand, if you put on a session or

workshop (or series of such) specifically dedicated to this area of grammar, I'm sure it will be quite

popular.

285

Article usage is a particularly hoary subject. Many advanced near-native academics working in UKHE

exhibit errors of one kind or another, often displaying consistent patterns of error influenced by L1. We

used to talk about 'error gravity' a long time ago, and I believe that inaccuracy in this particular area is

seen as 'error lite'. We correct such, almost instinctively, but we cannot hope that we can be fully

effective in view of the low functional load that this complex bit of grammatical kit brings to bear.

I generally ask students to proofread and peer check. I think that students can generally avoid article

errors when they concentrate on articles specifically. However, when grammar is not the focus I think

there is a tendency to translate from the student's L1 which means they make errors with articles.

review of general rules eg. specific / abstract / introducing for first time noticing in authentic texts ie.

when reading, especially academic articles in their subject discipline dictaglosses / writing up lecture

notes

Yes - I would get them to look at examples of successful student work in the context they are going into

and get them to notice the use of the articles particular to that context. I would then help them establish

some of the context rules for this (generalised) and highlight where they don't follow these at appropriate

times in their own work.

Yes. I use some example sentences (with overuse of 'the') and have students identify which uses of the

definite article is correct. After these sentences have been argued about and checked :-) students have

to create rules of use derived from them (i.e. 'the' should be used with e.g. superlatives, and rivers, but

not with e.g. cities or months - exceptions are discussed). Then students test each other on the rules.

A few weeks later, I test them again / recycle info / check understanding (repeat as necessary). In

addition, I have them check each other's work for correct use of articles (according to the rules they've

been learning) - often students simply don't notice their own errors. Possibly have them identify nouns,

noun phrases in their writing and decide whether the definite, indefinite or zero article needs to be used.

Sometimes students overuse 'the' at the beginning of a text, where they should initially use 'a/an' - this

leads on to a discussion of (reader/writer) shared knowledge & grammatical cohesion. While it's

important to teach use of articles, at the end of the day, mistakes with articles are usually non-impeding,

so it is important that marking criteria reflects this.

We can help but it is a slow and ongoing process. I try to focus on a limited number of rules at any one

time. Perhaps more useful to get students to explain correct usage in well written texts rather than

always responding to errors.

Yes, I think we can. Feedback and consciousness raising activities work well in addition to form-focused

instruction. I would say that I wouldn't focus on articles if there are other serious problems with a

learner's writing - I suppose I see articles as quite advanced; more advanced than a number of my

learners perhaps.

By analysing texts and the way articles are used there and why

I think this is the most difficult aspect of English for a NNS to acquire; even very high-level students

have the occasional problem with articles. I think by drawing students' attention to proof-reading their

own writtwn work for article mis-use is possibly the best we can do for them.

Some basic rules e.g. with plural country names (the UK, USA etc) are helpful. Ideas which are harder

to grasp quickly, e.g. talking about generic versus specific items, can be illustrated through reading and

recognising article use. In my experience, there is no quick fix to problems with article use and students

tend to be worried about it.

286

It seems a very persistent type of error, although students usually know 'the rules'. In teaching/clarifying

situations, I tend to focus on examples where the meaning is unclear because of mistakes in articles,

eg general or specific reference, which could confuse a reader/listener rather than dwell on easily

understood examples (eg 'UK')

Learning how to use articles takes a long time so we can't expect perfection, but if they can get a few

right, that';s a good thing and it increases the positive impression of the work.

To be honest, I think that the teaching of general English deals with this area more explicitly. Rarely

does EAP material address this, yet it is, unlike many grammar areas, something that appears in general

English materials from beginner to advanced. Therefore, course writers clearly perceive it as an

important topic to revisit. They acknowledge that learners rarely get the hang of it immediately. I would

say that I am less likely to address it explicitly as an EAP teacher because it is less important than other

features of EAP. This focuses on meaning, argument and expression - only when these are unclear

would I consider articles as a discrete item in a an EAP lesson. I would be interested to read this

research to see how the teaching of articles relates (or not) relates to EAP specifically.

I try to teach them some basic 'rules'. 'The' is used for something specific, one of a kind, unique. 'A, an'

are used for one of many. And such...

