Mestrado

Educação Multimédia

TESE

Connecting Corpora to Learner StyleTo what extent is the effectiveness of an online corpus-based approach to

grammar learning dependent on whether students prefer to learn grammar deductively or inductively?

Jonathan LewisFaculdade de Ciências, Universidade do Porto, Portugal

Orientador: Professor Doutor Duarte Costa Pereira

1

Table Of Contents Page List of Figures……………………………………………………………………........4List of Tables……………………………………………………………………..…...5Acknowledgements………………………………………………………………........6Abstract……………………………………………………………………………..…7Resumo……………………………………………………………………………..…9

Chapter 1: Introduction………………………………………………………….11

Chapter 2: Theoretical Perspectives: A Review of the Literature…….......19 Section 1: Developments in notions of grammar and approaches to grammar learning………….............................................202.1.1 Historical framework……………………………………………….….202.1.2 Deductive versus Inductive Approaches to Grammar Learning…….…26 Section 2: Developments in Corpora and Concordancing………...….. 302.2.1 What is a corpus?.....................................................................................302.2.2 What is a concordancer?..........................................................................312.2.3 Concordancing software…………………………………………….….322.2.4 Types of corpora………………………………………………………..332.2.5 Applications of corpora………………………………….……………..342.2.6 Advantages of using corpora in language learning………………….…352.2.7 Disadvantages of using corpora in language learning……………….…352.2.8 Applications of corpora in language learning: for the teacher………... 362.2.9 Applications of corpora in language learning: for the learner………....382.2.10 Lack of Research……………………………………………………….412.2.11 Online Applications of Corpora………………………………………. 43 Section 3: Developments in Learning Style theory…………………….452.3.1 Overview of factors involved in second language learning……….……452.3.2 Language Aptitude……………………………………………………...462.3.3 Cognitive Style and Learning Style……………………………….…….482.3.4 Gregorc……………………………………………………………….....512.3.5 Kolb……………………………………………………………………..532.3.6 Willing…………………………………………………………………..542.3.7 Skehan…………………………………………………………………..562.3.8 Myers-Briggs……………………………………………………………572.3.9 Felder-Silverman………………………………………………………..582.3.10 McCarthy………………………………………………………………..592.3.11 Gardner………………………………………………………………….602.3.12 Practical Application of Learning Style Theory……….………………..602.3.13 Learning Style Theory Applied to CALL………………………….……62

Chapter 3: The Experiment………………………………………..……………..633.1 Background………………………………………………………….......643.2 Aims……………………………………………………………………..653.3 Initial Experiment……………………………………………………….66

2

3.4 The Target Language……………………………………………………673.4.1 Linking Adverbials……………………………………………………...683.4.2 Subordinators……………………………………………………………693.4.3 Prepositions……………………………………………….……………..713.5 Method………………………………………………………………….723.6 The Concordancer………………………………………………………723.7 The Subjects…………………………………………………………….733.8 The Questionnaire………………………………………………………743.9 The Pre-test……………………………………………………………...753.10 The Inductive Lesson…………………………………………………...763.11 The Deductive Lesson…………………………………………………..893.12 The Post-test…………………………………………………………….89

Chapter 4: Results…………………………………………………………………904.1 Post-test results………………………………………………………….914.2 Statistical Analysis……………………………………………………...934.2.1 Correlations……………………………………………………………..934.2.2 Regression………………………………………………………………944.2.3 T-Test…………………………………………………………………...964.3 Post-project Questionnaire……………………………………………...97

Chapter 5: Conclusions………………………………………………...………...1015.1 Conclusions…………………………………………………………….1015.2 Future research…………………………………………………………105

Bibliography………………………………………………………………………….107

Appendix 1 Learning Style Questionnaire………………………………………….113Appendix 2 Pre-test....................................................................................................115Appendix 3 Inductive Approach printed handout…………………………………..117Appendix 4 Deductive Approach printed handout………………………………….123Appendix 5 Post-test………………………………………………………………...129Appendix 6 Post-project Questionnaire……………………………………………..131

3

List of Figures

page

Fig. 1 Screenshot of concordance lines drawn from the Compleat Lexical Tutor’s Online Concordancer ………………………………………………………………..13

Fig 2 Diagram identifying the key factors or variables which are believed to be essential to second language learning and suggesting probable ways in which the sets of different variables may interact…………………………………45

Fig. 3 Results of Learning Style questionnaire…………………………………….…….....75

Fig. 4 Average totals scored in the pre-test by each of the four groups (learning style / approach)………………………………………………...…..76.Fig. 5 Screenshot of ‘Set Up’ page from Inductive Mode……………….……………...….77

Fig. 6 Screenshot of BNC Concordancer Interface………………………………………....78

Fig. 7 Box of linkers to classify, from Exercise 1 of the Inductive Approach…………..….79

Fig. 8 Screenshot of chart from Ex.1 of the Inductive Approach, containing code of grammar items to enter into the concordancer………………………………….…80

Fig. 9 Screenshot of concordances for ‘during’…………………………………………..…81

Fig 10. Screenshot of Interface for Exercise 2 of the Inductive approach, made using Hot Potatoes software………………………………………………………………...82

Fig 11. Screenshot of Interface for Ex. 4 of the Inductive approach, made using Hot Potatoes………………………………………………………………………,,…83

Fig 12. Box of linkers from Exercise 5 of the Inductive Approach…………………………..84

Fig. 13 Screenshot of chart from Ex.5 of the Inductive Approach, containing codes of grammar items to enter into the concordancer……………………………………..…85

