Investigating common errors in the use of positions of adverbials of time by non english major students at the university of science m a 60 14 10

Figure 2.1: Illustration of functions of a corpus-based program Figure 2.2: Screenshot of concordance lines with errors of prepositions Figure 2.3: Steps in the process of corpus-based e

INVESTIGATING COMMON ERRORS OF

POSITIONS OF ADVERBIALS

BY NON-ENGLISH-MAJOR STUDENTS

AT THE UNIVERSITY OF SCIENCE

A thesis submitted in partial fulfillment

of the requirements for the degree of

Master of Arts (TESOL)

Submitted by LƯƠNG UYÊN PHƯƠNG

Supervisor Assoc Prof - Dr ĐINH ĐIỀN

Ho Chi Minh City, December 2009

I certificate my authorship of the thesis submitted today entitled:

INVESTIGATING COMMON ERRORS OF POSITIONS OF ADVERBIALS BY NON-ENGLISH-MAJOR STUDENTS

AT THE UNIVERSITY OF SCIENCE

in terms of the statement of Requirements for Thesis in Master’s Programs issued

by the Higher Degree Committee This thesis has not been submitted for the award

of any degree or diploma in any other institution

Ho Chi Minh City, December 30, 2009

Lương Uyên Phương

I hereby state that I, LƯƠNG UYÊN PHƯƠNG, being a candidate for the degree of Master of Arts (TESOL) accepted the requirements of the University relating to the retention and use of Master’s Theses deposited in the Library

In terms of these conditions, I agree that the original of my thesis deposited

in the Library should be accessible for purposes of study and research, in accordance with the normal conditions established by the Library for the care, loan

or reproduction of theses

Ho Chi Minh City, December 30, 2009

Lương Uyên Phương

ACKNOWLEDGEMENTS

First of all, I wish to express my profound gratefulness to my supervisor, Associate Professor - Dr Dinh Dien, Lecturer of the Department of Comparative Linguistics – HCMC University of Social Sciences and Humanities; Lecturer of the Department of Computing Science – HCMC University of Science for his invaluable guidance, assistance and encouragement during the preparation and completion of this thesis

I am extremely grateful to the Head of English Department of the University

of Science - Dr Nguyen Hoang Tuan for his patience in reading the thesis again and again and clarifying my ideas

I would like to express my deep gratefulness to Dr Nguyen Thai An who willingly gave me valuable evaluations and comments on my project

Also, I must thank the students at the University of Science who wrote the compositions and particularly those who sat for the Diagnostic Test under such difficult circumstances

On a personal level, I am grateful to my husband and my parents who were there to share the ups and downs with me More especially, my little son has given

me strength and motivation to complete this thesis

Last but not least, never would this thesis have been accomplished without all those who helped me their handful hands in the research project: Mr Lưu Vĩnh Tấn, Ms Nguyễn Thị Xuyên and their precious remarks

ABSTRACT

This study is a corpus-based study of common errors in adverbial placement

of EFL students at the University of Science The conceptual motivation for this project is based on the ideas behind Error Analysis and Contrastive Analysis and on the role of learner corpora in relation to these two paradigms

The hypotheses to be tested in this study are:

1 There is no difference between the errors made by the students due to mother tongue interference and those due to difficulties of the target language

2 There is no statistically significant correlation at the 0.05 level between students’ scores on the diagnostic test and their scores in English writing essay in the study

A database of errors extracted from 100 written essays was typed and tagged for error tagging The tagged errors which extracted from the Corpus of learner language then compared in turn with occurrences in The Collins Cobuild Corpus -

a corpus composed of 56 million words of contemporary written and spoken text

of native speaker A diagnostic test was constructed from the students’ authentic written compositions added to the validity and reliability Then, the Correlation Matrix was computed to assess whether the means of two groups are statistically different from each other

The first hypothesis of the study was rejected The results showed that there was a signification difference between errors due to the students’ mother tongue interference and those due to difficulties of the target language

The second hypothesis was also rejected at 0.05 The results showed that there was significant correlation between the students’ achievement in English writing and their achievement in the diagnostic test of English adverbial positions