1. Raising awareness of use of articles as well as understanding how they work with countable and

uncountable nouns, certain phrases and usages is very important, especially for speakers of languages

which have no articles, or use them differently. So some activities where students can work out some

patterns/'rules' can help. 2. Otherwise, extensive reading can be a strategy to develop article

awareness, perhaps unconsciously, as well as other areas or grammar. 3. Learning collocations which

contain articles is useful to build up an understanding of their use. (hope this helps)

Not sure if this can really be taught without including determiners. If students are of a high enough level

they can usually conceptually grasp the starting point that an English NP (that is not a person, place or

uncountable) will need something more: a determiner, pluralisation, or both. I find this particularly useful

in trying to get past the frustration of the exceptions/sometimes arbitrary nature of article/determiner

use with speakers of language groups which don't use articles and who are more likely to struggle with

their use.

https://www.amazon.com/Book-Practice-toward-Mastering/dp/0472086391 This book outlines 125

rules for articles, some of which are American "in the hospital v in hospital". A further inconvenient truth

is that he writes the answers to each exercise directly below the exercise. Clearly, Cole knows more

about articles than how to write an effective textbook. To me, the book is proof of what a Herculean task

we face and also proof that we are following a tone of article rules that we were never explicitly taught.

I believe that students can be taught basic rules of when an article is needed, but true finesse with

choosing defining v non-defining seems more like something that has to be picked up the same way

that natives do it: through massive exposure and experience with the language. To be a bit provocative,

we could help students learn articles by admitting that the vast majority of native English teachers have

no idea what the actual rules are. For example, it's Oxford University (with no article) but it's the

University of London. This rule is covered in Cole. They just have a feel for what's right. That's the

difference between being able to grab a guitar and play a song and being able to write the song down

with full musical annotation. If all you can do is grab a guitar and play, maybe you shouldn't attempt to

teach people annotation. Teachers, on the other hand, will generally as a point of pride have a go at it,

trying to apply simple rules to too many contexts, making up rules they think fit, and possibly just lying

to avoid telling a student they simply don't know. ^That was me many years ago until I realized how it

is the opposite of teaching to do that. Teachers should never be afraid to say "I don't know" and admit

to the limits of their knowledge.

287

My response to this is to get students to closely analyse reading texts which have either been introduced

as part of the syllabus or which they independently chosen and to look for specific language use issues

- in this case the use of the article (pre-teaching article use if appropriate) and to compare in pairs, then

larger groups and finally in plenary discussion. I might add that, for example, I often work with very able

Kazakh students who really do struggle with the use of the article even when verging on or even

reaching C1 level.

Articles, by comparison with many other features of English, are cosmetic and have no real function i.e.

they don't carry any additional information or tell you anything you wouldn't know if they were not there.

therefore these are not a top priority. However, if students have chosen to study in English that means

they must do their best to achieve a good level of proficiency and after mastering all the rest of the

grammar getting details such as articles and prepositions adds polish to the end product [spoken or

written]. Another view is that these errors are relatively 'quick-fixes' and I've taught on courses where

the aim seems to have been to sort out the easy problems while glossing over the more fundamental

ones. Regarding actually teaching - I'm afraid I have no magic formula to offer, only constant practice

and feedback.

I combine teaching of articles with the notions of specificity/definiteness and generality. I use circles (as

in set theory) to draw these notions and I also try to mention the use of ind/def art. and bare plural for

generics (all in very simple terms)

Rather depends on what is meant by improve! I think we can raise awareness and use noticing skills to

help them to see how article use is accurate in some texts and inaccurate in others, including in their

own writing. To what extent this has an impact on their real time accuracy, either in writing or in

speaking, is of course a much more difficult point to measure

I think seeing the definite article as part of teaching text cohesion (its function in signalling previously

mentioned items, assumed prior knowledge or something fully defined or unique) is the important

teaching point. I would teach it alongside the use of cohesive general nouns and the use of 'This' as a

cohesive marker, especially in topic sentences - This approach/ The reason for this etc. Attempts to

devise 'rules' are a waste of time and show lack of understanding in teachers and materials writers.

Real minefield. The so-called rules on articles are not at all useful. Constant exposure to and discussion

of instances of articles - or not - helps but I do think correction is important when done in a positive and

sympathetic way. Learners won't thank a teacher if she/he simply avoids correcting in this vital area.

George

I 'teach' articles mainly with PhD students, who are usually aiming for a high level of accuracy and want

to focus on the nitty-gritty of language. I take their frequent errors as a starting point and focus on

common patterns in academic discourse (e.g. 'in the 19th century' vs. '19th century literature') with the

aim of enabling them to reduce, if not eliminate, their errors in this area. Generally speaking, whether

or not I would correct the use of articles in your example sentence would depend on what the rest of

the student's writing was like. Missing/unnecessary articles here don't seem to impede understanding

or create ambiguity, so I wouldn't bother if the student had, for example, problems with sentence

structure that did prevent them from putting their points across.

I suppose the use of a copus could be useful.

Making them aware of errors (at least in their writing - orally is another matter?!) and asking them to

work out why these are so, in tandem with discussing the use of articles in pieces of text and

encouraging them to work out why articles/non-Articles are used. A good example of this (which I've

288

adapted to use with other relevant texts) is an activity about traffic in New Zealand in Bruce, I. (2011)

Theory and concepts of EAP, Palgrave Macmillan (sorry, full ref. and page number not to hand!)

Correcting surface errors like articles are part of our role as EAP tutors. I feel this is best done using

error codes and the student's own work rather than abstract sentence level grammar exercises.

I think it is important to sensitise students to the issue of article use and encourage them to self-correct

in revision drafts. So an overview of article use (quite well covered in Bailey) and then pointing out

examples little and often. Some students don't seem to think the issue important, while others agonise

over accuracy. I tell my students that, while clear expression of meaning is more important that total

grammatical accuracy, poor use of articles can be like an irritating scratching noise as you read, so it is

helps the reader if you get it right. I try to maintain a balance between not sweating the small stuff - as

often a meaning is clear even with the poor article use (sic) - and producing grammatically accurate

(and therefore more readable and possibly more authoritative) text. I also teach the use of articles as

part of coherence.

Yes, we can help. Noticing - encourage students when reading to notice use of articles. Read

(interesting/easier article) first time for content, second time for language, foussing on articles. Students

choose sentences to write down to refer to when writing.

I have not yet come across an effective way of correcting the use of articles where they have been

wrongly learned in the first place, which is often the case with international students I have taught. My

EAP experience consists mostly in fairly short presessional courses, on which there is usually little to

no time for language development. I think scheduling more time for articles, as well as other aspects of

language which pose difficulties, incorporating follow-up sessions, and perhaps focusing exclusively on

general language feedback on some of students' written assignments (and asking for revisions of what

they have writen with a focus on articles) could be helpful.

The students I teach are on a foundation year for undergraduate study in English speaking universities.

They have come from a background of learning English with a strong focus on grammar and speak a

first language that does not include articles. I tend to highlight examples as they come up in class and

in written work asking for suggestions and focusing on generalities of use as appropriate. I tend not to

look at article use with planned focuses on any particular use, however, as students tend to recognise

the points from previous study quite quickly and this does not seem to reduce error frequency. I believe

highlighting the errors they make will over time increase the likelihood of self correction taking place

especially if I regularly highlight the most commonly made errors .

We can help them but it is a slow drip feeding process.

I'm a second lang learner myself at PhD level and make mistakes with articles and prepositions. I'd

teach the rules which are sort of vague anyway but in general I'd let it go. They will never be perfect but

comprehensible and comprhensive input exposure to language will eventually reduce the occurances

;)

I don't see why not - after all, it's a basic building block of TEFL. However, as a pre-sessional EAP

teacher, there was hardly any time on the course to focus on language systems - our syllabus was

mostly skills-focussed.

It`s a huge challenge but students can improve. One way of helping them, of course, is through

consciousness-raising, i.e. having them puzzle out differences in meaning caused by the presence or

absence of certain articles. E.g. `Japan` vs. `the USA`, it`s also worth using the `automaticity` principle

and getting students to learn key examples as formulaic phrases.

289

Clearly we can help them, and these errors need to be highlighted as they are non-salient. However,

as ever with error correction are you sincere in your use of the word 'improve', or are you seeking

perfection. Incremental improvement and awareness is feasible, but complete accuracy is likely to

remain elusive. As my focus here is on awareness, the classroom intervention on articles is one which

occurs early, and creates the vocabulary of feedback which will be used throughout the year. The aim

is to develop triggers which can rapidly indicate errors in article use and aim for self-correction and,

later, self-regulation.

Analysis of example texts focusing on sentence and paragraph level patterns and structures.