CERTIFICATE OF ORIGINAL i

RETENTION AND USE OF THESES iii

ACKNOWLEDGEMENTS iv

ABSTRACT v

TABLE OF CONTENTS vi

LIST OF FIGURES x

LIST OF TABLES xi

LIST OF ABBREVIATIONS xiii

Chapter 1 INTRODUCTION 1

1.1 BACKGROUND TO THE STUDY 1

1.2 AIMS OF THE STUDY 2

1.3 RESEARCH QUESTION 2

1.4 TERMINOLOGY 3

1.5 SIGNIFICANCE OF THE STUDY 5

1.6 OVERVIEW OF THE STUDY 6

Chapter 2 LITERATURE REVIEW 7

2.1 ERRORS ANALYSIS 7

2.1.1 LEARNER ERRORS 7

2.1.1.1 Errors versus mistakes 7

2.1.1.2 Sources and Causes of Errors .8

2.1.1.2.1 Interlingual transfer .8

2.1.1.2.2 Intralingual transfer .9

2.1.1.2.3 Context of learning 10

2.1.2 APPROACHES OF ERROR ANALYSIS 11

2.1.2.2 Traditional Error Analysis 11

2.1.2.2 Corpus-based approaches for error analysis 12

2.1.2.2.1 Contrastive Interlanguage Analysis (CIA) 13

2.1.2.2.4a Principles of error tagging 15

2.1.2.2.4b Illustrated examples for error tagging 15

2.1.2.3 Steps in the process of corpus-based error analysis 17

2.1.2.4 Error correction 20

2.2 THE POSITIONS OF ADVERBIALS IN AN ENGLISH SENTENCE 21

2.2.1 Basic points of the positions of adverbial in an English sentence 21

2.2.1.1 Initial position 22

2.2.1.2 Medial position 23

2.2.1.3 Final position 24

2.2.2 Exceptions related to the position of adverbials 25

2.2.2.1 Adverbials and arguments 25

2.2.2.2 Adverbials and subject in negative sentences and questions 27

2.2.2.3 Adverbials in gap constructions 28

2.2.2.4 The position of adverbials in multiple adverbial clauses 29

2.2.2.5 Order of importance vs order of time 30

2.3 POSITIONS OF ADVERBIALS IN A VIETNAMESE SENTENCE 31

2.3.1 Basic points of the positions of adverbial in an English sentence 31

2.3.2 Order of importance vs order of time 32

2.4 PREVIOUS STUDIES 34

2.5 SUMMARY 37

Chapter 3 METHODOLOGY 38

3.1 Research design 38

3.2 Subjects 38

3.3 Description of corpora 41

3.3.1 The Learner Corpus 41

3.4.2 Identifying errors 44

3.4.3 Describing errors 46

3.4.4 Designing a diagnostic test 47

3.4.5 Comparing and Explaining errors 51

3.4.6 Testing two hypotheses of the study 54

3.5 SUMMARY 55

Chapter 4 RESULTS AND DISCUSSIONS 56

4.1 RESULTS FOR HYPOTHESIS 1 56

4.1.1 Errors due to MTI 59

4.1.2 Errors due to DTL 60

4.2 ERROR ANALYSIS 61

4.2.1 MTI errors 61

4.2.2 DTL errors 65

4.3 RESULTS FOR HYPOTHESIS 2: 70

4.3.1 The frequency distribution of the students’ scores from the diagnostic test 71

4.3.2 The frequency distribution of the students’ writing scores from the their essays 72

4.3.3 Correlation between the students’ English mastery of adverbial positions

and their English writing proficiency 73

4.3 SUMMARY 75

Chapter 5 IMPLICATIONS AND RECOMMENDATIONS 76

5.1 Implications and Recommendations 76

5.2 Suggested classroom techniques in teaching adverbials 77

5.2.1 Teaching English adverbials through corpora 78

5.4 CONCLUSION 85

BIBLIOGRAPHY 87

APPENDICES

APPENDIX 2 Common errors of adverbial positions in randomly chosen

Figure 2.1: Illustration of functions of a corpus-based program

Figure 2.2: Screenshot of concordance lines with errors of prepositions

Figure 2.3: Steps in the process of corpus-based error analysis

Figure 2.4: Description of word categories and their frequencies of errors

Figure 2.5: Description of word categories and their frequencies of errors types Figure 2.6: Screenshot of types and explanations for tagged errors

Figure 2.7: Contrastive Interlanguage Analaysis

Chapter 3

Figure 3.1: Subjects’ age

Figure 3.2: Subjects’ gender

Figure 3.3: The place where the students attended high school

Figure 3.4: The time when the students started to learn English

Figure 3.5: Screenshot of concordance lines extracted from CUS

Figure 3.6: Screenshot of concordance lines of error tagging

Figure 3.7: Screenshot of error-correction process of CASEC-G software

Figure 3.8: Random Test Generator – PRO 8.3 screen shot

Figure 3.9: Test-Builder screenshot of Random Test Generator-PRO 8.3

Figure 3.10: Screenshot of searching the word “highly” followed with adverbs Figure 3.11: Screenshot of concordance lines of “highly” followed with adverbs Figure 3.12: Screenshot of a sample Correlation Matrix of SPC XL software Figure 3.13: Screenshot of a sample scatter chart of SPC XL software

Chapter 4

Figure 4.1: Distribution of errors in types of adverbial

Figure 4.2: Distribution of errors of English adverbial positions due to MTI

Figure 4.3: Distribution of errors of English adverbial positions due to DTL

Figure 4.4: Histogram of the frequency distribution of scores on the diagnostic test

Figure 5.1: Screenshot of a corpus-based task for teaching adverb “never”

Figure 5.2: Screenshot of concordance lines of “never”

Figure 5.3: Screenshot of a clip of adverb song

Figure 5.3: Screenshot of the song “My love affair with adverb”

Figure 5.4: Screenshot of the clip “How to parent like the President”

Figure 5.5: Screenshot of a video clip of adverbial phrases

Figure 5.6: Screenshot of a video clip of adverbial positions

Chapter 2

Table 2.1: The position of adverbials in multiple adverbial clause

Table 2.2: Illustrations extracted from The Da Vinci Code

Chapter 3

Table 3.1: Distribution of errors found in the students’ essays and later used as test items Table 3.2: POS tags used to search in Collins Cobuild Corpus

Chapter 4

Table 4.1: Test items removed from the list of the students’ common errors

Table 4.2: Summary statistics of the diagnostic test

Table 4.3: Summary statistics of the writing scores

Table 4.4: The pairwise correlation coefficient

Chapter 5

Table 5.1: The differences of task-based and corpus-based approach

CUS Corpus of learner language at the University of Science CCC Collins Cobuild Corpus

CLC Cambridge Learners’ Corpus

JEFLL Japanese EFL Learner Corpus

LLC Longman Learner Corpus

ICLE International Corpus of Learner English

CALL Computer-Aided Language Learning

L1 First Language / Native language

L2 Second Language / Foreign Language

SLA Second Language Acquisition

NS Native Speakers

NNS Non-native Speakers - examples taken from student essays

CA Error Analysis

CA Contrastive Analysis

CEA Computer-aided error analysis

CIA Contrastive interlanguage analysis

EFL English as a Foreign Language

ESOL English for Speakers of Other Languages

IL Interlanguage

MTI Mother tongue interference

DTL Difficulties of the Target Language

POS Part of Speech

* Unacceptable form

 Suggested correction

Chapter 1

INTRODUCTION

1.1 BACKGROUND TO THE STUDY

Learning a foreign language generally implies making errors in various areas, especially in grammar A very common type of error which learners make when producing grammatical structures involves the use of adverbs, specially misplacing them in the sentence A tricky problem for Vietnamese learners of English is to know where to put adverbials in relation to the verb, as can be seen in these examples taken from undergraduate essays of Corpus of learner language at the University of Science (CUS):