Firstly, students need to be aware that they have a problem with their use of articles. Show them some

examples of how use of articles can change the meaning of sentences; this will demonstrate to your

students that misuse of articles can be a serious issue. Having explained to students how articles work

in English, have students peer-correct one another's work, and self-correct their own work. Ultimately,

students acquire English article usage by doing a massive amount of reading.

Analysing/deducing rules; controlled practice; reading and noticing article use; focused writing practice

There's no magic wand - they just have to learn the rules and develop an ear for how articles are used.

This latter is tricky of course because articles tend to be unstressed and therefore difficult to pick out.

Can I just point out that 1. you've used 'on' instead of 'in' 2. 'deduct' instead of 'deduce' 3. the first

question is extraordinary given that it's addressed to English teachers!

Peer-correction Reference to actual texts/sources

The resources or explanations I use depend on how much detail I want to get into, on the level of the

student, and on how eager they are to use accurate language. General guidelines on usage only work

as a starting point, but in a grammar lesson, it is often a question of practice. In any case, I would use

the student's own production to try and understand their thinking process first.

With some high level students (C1) I encourage them to produce their own cloze exercises by taking a

text, and using Word, doing 'replace all' for the articles and substituting them for an underlined gap and

then trying to replace the missing articles and checking with the original, as it is only by huge amounts

of practice that they will gain a feeling for article use (which is not 100% consistent even by natives!)

However, this practice does not help with zero articles - that would require a cloze produced by a

teacher. This develops learner autonomy and encourages them an investigative frame of mind.

Depends on the students first language. For those whose first language does not feature an article,

students most frequently depict huge problems with the concept and ineptitude with the usage.

Prolonged, regular teaching is wasteful of time seldom resulting in apparent gains in knowledge and

usage.

By drawing their attention to the rules and meaning of articles and reinforcing this when giving feedback.

Encouraging them to notice and think about why articles are used in any reading they do.

Use newspaper and magazine articles, asking students to highlight a/an/the/0 examples on a regular

basis then cross-reference to the article/zero rule. Doing this regularly can focus their attention on their

own use.

Firstly, I try to set appropriate expectations: this is an area that even very advanced users find difficult

if their L1 does not have articles (I use the example of a senior UN official from Vietnam). I tell students

that they are unlikely to make dramatic progress in five weeks or in eighteen hours of in-sessional,

290

though they can make progress over longer timescales. Secondly, I try to teach both relatively

straightforward rules (country names with common nouns; superlative adjectives) and the general

principle. For the latter I use a flowchart produced in-house, which encourages students to ask the

question "Is this use general or specific?". I tend to rephrase this as "Do the writer/speaker and the

reader/listener both know which one they are talking about?", which students seem to find helpful.

Thirdly, I find it interesting that you describe the options as [a/an, Ø, the]. I know this is the standard

formulation, but it seems to me that many student errors with plural -s are actually part of the same

system network (in the SFL sense of a series of choices). However, I am not an SLA specialist, so this

maybe misguided.

What an interesting survey! I coordinate an in-sessional service in an ancient university. The service is

predominantly used by PG students, of which about 30% are PGRs. Much of our teaching takes place

in one-to-one tutorials. I'm answering your questions in this context. I spend a lot of time teaching article

use to PGs. We tend to approach teaching articles using Caplan's material: Caplan (2012) Grammar

Choices for Graduate and Professional Writers, Ch.5. One of the strengths of Caplan's approach is not

to teach articles, but the noun phrase. Another is to start the unit with a diagnostic test so that the one-

to-one teaching can be highly targeted. We combine this with getting students to notice and explain the

meaning of articles in their disciplinary reading. We then get students to nominate electronic articles in

their discipline and we (teach them to) make (self-)study material using their reading texts. I've seen

able students in challenging disciplines (in terms of article use) go from 30% accuracy to 95% accuracy

in a semester. In short, the key is to connect article use to meaning, to motivate learners by getting

them to notice articles and to interpret them more sensitively when reading, and (as a teacher) to take

account of disciplinary variation. BTW - Caplan is coming to our EAP conference at St Andrews in Feb

2017. You would make a very interesting conference speaker/delegate if you have time to come to StA.

The conference theme is Coherence & Cohesion. My contact details: Janie ajb31@st-andrews.ac.uk

In any case, good luck with your thesis.

Of course. This needs to be progressively trained. Accurate determination is an essential aspect of

academic writing (almost by definition!).

Dear Richard, this is a very interesting area indeed. I first came across the problem in a major way in

South Africa, where it is widely accepted now that speakers of Bantu languages do not use the article

correctly, even in academic writing. If you are interested, Vivien de Klerk, Rhodes University in

Grahamstown, has done research on this. However, in an academic context in Europe I do teach and

correct the incorrect use, mainly because many subject tutors are not as "liberal". Good luck with your

Phd! Regards, Monika Marienhagen

Helping them to identify usage in context, then awareness of their own errors through teacher correction

and finally, self correction.

This certainly is a difficult aspect of grammar of teach and sometimes the use of the article does not

even affect meaning. I'm not sure that 'grammar lessons' help so this is not really something I do.

Individual feedback followed by tutorials is the best way to tackle this area of grammar. Independent

work that is checked and discussed with the student is also helpful.

Explicit examples and practice activities, particularly showing how strange English sounds without

articles. NB "deduct" does NOT mean "deduce."

English articles are definitely a tricky issue to teach. There seem to be some very clear rules about their

use, but because there are so many, students often confuse them. Students whose L1 has this

grammatical feature (e.g. French, Spanish, Italian) will make fewer mistakes, but students whose L1 is

a case-based language (e.g. Polish, Serbo-Croat, Russian, Ukrainian), or students who are Arabic or

291

Chinese speakers will need to be continually supported with frequent exercises and focused correction

of their writing.

Yes, we can. Written and oral feedback on written work. Remedial teaching 'of article use' in class.

I point out to them the frequency of these errors, and ask them if they want me to indicate where they

are, and go over the rules with them. I also let them know whether or not they interfere with

communication or change the meaning of the sentence. However, the errors are so widespread (and

difficult to address) in speakers of EFL/ESL that my own feeling is that the use of articles is changing

in most varieties of English, especially in ELF contexts, and many students feel they are probably not

worth concentrating on.

The articles in the example are being used correctly in Indian English.

Many students see improving their use of articles as a difficult task. The rules might seem clear, but

when it comes to applying them students continue to make mistakes and feel frustrated and

discouraged about their inability to get them "right". As in the sentence above, often the errors do not

impede meaning; a proficient user of English can "fill in the gaps" without any trouble and there is no

ambiguity. However, if students' work is going to be published or produced for a wider audience, they

need to aim to produce work which is as accurate as possible. Not all readers are tolerant of

grammatical inaccuracy and it could undermine the authors' authority. I think if we don't point this out to

students, we do them a disservice. As for practical ways of improving students' accuracy in improving

English article use, I don't think there is any magical fix! I think most of it is in the hands of the students.

They need the will to improve this particular aspect of grammar, continued noticing of its use in written

texts, proof-reading of their own writing, perhaps practice grammar texts as part of independent study.

The teacher can help by focusing students' attention on articles regularly as part of feedback on

students' written production. Whilst it is useful to dedicate a lesson to article use, as it fulfills student

expectations to "do grammar", it is more important for teacher intervention to be timely and consistent.

Through text analysis and students chunking noun phrases. These chunks should then be recorded in

their unique and originally created academic phrasebanks

Exposure to natural writen / spoken English and awareness raising of patterns in article use.

292

293

APPENDIX 3

Transcripts of Six Tutor Interviews (Chapter 4)

The interviews were semi-structured. All tutors were asked questions 1-2 as worded below,

but answers were followed up with various prompts.

1. Question 2 asked about how often you teach the article system to your students. Can you

talk a little more about when you teach them and how you teach them? When are you most

likely to teach article use?

2. Question 3 asked about how likely you’d be to highlight the errors/give feedback in the

example. In general, do you give corrective feedback? How often and when do you give

corrective feedback? How do you do it?

Interview Transcripts

Interview Tutor #1

Question 2 asked about how often you teach the article system to your students. Can you

talk a little more about when you teach them and how you teach them? When are you

most likely to teach article use?

Teaching the article system? To be frank, I don’t have enough time to teach it in the classroom

in one session of its own. With the time pressure and everything we cover, I don’t have time

to dedicate time to this. We should be focussing on the process of researching and drafting their

work and looking at how texts hang together. It might come up in cohesion and I might send

them to a grammar reference for homework. But I don’t formally teach it in EAP. I think it’s

quite a trivial surface error. But each time when I correct writing…when I check students’

writing, I will check and point out any errors in articles most of the time.