*…some of them even do not know how much money they have to pay

* I even never opened the letter

*I hope I can speak fluently English

The problem with lumping all of these very diverse words into one category

is that it can make learning the rules about how to use them seem complicated From the actuality of teaching and learning as well as the approach to current issues

in the domain of adverbial which are widely discussed nowadays, the author of the present study can realize that there are some subjective ones in the process of teaching and learning this grammatical category As Chomsky said, “adverbials are

a rich and as yet relatively unexplored system.” [Chomsky, 1965: 219] This is true not only for non-majored-English students but also for English majors

In fact, the question of adverbial placement in learner language has received considerable attention from researchers in Second Language Acquisition, but there

is a relative lack of corpus-based studies To complement elicited data, it is useful to have information about spontaneous production of adverbials in learner language Given the infrequent nature of unsolicited adverb use, access to corpus data is the only feasible way of collecting enough occurrences to reveal possible patterns in the spontaneous placement of adverbs

1.2 AIMS OF THE STUDY

Descriptively, the study was designed to diagnose and investigate common

errors and their possible causes in the use of English adverbials as well as to determine whether or not there exists a relationship between learners’ achievement

in English writing on the one hand and their achievement in a diagnostic test focused on English adverbials on the other Those errors were then compared in turn with occurrences in the Collins Cobuild Corpus (CCC) Through that finding,

it shows a comparison of adverb placement in the written productions of EFL learners of the University of Science and in those of native speakers

Theoretically, the study aims to find out two hypotheses: (1) there is no

difference between the errors made by the students due to mother tongue interference (abbreviated to MTI) and those due to difficulties of the target language

(abbreviated to DTL); (2) there is no statistically significance correlation between

the score given to the students’ diagnostic tests and the score given to their essays

written in English

Practically, this study is based on corpus linguistic, especially of the

application “error tagging” for error analysis in the hope that they can act as something new for language research in general For specific purposes, in order to help students to overcome difficulties and master the use of English adverbials, the author of the thesis collected examples of errors of adverbial position from the corpus of learner to determine the frequency of occurrence, and, more importantly,

if these errors are persistent problems that warrants targeting in the classroom, it is useful in design practical and applicable solutions for teaching

In order to clarify the matters related to this question and to support the general purposes, the study also deals with some more sub-questions:

- What are the causes which were behind the errors that the students made?

- Can the students’ use of English adverbial as observed from the diagnostic test be seen through their ability to write essays in English?

- What pedagogical implications for teachers can be drawn from the findings

of this study to help students deal with such errors?

1.4 TERMINOLOGY

1.4.1 Adverbial

Richards and Platt [1992: 9] note that the word adverb refers to single words

that modify verbs, adjectives, and adverbs Adverbs answer the questions how, when, where, and why, and most of them modify or describe actions, states, or

qualities Any phrase or clause that is used as an adverb is called adverbial

1.4.2 Learner corpora

Learner corpora, also called inter-language (IL) or L2 corpora, are electronic

collections of authentic foreign or second language data There is an increasing interest in learner corpora both as a pedagogical tool, as well as a research tool Learner corpora have been used mostly to provide information on learners’ common errors, especially useful when annotated with the help of a standardized system of error tags “They are also characterized by a high rate of misuse, i.e orthographic, lexical, and grammatical errors, as in the Longman Dictionary of Common Errors (based on the Longman Learner Corpus) for a state of the art account of the use of learner corpus for teaching purposes.” [Nesselhauf, 2004: 18]

1.4.3 Concordance

A concordance is a list of occurrences of a particular word, part of a word or combination of words, in its contexts drawn from a text corpus The search word is sometimes also referred to as key word The most common way of displaying a

concordance is by a series of lines with the keyword in context [Bernhard Kettemann, 2006]

1.4.4 Concordancing is a method of analyzing language by studying structures found in effective communication [Garry N Dyck, 1999] It is also a descriptive method, but instead of focusing on the data of a single expression as in the introspective approach, the expression is shown in a large number of contexts and examined quantitatively In this way, meaning and grammar can be discovered by an examination of patterns in the examples In contrast to the introspective method which focuses on the teacher’s intuition, there is a greater focus on the data in this method

1.4.5 Concordancer

A concordancer is a software program that is used to analyze corpora and list the results [Daniel Krieger, 2003] “The concordancer can find a selected word and list sentences or portions of sentences containing that word, called key word in context It can also identify collocations or words most often found together with the key word…” [Garry N Dyckn, 1999] providing students with information on

patterns in sample sentences of real language

1.4.6 Token words

Token words are the word forms that are counted according to their occurrence in the corpus In this example: “I am reading a book”, there are 5 tokens

1.4.7 Error tagging

Error tagging system is the construction of an error annotation system

includes the design of a taxonomy of errors alongside its pertinent tags, which are to

be inserted in the learner corpus manually [Izumi et al., 2004: 35]

1.5 SIGNIFICANCE OF THE STUDY

In Vietnam, there is almost been no scholar work developing deeply the

issues related to English adverbials In such a situation, theoretically, the study will

make a contribution to the theoretical studies of Vietnamese linguistic on the one hand and to provide implications for the teaching and learning on the other hand By creating our own corpus and identifying and classifying errors in our database, we are able to design pedagogical materials which are more “locally” oriented for learners in a particular context

Practically, this corpus-based study not only gives “learners who receive

only corrective feedback” but also compares with occurrences in the Collins Cobuild Corpus which is composed of 56 million words of contemporary written and spoken text of native speaker in the hope to “go a long way to improve their target language writing style” It helps students to get a general and systematic view

on English adverbials in order to understand firmly and use them more confidently, not merely for the sake of passing examinations, but also to achieve higher communicative competence with the English language

Furthermore, this corpus-based approach provides new insights into the way the language operates opening new perspectives of grammar of the target language This characteristic is also very useful for teachers in teaching foreign languages and for linguists in doing contrastive studies

Last but not least, the study also offers a new model named “error tagging”

to search for particular errors and find plenty of examples of error description and specificity to use on English written material by Vietnamese learners so that the linguistics can apply it to build other corpora in other languages to serve a variety of fields Moreover, this corpus analysis opens the gates of a new era for language research

1.6 OVERVIEW OF THE STUDY

The structure of the dissertation is organized in chapters, sections and subsections In detail, Chapter 1 discusses the practical background, the aim and

research question as well as terminologies related to the thesis

Chapter 2 presents an overview of the literature and the research relevant to this study It consists of three main parts: (i) error analysis including limitations of traditional error analysis and the application of corpus-based error analysis, especially of error-tagging; (ii) the positions of adverbials in an English sentence and (iii) positions of adverbials in a Vietnamese sentence

Chapter 3 is the research design with a description of the student subjects as well as the two corpora, instruments and the collection procedures These parts are the core of the present study They are concerned with the steps of a corpus – based error analysis In detail, corpus-based methodology is used to investigate data, concordances and tagging tools are used to provide semantic profiles of specific words highlighting differences, and error analysis is used to identify these common errors

Chapter 4 discusses and analyzes the findings, seeking for satisfactory answers for the four research questions and two null hypotheses

Chapter 5 draws conclusion about the implications of the findings and suggests some techniques and methods of teaching English adverbials to enhance students’ ability to use them precisely, especially of adverbial placement Furthermore, the limitations of this thesis and the further recommended are also stated out in the last chapter

Chapter 2 LITERATURE REVIEW

This chapter reviews the literature relevant to the study of the M.A thesis The four main and distinguishing parts are needed to be involved:

(1) Error analysis (abbreviated to EA) consisting of the definitions of errors and mistakes, limitations of traditional error analysis as well as corpus-based approaches

to error analysis with the process of error tagging, successive steps in EA research and possible sources of errors;

(2) Basic points and exceptions of the adverbial positions in an English sentence; (3) Basic points of the adverbial positions in an English sentence;

(4) Previous studies of adverbial positions

2.1 ERRORS ANALYSIS

2.1.1 Learner Errors

2.1.1.1 Errors versus mistakes

Corder introduced the distinction between errors (in competence) and mistakes (in performance) An error is produced when the learner lacks linguistic knowledge or competence about a rule or the system of the second language This can be seen in texts where a specific type of error occurs several times A

“mistake”, on the other hand, is non-systematic and appears more randomly It does not necessarily indicate the knowledge of a learner, but rather problems in production Mistakes are most common in speech but can also be found in writing

This study is based on errors, but since it is difficult to rule out mistakes, some mistakes will be sorted under errors The examples from the material will be cited as they were written, without corrections, except a few cases when slight corrections are needed to make sense of the sentence Each type of error will contain a description of the error, followed by examples and a brief summary of the instances of that error found in the material

2.1.1.2 Sources of Errors

Learners make errors in both comprehension and production Corder [1974: 25] has pointed out: "It is very difficult to assign the cause of failures in comprehension to an inadequate knowledge of a particular syntactic feature of a

misunderstood utterance"

However, based on the magnitude of errors which includes errors at different levels such as a phoneme, morpheme, word, sentence or paragraph, Richards [1984: 123] divides errors into two kinds: global errors and local errors According to him,

“global errors” are those which involve the “overall structure of the sentence” Therefore, it can hinder the communication and prevent the message from being understood On the contrary, local errors are those which affect “a particular constituent” of the sentence and they don’t prevent the message from being comprehended because the hearer can still guess the intended meaning easily

Basing on the source of errors, Richards [1974: 206], Jack C Richard, John Platt, Heidi Platt [1992: 127-128] supposed that there are two types of errors: interlingual and intralingual errors The former is caused by the interference of the mother tongue while the latter is the result of interference within the target language

2.1.1.2.1 Interlingual transfer

Wilkins observes:

"When learning a foreign language an individual already knows his mother tongue, and it is this which he attempts to transfer The transfer may prove to be justified because the structure of the two languages is similar - in that case we get

“positive transfer” or “facilitation” - or it may prove unjustified because the structure of the two languages are different - in that case we get “negative transfer” - or “interference”

[Wilkins, 1972: 199] Interlingual errors are those which result from language transfer, that is, which are caused by the learners’ native language For example, Vietnamese students may say that “Yesterday, I visit Hanoi capital” In this sentence, they forget to put the verb

“visit” in the past tense It is easy to understand because in Vietnamese, when

talking about the past tense, they do not have to conjugate Instead, they just use some adverbs of time such as “Yesterday, last year, last month, two years ago, etc”, which is enough to describe the past tense In contrast, intralingual errors are those which result from faulty or partial learning of the target language, rather than from language transfer In other words, it reflects the learners’ competence at a particular stage

Richards [1971] focuses on several types of errors which do not derive from

transfer from the mother tongue These kinds of errors are called intralingual and

Intralingual errors are errors include false analogy (e.g boy and boys vs child and

*childs), misanalysis, incomplete rule application (under generalization), exploiting redundancy, overlooking co-occurrence restrictions, hypercorrection and overgeneralization [185-187] These errors include the use of holistic strategies (e.g Students do not find the required form, so they try to use another near-equivalent second language item which they have learnt) For them there are four main types of errors including omissions, additions, misformation and misordering

To be more specific, they point out the definition of each type of error as

followed: (1) Omissions: the absence of an item that must appear in a well-formed

utterance (e.g in early stages of learning, the omission of function words rather than

content words) [p.107] (2) Additions: the presence of an item that must not appear

in well-formed utterances (e.g failure to delete certain items: He doesn’t know*s

me) [p.107] (3) Misformations: the use of wrong form of the morpheme or structure (sometimes called misselections [p.108-109] (4) Misorderings: the incorrect

placement of a morpheme or group of morphemes in an utterance (e.g The

misplacement of adverbials, interrogatives and adjectives) (p.110) [Cited in Ellis,

1995, p.56]

In general, an intralingual error may be caused by the influence of on target language item upon another while developmental errors occur when the learner attempts to build up hypothesis about the target language on the basis of limited experience” In this thesis, intralingual errors are also named errors due to