Question 3 asked about how likely you’d be to highlight the errors/give feedback in the

example. In general, do you give corrective feedback? How often and when do you give

corrective feedback? How do you do it?

294

I would like to say I do peer correction and things like that…but I just don’t have time. Each

time… I correct the errors – I often use codes for all grammar errors. Students will get articles

marked two or three times…sometimes more if there’s a high stakes assignment. Maybe that’s

not ideal, but we are so short of time. So, each time if I spot any error I will just mark it. And

then in the tutorial time I will talk through it. And…if..actually if there are more than one

students making similar mistakes…in sort…maybe in of in similar usages, I will collect

examples and talk through very briefly in the classroom as a group.

How often do you correct their errors?

Most weeks I’ll micro-mark a few paragraphs of their writing. For the important assessed work

I’ll probably mark two pages of the first draft of the assignment.

Do you think that helps?

Definitely. Without correction they wouldn’t normally realise that they ‘ve used the article in

the wrong way and after I’ve picked it up they normally ask questions and try to reason…find

the correct way and they ask questions. OK, no I wouldn’t be able to evaluate how effective

this practice this. But I think in my opinion they do get better over time.

In question 4 you were asked for recommendations/suggestions about helping students

improve their accuracy of English Articles in writing….(do you have any further

comments?).

I gave a reference to a book which I’d recommend. Really, it’s also about talking to them about

their expectations. They can improve but they shouldn’t aim for perfect – that’s really unlikely

to happen.

295

Interview Tutor #2

Question 2 asked about how often you teach the article system to your students. Can you

talk a little more about when you teach them and how you teach them? When are you

most likely to teach article use?

Yes I do. So, this year I had a really high level group at the 10-week stage and I didn’t do

some things I normally do. But then I realised these students still had to do a presentation and

I often teach articles related to speaking and presentation skills. And yes, these students are

high level, but they still needed to stress articles in their speaking more, and I think in future

I’ll do it like I normally do.

In writing it’s really complicated. What I’ve found is…in a way some of them are very clear

about articles and …this is always the case…if you give them a gap-fill they would get it right

most of the time, but in their own writing it’s another thing. I think it’s because the flow is

interrupted too many times, so they don’t feel it.

So, thinking about what I normally do….In the 15-week Presessional I did this - I had one

session which was..like a 45 minute session where we look specifically at articles and then

there were two or three times when it came up and we had 5-ten minutes each time. And then

there’s the ARP feedback, so maybe I help them with articles 5 or 6 times on a longer

presessional. But I didn’t do this so much with the 10-week students.

Right, how do you do that?

OK…so…we had…we used the academic skills coursebook and there was a reference I think

in the extra materials to proofreading. It comes up really late in the materials, apart from the

Unit 1 focus on noun phrases. I think there was a proofreading exercise at the end. There was

also that poster like sheet explaining things that we can give them to download. It was helpful-

ish. But for me this EAP is a big change, I’m new to it. For me, most of my teaching in the

past 5 years has been heavily European so for an awful lot of articles it’s identical usage. They

just overuse the definite articles in generalisations. So I normally use that old Headway thing

– give them something to look at and then look through a text to think about why they’re there.

Then they have information about usage that they then read through. So it’s more awareness

raising than anything else.

Question 3 asked about how likely you’d be to highlight the errors/give feedback in the

example. In general, do you give corrective feedback? How often and when do you give

corrective feedback? How do you do it?

So on your questionnaire I put not sure for the first one (do you correct article errors during in

class writing?) because it would really depend what the focus of the in class writing was. If

it’s got nothing to do with the focus, no I probably wouldn’t. For the homework one…not sure

but likely. It would depend again. But for more regular writing tasks, it may depend on the

student. If they normally are good at them, then I would mention errors. But if they are terrible

296

all the time I tend to mark errors less. For the ARP and high stakes writing, I have to admit

that I mark article errors. I highlight the words and write ‘articles’ on areas where there are

problems. But I only do that for a few pages, then I tell them ‘that’s for you to now go and

find other errors in the rest of your work’. For the last one – when co-presenting with another

person – it would depend on the personality. If that person wants everything to be perfect, I’d

tell them.

OK, last question. In question 4 you were asked for recommendations/suggestions about

helping students improve their accuracy of English Articles in writing….(do you have

any further comments?). Do you have any more tips? Can we help students improve their

English article accuracy and if so, how?