“difficulties of the target language” (abbreviated to DTL)

2.1.1.2.3 Context of learning

One more kind of sources of errors is context of learning which refers to “the

social situation.”[Brown, 1980: 174] The students make these errors due to a

“misleading explanation from the teacher, faulty presentation of a structure or word

in the textbook” or due to “a pattern that was rotely memorized in a drill but not probably contextualized” However, this kind of source has not been explored and identified easily in this thesis, except for an interview that somehow makes it clearer

In short, there are many different ways to classify the errors students of second language make In this paper, the researcher chooses the way of classifying the types of errors by the authors Richards [1974] and Wilkins as mentioned above

as criteria for analysis of common errors in the placement of adverbials Although

“it is very difficult to distinguish transfer and intralingual errors, it is not easy to recognize various kinds of intralingual error the author of the thesis had an attempt

to identify all possible sources of the error identified in this M.A thesis, those which

have finally been divided into two main kinds: MTI errors, i.e interlingual errors and DTL errors, i.e intralingual errors

2.1.2 APPROACHES OF ERROR ANALYSIS

2.1.2.1 Traditional Error Analysis (TEA)

Although it is widely recognized that error analysis (abbreviated as EA) contributes to describing learner language and the improving second language pedagogy, several problems and limitations have been pointed out mainly because a concrete methodology of EA has not been established yet

Most importantly, EA cannot be successful without robust error typology, which is often very difficult to obtain Since it used to be difficult to collect or access large databases of learner language, a robust error typology that covers almost all error types was not established in traditional EA Another criticism against EA is that errors reflect only one side of learner language A lot of people point out that if a researcher analyzes only errors and neglects what learners can do correctly, he/she will fail to capture the entire picture of learner language It is time-consuming to count both correct and incorrect usages in learner data, and this must have been quite difficult to do in the past before computing technology was developed

Furthermore, the real significance of EA cannot be identified without using diachronic data in order to describe learners’ developmental stages The types and frequencies of errors change with each acquisition phase Without longitudinal data

of learner language, it is difficult to obtain a reliable result by EA

In short, the problems and limitations of traditional EA are mainly due to the deficiency of computing technology and the lack of large databases in early times However, now that computing technology has advanced, and a lot of learner data is available, it might be possible to perform EA more effectively mainly by annotating errors And “if you are a student of English, a teacher, a translator or if you are writing in English, analysing English, or have any questions about how English works, a corpus-based approach can be of great benefit to you.” [James Thomas, 2008:33]

2.1.2.2 Corpus-based approaches for error analysis

The plural form of corpus is usually “corpora” Theoretically, corpora should

be (C)apable (O)f (R)epresenting (P)otentially (U)nlimited (S)elections of texts In fact, the term CORPUS can be derived from its features it implies:

C : Compatible to computers

O : Operational in research and application

R : Representative of the source language

P : Processable by both man and machine

U : Unlimited in amount of data, and

S : Synchronic in formation and representation [James Thomas, 2008: 35] Language corpora usually represent a large collection of representative samples obtained from texts covering different varieties of language used in various domains of linguistic activities A corpus helps us to understand more about the language and see how people use it when they speak and when they write According to the website of Cambridge University Press at the link http://www.cambridge.org/elt/corpus/what_is_a_corpus.htm, it can give us answers

to questions such as these below

Figure 2.1: Illustration of functions of a corpus-based program

The use of computer tools has allowed researchers to handle vast corpora, to gain better insight into authentic learner language at different levels – lexis,

grammar, and discourse It gives a new dimension to both traditional EA and Contrastive Analysis (abbreviated as CA) [Pravec, 2002; Granger, 2003] As Granger [2002: 11-12] points out, linguistic exploitation of learner corpora may

involve one of the following two methodological approaches: (i) contrastive

interlanguage analysis (CIA), involving quantitative and qualitative comparisons

between (a) native and non-native data or (b) different varieties of non-native data,

from learners with different mother tongue [Granger, 1996; Gilquin, 2001], and (ii)

computer-aided error analysis (CEA), focusing on errors in IL and using computer

tools to retrieve them

2.1.2.2.1 Contrastive Interlanguage Analysis (CIA)

Learner corpus research has concentrated on Contrastive Interlanguage Analysis (CIA), which involves two types of comparison – 1) native productions (NS) vs non-native productions (NNS) to highlight the features of non-nativeness

in the learner language; 2) two or more varieties of NNS to determine whether native features are limited to one group of learners, in which case they are most probably transfer-related phenomena, or whether they are shared by several groups

non-of learners, which would point to a developmental or inter-language issue

2.1.2.2.2 Computer-Aided Error Analysis (CEA)

Computer-aided Error Analysis (CEA) has led to a much more limited number

of publications than CIA due to the cost of manual error annotation Apart from articles describing error tagging systems, there are a few articles covering certain specific error categories including lexical errors (Man-Lai et al., 1994; Källkvist, 1995; Lenko-Szymanska, 2003), tense errors (Granger, 1999; Fitzpatrick and Seegmiller, 2004) and a more recent article (Neff et al., 2007) covering the range of error types in the ICLE corpus from Spain These analyses offer great promise for identifying the sources of error (L1 interference, features of novice writing, limited vocabulary and language structure, etc.) so the need to annotate for error and to reduce the cost of annotation by automating where possible is great

2.1.2.2.3 Bringing CIA and CEA together

Computer-aided error analysis (CEA) and contrastive inter-language analysis (CIA) can of course be integrated Granger [1999], for instance, compared errors concerning the use of verbs in post-intermediate and advanced students, showing that the progress rate was lower for certain subcategories (i.e tense errors) than for others (e.g auxiliaries errors); she also found that a large number

of tense errors were transfer-related

Similarly, Altenberg and Granger [2001] analysed errors in the use of the

high-frequency and polysemous verb to make by Swedish and French

learners, showing important correlations between errors and L1

The studies mentioned above and in the previous sections are highly heterogeneous and disconnected, which implies that their results are often not directly comparable Some general findings, however, seem to emerge [Nesselhauf 2004: 134-135] For instance, that the learners’ L1 has a great influence on their

L2 (which is why some researchers have started adding corpora of the learners’

L1 into their analyses [Tono 2004, Borin and Prütz 2004], that learner writing is often less qualified than NS, and that it tends to be more speech-like According to Nesselhauf, the lack of well-defined findings is due to the relative newness of learner corpus research, which means that general results will hopefully emerge when more and complementary studies are carried out

2.1.2.2.4 Error Tagging system for learner corpora

One of the methodologies that can be applied to corpus-based error analysis

is error tagging [Granger, 2002: 98] Error tagging system is the construction of an error annotation system includes the design of a taxonomy of errors alongside its pertinent tags, which are to be inserted in the learner corpus manually [Izumi et al., 2004: 35] It includes tag-associated error categories arranged on a menu-driven interface which the user can select and insert in the text as he/she revises the learner material, and decides on the nature of the error confronted In addition to error tags, correction or reconstructions of the target version are often inserted in the tagging

process Once errors are fully tagged, error tags can be retrieved with the aid of software retrieval tools and analysed quantitatively and qualitatively according to the researcher’s interest

Error tagging is a highly time-and labor-consuming task At the same time, a corpus annotated for error provides an invaluable resource for SLA research and practice For researchers, errors can reveal much about the process by which L2 is acquired and the kinds of strategies the learners use in that process For language instructors, errors can give hints about the extent to which learners have acquired the language system and what they still need to learn Finally, for learners themselves, access to the data marked for error provides important feedback for improvement

2.1.2.2.4a Principles of error tagging

According to Granger [2003: 467], for an error annotation system to be fully

effective, it should be: (1) informative but manageable: it should be detailed enough

to provide useful information on learner errors, but not so detailed that it becomes unmanageable for the annotator;

(2) reusable: the categories should be general enough to be used for a variety

of languages;

(3) flexible: it should allow for addition or deletion of tags at the annotation

stage and for quick and versatile retrieval at the post-annotation stage;

(4) consistent: to ensure maximum consistency between the annotators,

detailed descriptions of the error categories and error tagging principles should be included in an error tagging manual

2.1.2.2.4b Illustrated examples for error tagging

There are two ways to annotate learner data for error One approach is to reconstruct the correct form [Fitzpatrick and Seegmiller, 2001] The other approach

is to mark different types of errors with special tags [Granger, 2003] The former is used for developing instructional materials that can provide (automatic) feedback to

learners; the latter is used for SLA research to compare type of error and error frequency among different learners at different levels of language development The present study is belong to the latter one In order to make it clearer, Cambridge Learner Error Coding system is an illustrated example of this type of error tagging

Cambridge Learner Error Coding system, a unique feature of the CLC is that over 30 million words or about 95,000 scripts, have been coded with a Learner Error Coding system devised by Cambridge University Press This means that we can see which words or structures produce the most errors in Learner English It also means that we can search for particular errors and always find plenty of examples

Here is what a Cambridge University Press author would see if they wanted

to find out where learners make the errors of missing out a preposition The words

in red are the prepositions that the learners should have used but didn't and <#MT>

is the code for a missing preposition The concordance lines of learner text around the statistics window show some common errors like these:

“…*I would like to suggest you that it is extremely important…”

“…*but she explained me that she was in London…”

Figure 2.2: Screenshot of concordance lines with errors of prepositions

As can be seen from the Figure 2.2, the grey statistics window which is on top of the lines of text shows which preposition is missed out most frequently - it was omitted 197 times in a sample of 1000 cites The program could equally show which words most frequently have a preposition omitted after them In this sample

“explain” was the word where students had most frequently omitted the preposition

2.1.2.3 Steps in the process of corpus-based error analysis

Degneaux, et al.[1998] call EA based on learner corpora “computer-aided error analysis [CEA]”, and expect that the rapid progress of computing technology and learner corpora will be able to solve the problems and overcome the limitations

of traditional EA Surely, thanks to the quantitative database of learner language, we will become able to cover a wider range of learner errors Advances in computing technology make it possible to perform statistical analysis with quantitative data more easily

However, it must be noted that human researchers still have a lot of work to

do in the same manner as in traditional EA, such as establishing an error typology for error tagging or examining results obtained from CEA carefully In the other words, “the basic motivations concerning with language teaching” [Corder, 1974:

28] for error annotation and the steps of the corpus-based error analysis are the

same as those of traditional EA, but they are different from the ways to work That

is why we still choose Corder’s theory with five steps in error analysis research in order to reach that objective These steps are:

Figure 2.3: Steps in the process of corpus-based error analysis

In the first step, three broad types of error analysis according to the size of

the sample have to be identified These types are: massive, specific and incidental

samples All of them are relevant in the corpus collection but the relative utility and

proficiency of each varies in relation to the main goal In other words, in this first step, the researcher has to be aware of her research, and the main objective of this

stage is selecting a proper collection system The first type of sample, massive,

mentioned involves collecting several samples of language use from a large number

of learners in order to compile a comprehensive list of errors, representative of the

entire population A specific sample consists of one sample of language used, collected from a limited number of learners Then, an incidental sample uses only

one sample of language provided to a single learner In practice, the most common samples used by researchers are specific and incidental in order to avoid the difficult task of processing, organizing and evaluating the large quantities of samples taken in a massive sample collection

In the second step, identifying errors, it is necessary to localize errors by

pointing out which letters, words, and phrases, or how sentence structures or word order, are incorrect

Figure 2.4: Description of word categories and their frequencies of tagged errors

Thirdly, identified errors should be described by being linguistically

categorized depending on, for example, their POS (part-of-speech), linguistic level (morpheme, syntax, lexis, or discourse), or how they deviate from the correct usage

on the surface structure (redundancy, omission, or replacement)