OK, so from what I have….from what I’ve understood. OK look I’ve got my Diploma but I’ve

not done …I’m not someone who’s done a lot of reading on this. But from what I’ve

understood, articles are late acquired. So, to a certain extent, it’s one of those things that is

cyclical. It comes up at various points in what you’re teaching. It’s just a complicated thing.

So, what I’ve done in the past with …because I have taught Russians and East Europeans…so

for speaking, making it part of the rhythm. I teach it more in the speaking skills. You can

make them understand that there’s a space for the article, even by clapping it out. For example

‘as a result’…you clap on the syllable….include them in a group. The focus shouldn’t be on

‘why’, it’s just look it’s there in that chunk of language. As with everything, for some students

it really helps them, for some students it’s like water of a duck’s back. Another thing is that

box where is says known and unknown and in the next rows ‘listener knows’, ‘speaker

knows’…and students see they have to click both boxes to see they can use articles. It shows

them clearly that this thing needs to be known by both the reader and the writer to use the. And

if it’s not known by both reader and writer then it will be a or an. But it’s one of those things

that’s really hard to get them right. They make mistakes even at higher levels.

Interview Tutor # 3 (Appendix 3)

Question 2 asked about how often you teach the article system to your students. Can you

talk a little more about when you teach them and how you teach them? When are you

most likely to teach article use?

Not so much teach, but I ask students to focus on some article effects and then get them to

check their own work. In EAP it’s about getting them to notice language patterns in academic

contexts. And after that it’s also about peer marking and self-correction, more than teaching.

We have to get them to notice these patterns first…. Yes I do introduce it more formally at

297

the start. I often look at correct use first, maybe at how it works with definitions of

generalisations in the introduction, which is normally a plural form without article. This is

important because I can’t assume they’ve been taught it (article system) in the way I want them

to understand it. It fitted in quite nicely in the 10-week course this year at the beginning – I

brought forward the stuff about proofreading. But rather than focus on errrors, I first looked at

how correct use worked. Then I put a powerpoint slide up with a research article I’d chosen

but simplified a bit. I got them to focus on how articles determining specific examples at some

points and generalisations in others – I think I got this idea from Swales. Then I gave them

some proofreading exercises with correct and incorrect article use. Then I got them to start peer

marking their own work. I actually do that with every written task – they peer mark it every

time….before it comes to me. And that first time it was with aricles.

Question 3 asked about how likely you’d be to highlight the errors/give feedback in the

example. In general, do you give corrective feedback? How often and when do you give

corrective feedback? How do you do it? It’s just peer marking?

Yes, going back to your questionnaire, the only thing I think teachers should be micro-marking

themselves – because it’s so high stakes – is the ARP extended writing assessment. Otherwise

no. I think there’s a lot of teachers overmarking articles. The problem is, the student is just

going to depend on your spoonfeedinng. And to they actually learn? You’re not going to be

there in their main programme – they have to learn to do this themselves. And we have these

really amazing motivated students in EAP which is so great – I come from an ESOL

background where I spent so long trying to get students to take responsibility, but our

Presessional students get this more quickly.

Do you have any more tips? Can we help students improve their English article accuracy

and if so, how?

I think we can accelerate their progress, definitely, by helping them see the importance of

articles for moving us from non-specific to unspecific ideas, but they need to leave our course

ready to help themselves and learn how they do this in their own subject. We have lots of

independent learning resources for grammar and academic writing online – we can signpost

them to this in tutorials, but then turn it around and ask them to show you what they’ve been

working at on their own initiative in tutorials.

What place do you think grammar and a focus on form has in a Presessional EAP course?

Well my understanding of EAP is that we’re focussed on meaning in texts and I….I mean the

object is to help students understand the academic community and the expectations of this

community. OK, no we’re not language teachers and we shouldn’t be structuring what we do

around language form, but let’s be honest - it’s impossible to not teach some aspects of

298

grammar alongside the functions of academic language in a text. So when we teach academic

caution, I’m focussing students incidentally on the modal verbs, tense choice and adverbs in

the text that helps the writer hedge their language. How can we focus students on genre

differences of reflective writing and a lab report without focussing on how personal pronouns

are used in the former and greater use of passives might be found in the latter? Equally, I think

a focus on articles is useful when we look at definitions or generalisations.

Interview Tutor #4 (Appendix 3)