Figure 2.5: Description of frequencies and types of tagged errors

The next step is “explaining errors” It means identifying why those errors

occurred This is a very important task in order to figure out the learners’ cognitive stage Some causes of learner errors have been recognized in common such as errors caused by language transfer, learning and communication strategy-based errors, and the transfer of training and induced errors

Figure 2.6: Screenshot of types and explanations for tagged errors

Finally, errors are evaluated This can be done by estimating intelligibility or

near-nativeness of erroneous outputs In other words, “error gravity” is estimated by examining how each error interferes with the intelligibility of the entire outputs

In short, although the basic motivations for error annotation are the same as

those of traditional EA, such as describing learner language and improving language pedagogy, several new applications of EA might become possible such as the development of a new computer-aided language learning (CALL) environment that can process learners’ erroneous input and give feedback automatically

second-2.1.2.4 Error correction

To many scholars, error correction plays a significant role in improving learners’ accuracy in language learning especially in L2 writing, which is grammatically demanding As Brown [1994: 219] states, “one of the keys to successful learning lies in the feedback that a learner receives from others”, when giving feedback to learners’ written work, “teachers normally focus on correcting the wrong use of basic vocabulary, grammatical forms, spelling and punctuation to make the written work acceptable” [Bartram and Walton, 1991: 84]

However, Cohen [1990: 117] claims that the evaluation is “partial” since it mainly focuses on the “low-level” accuracy, but ignores the “higher-level” style, such as appropriate word dictions, native-like organizations of the whole writing That is to say, learners who receive only corrective feedback still need to go a long way to improve their target language writing style Actually, learners with a certain level in the target language have the intention to produce natural target language writing and have a stronger desire for evaluation on this aspect They are not satisfied with their errors being corrected, but also want to know how to rework their expression to make it sound natural [Bartram & Walton, 1991]

Error correction treatment options can be classified in a number of possible ways with “basic options” and ‘possible features” within each option Discussing in brief about this issue, Brown [1994: 222] suggests, “the teacher needs to develop

the intuition, though experience and solid theoretical foundations, for ascertaining which option or combination of options is appropriate at given moment.”

Based on this theory of Brown, Bartram, Walton and Cohen, the present study not only gives “learners who receive only corrective feedback” but also compares with occurrences in the Collins Cobuild Corpus which is composed of 56 million words of contemporary written and spoken text of native speaker in the hope to “go a long way to improve their target language writing style”

2.2 THE POSITIONS OF ADVERBIALS IN AN ENGLISH SENTENCE

2.2.1 Basic points of the positions of adverbial in an English sentence

According to Angeda Downing and Philip Locke [1995, 557], “adverbials occupy fixed positions, these varying with their function As adjuncts or disjuncts they are more mobile, as the following examples show:

Legally, the man couldn’t have been sent to prison

The man legally couldn’t have been sent to prison

The man couldn’t legally have been sent to prison

The man couldn’t have legally been sent to prison

The man couldn’t have been legally sent to prison

The man couldn’t have been sent legally to prison

The man couldn’t have been sent to prison legally

[Angeda Downing and Philip Locke, 1995: 558] However, these two authors also state that “not all adverbials are equally mobile The choice of position is determined by its type (circumstantial, modal, degree, ect.), the scope of its meaning (whole clause or part of a clause), and the general information structure of the clause…” [Angeda Downing and Philip Locke, 1995: 558]

Allsop [1985:244] also mentions some of the cases that adverbs show a fair amount of variability in placement, but are subject to strict ordering restrictions For

example, from (1b) it can be seen that already can appear before the finite auxiliary

has (although this order is slightly marked), and from (1c) it is clear that probably

can appear after has

(1) a John has already left

b John already has left

c John has probably left

d John probably has left

However, although the two adverbs can co-occur in a single clause, as shown in

(2a), they cannot appear in the order already-probably Their relative order is strict

(2) a John probably has already left

b *John already has probably left

or

(3) a Howard will probably already have been finishing up by then

b * Howard will already probably have been finishing up by then

Due to “the choice of position is determined by its type (circumstantial, modal, degree, ect.)” [Allsop, 1985: 225], the positions of adverbials are often presented according to their types of meanings expressed There are “six board types of meaning of adverbials in clauses and groups: circumstantial, process, modal, degree, focusing, conjunctive Each of the six main types is made up of several subtypes, which are exemplified by lexical exponents” [Angeda Downing and Philip Locke, 1995: 551] Beside the basic error to avoid is that of putting an adverb between a transitive verb and its direct complement as the following examples (4a) and (4b), “adverbs can usually be placed in three different positions

in a sentence: initial, medial and final.” [Firsten and Killian, 1994: 224]

(4) a He ate the chocolate cake greedily

b *He ate greedily the chocolate cake

2.2.1.1 Initial position

When an adverb is placed at the beginning of a clause, its meaning extends to the whole clause and not simply to the predication or to an element of the predication In this position, the meaning may be of two boards of kinds:

(i) It functions as an element within the clause and has the same status as the other

elements, though referring to all of them together, as in:

(5) a Slowly, the rising sun appeared over the distant horizon

b Suddenly, a fresh breeze began to ruffle the surface of the sea

In this way, the scope of the adverb ranges more widely over the clause than it would do in medial or final position:

(6) a The rising sun slowly appeared over the distant horizon

b The rising sun appeared slowly over the distant horizon

[As Angeda Downing and Philip Locke,1995: 561]

(ii) Other adverbs occur in initial position but are considered, both syntactically and

semantically, to be outside the clause which expresses either the speaker’s attitude

to what he is saying Set off by one or more commas, these adverbials are a comment on the entire sentence

(7) Frankly, I don’t believe you

(8) Hopefully, the new plan will lead to some improvements

(9) He neglected to tell her, unfortunately

Specially, when negative frequency adverbs such as never and rarely are

used at the beginning of a sentence, the sentence structure needs to be rearranged to accommodate auxiliary-subject inversion:

(10) Never have I seen such a crowd!

(11) Rarely do we invite so many people

2.2.1.2 Medial position

An adverb may appear between the obligatory initial and final clausal elements

(i) Pre-verbal position, e.g between the subject and the verb phrase:

(12) She actually expects to marry him

(13) Marry in her own way was a darling

(ii) After the modal or first auxiliary verb and before the lexical verb:

(14) You may in this way be of great assistance

(15) He had often dreamed about coming back

(16) The utilization of computers is not of course limited to business

(iii) After the lexical verb but before other obligatory elements:

(17) It is no longer a casino

(18) He is certainly/ without doubt an expert

(19) He deals cleverly with her clients

2.2.1.3 Final position

Adverbs of time and definite frequency (e.g last week, every year), adverbs of manner which focus on how something is done (e.g well, slowly, evenly) and adverbs of place (e.g in the countryside, at the window) usually go in final position Sentences (20a) through (20f) show adverbs of various types in sentence-final position:

(20) a He left the room quickly Manner

b She enjoyed the party tremendously Degree

c Tom was a doctor for many years Duration

d He comes in late sometimes Frequency

e Tom believed the man crazy after questioning him Time

f Tom put his watch where he can find it in the dark Place

In brief, some usages of adverbs make it possible to place the adverb in several locations in the sentence In other cases, there is only one location where the adverb can be placed in the sentence without dramatically altering its meaning

2.2.2 Exceptions related to the position of adverbials

Many authors have not only defined adverbs but also described their position

in the English sentence; many of them agree on the fact that adverbs are rather complex structures because of their various classifications, meanings and positions

in the sentence This complexity is due to the syntactical and semantic behavior of the adverb which is determined by its position in the sentence Adverb position may not only change the meaning of the sentence but also make the sentence grammatical or ungrammatical [Carter, Hughes, and McCarthy 2000; Bing 1989; Parrot 2000; Raimes 2001; Swan 2006; Celce-Murcia and D Larsen-Freeman 1999] These authors exemplify the complexity of adverb usage by giving the following considerations regarding adverb positions in the sentence:

2.2.2.1 Adverbials and arguments

2.2.2.1a Adverbials and Finite Verbs/Auxiliaries

In English adverb placement relative to a finite verb depends on the type of verb, lexical verb vs auxiliary The choice between pre- and post-auxiliary adverb positioning seems to be influenced by stress An unstressed auxiliary normally precedes an adverb; the sequence adverb - auxiliary, by contrast, is most natural if the auxiliary is stressed [Baker 1971, 1981, 1991, and Wilder 1997]

(21) a John is often in his office [Wilder, 1997: 327]

b John often is in his office [Wilder, 1997: 328]

However, a modal verb following a frequency adverb may only receive a deontic reading; if the modal is to be interpreted as epistemic as in (22b), it has to precede the adverb

(22) a Charles (frequently) must (frequently) see the doctor

b Sue (*frequently) must (frequently) have hit the dog

The linear order of an adverb and a finite auxiliary with cliticized negation unambiguously reflects their scopal relation: an adverb may only be interpreted as taking scope over negation if it precedes the negation marker, whereas the adverb

receives a narrow scope reading if placed behind the negation marker (23a) expresses that there was a regularity of John's not taking his medicine while it is stated in (23b) that there was no regularity in John's taking his medicine

(23) a John regularly didn't take his medicine

i) “It was regularly the case that John did not take his medicine.” ii) # “It is not the case that John took his medicine regularly.”

b John didn't regularly take his medicine

i) #“It was regularly the case that John did not take his medicine.” ii) “It is not the case that John took his medicine regularly.”

[Rizzi, 1997: 312] Similarly, linearization among several medial adverbs corresponds to scope Frequency and subject-oriented adverbs, for example, may co-occur in either order with distinct interpretations: what is considered to be wise of Richard in (24a) is the frequency of his calls, whereas every single call is judged as wise in (24b) Other combinations of adverbs are restricted in their ordering Like the epistemic modal in (25b), an epistemic adverb has to precede a frequency or subject-oriented adverb, while it follows an evaluative one

(24) a Richard wisely has often called his rich aunt

b Richard often has wisely called his rich aunt

(25) a Ed probably has (wisely / often) called Bob

b Ed (*wisely / *often) has probably called Bob

[Rizzi, 1997: 310]

2.2.2.1b P ositions of Adverbs and non-finite verbs / auxiliaries

While adverbs may always follow a finite auxiliary in English, the acceptability of adverb placement behind a non-finite auxiliary depends on the type

of adverb For example, epistemic and evaluative sentence adverbs cannot follow a non-finite auxiliary, let alone a non-finite lexical verb By contrast, subject-oriented

and frequency adverbs may appear behind a non-finite auxiliary Note that adverb

placement in front of the auxiliary and behind it as in (30) gives rise to different

interpretations: (30a) states that the event of their being knocked off their feet occurred frequently while (30b) describes one event in which they were knocked off their feet many times during one training session [Ernst, 2002:99]

(26) *Oskar had been luckily leaving the office at the time

(27) John (probably) has (probably) been (*probably) elected (*probably)

by the majority of the candidates

[Ernst, 2002: 100]

(28) She has been wisely insisting on total control of her films

(29) All during that time, they had been cleverly attending every church

meeting they could find, as a way of bolstering their social standing

[Ernst, 2002: 106]

(30) a They have frequently been knocked off their feet during training

b They have been frequently knocked off their feet during training

[Ernst, 2002: 348]

2.2.2.2 A dverbials and subject in negative sentences and questions

In English, an adverb cannot precede the subject in questions (31), inverted conditionals (32), and Negative Inversion constructions (33) while it may occur in pre-subject position in simple declarative clauses

(31) a Yesterday, Bill gave a book to Mary

b.*What did yesterday Bill give to Mary?

c John claimed that yesterday Bill gave a book to Mary

d.*I wonder what yesterday Bill gave to Mary

(32) a If yesterday John had done that,

b.*Had yesterday John done that,

(33) a Not only did John usually do the dishes, he also cleaned the windows

b.*Not only did usually John do the dishes, he also cleaned the

windows

(34) a Bill claimed that under no circumstances would Susan regularly go to

school