Fig. 14 Screenshot of concordances for ‘Still’ as a conjunct………….……………………...86

Fig. 15 Screenshot of completed chart for Ex.5 and table of codes for Ex.6…………………86

Fig.16 Screenshot of concordances for ‘though’ as a conjunct ……………………………..87

Fig. 17 Screenshot of interface for Ex. 7, made using Hot Potatoes software………………..88

Fig. 18 Screenshot of interface for Ex. 8, made using Hot Potatoes software………………..88

Fig. 19 Screenshot of information about conjuncts from Ex. 1 of the deductive mode……...89

Fig. 20 Chart showing the average scores obtained by each group in the post-test compared to those in the pre-test……………………………………………...……....91

4

Fig. 21 Charts showing to what extent the post-test scores obtained by students in each group were higher than the pre-test………………………………...…….……..91

Fig. 22 Charts showing results of Qu. How easy was the concordancer tool to use?...............98

Fig. 23 Charts showing results of Qu. How useful did you find the concordancer as a means of studying grammar?..................................................................................98

Fig. 24 Charts showing results of Qu. How easy was it for you to identify patterns of grammar and work out ‘rules’ using the concordancer tool?....................................99

Fig. 25 Charts showing results of Qu. Would you have also liked to have the grammar rules given to you in a more traditional and straight-forward way? ………………….99

Fig. 26 Charts showing results of Qu. Would you like to study grammar in this way again in the future? …………………………………………………………………..100

List of Tables

Table 1. Linkers studied in the virtual lesson…………………………………………………67

Table. 2 Completed chart from Exercise 1 of the Inductive Approach………….……………79

Table 3. Chart to be completed in Exercise 5 of the Inductive Approach…………………….84

Table 4: SPSS Correlations table showing strength of relationship between pre/post test results and matching learner style/approach……………………………93

Table 5: SPSS Model Summary table showing results of simple linear regression analysis.…94

Table 6: SMSS ANOVA, or F-Test, table, showing analysis of variation between the means in each group………………………………………………………...……95

Table 7: SPSS Coefficients table showing estimated regression coefficients…………………95

Table 8: SPSS T-Test table showing the descriptive statistics for the two groups, that where learner style matches approach and that where it does not……………….96

Table 9: SPSS Independent Samples Test……………………………………………………..96

5

Acknowledgements

I wish to offer thanks to the following people for their help and assistance:

To begin with, to my ‘orientador’ Professor Doutor Duarte Costa Pereira, who provided

much needed orientation at the outset of the project and guidance whenever it was

needed.

Secondly, to Professora Doutora Belinda Maia at FLUP, who also gave much needed

advice and guidance in the initial stages and the loan of several key books to kick-start

the research.

Thirdly, to Professora Doutora Susan Howcroft at the Universidade de Aveiro, whose

advice and input I greatly appreciated.

Special thanks must be extended also to my great friend and colleague Mestre Alan

Dawber for whose loan of his own master’s thesis I will be forever grateful.

Thanks also to the library staff at FLUP who were very helpful whenever I needed to

borrow or buy books for the research.

Finally, I wish also to thank all the students who participated so willingly in this project

and without whose help it would not have been possible.

6

Abstract

This thesis describes a research project carried out at the Faculdade de Letras da

Universidade do Porto (FLUP) which focused on applying an inductive corpus-based

approach to grammar learning; specifically, to the study of the differences between

concessive connectors such as even so, even though and despite.

The project took inspiration from Tim John’s ideas about ‘Data Driven Learning’

(1994) and his belief that hands-on concordancing allows the learner to interact with the

corpora data and thereby discover rules and patterns in the language for himself.

The experiment, carried out on 56 third year students of English at FLUP, aimed to

find out whether advanced learners’ preferred way of studying grammar - the extent to

which they employ a deductive or inductive learning style (Felder, 1995) - would

influence how successfully they assimilated the target language using a corpus-based,

guided-discovery approach.

After completing a questionnaire to determine their learning style, the subjects were

divided into two groups; those favouring a more traditional, explicit rule-based

approach to grammar learning, and those who showed a greater tendency to induce rules

from natural contexts. Each of these groups was then subdivided such that half studied

the target language with the aid of corpora and were encouraged to work out rules for

themselves by analysing concordances, while the other half studied the same grammar

items via a purely deductive approach.

A website was set up to provide all the instructions and additional practice exercises

for both modes of instruction. The inductive version used a concordancer program

linked to the BNC.

All the subjects did the same pre- and post-test as well as a post-experiment

questionnaire in order to yield both quantitative and qualitative data.

7

After statistical analysis had been performed, results revealed that there was a

‘significant’ link between students’ learning style and the approach they followed. In

the groups where learning style matched the approach, subjects generally did better in

the post-test than in the pre-test, while in the groups where there was a mismatch

between style and approach the opposite was true.

The study thus raises questions about how effective hands-on concordancing is for

many students, and shows that teachers need to tailor computer-based methods of

instruction according to their students’ preferred way of learning as well as provide the

necessary training if concordancing is to be used as a tool to facilitate grammar study.

8

Resumo

Esta tese descreve um projecto de pesquisa levado a cabo pela Faculdade de

Letras da Universidade do Porto (FLUP) que se focou na aplicação base de uma

vertente indutiva usando os corpora na aprendizagem gramatical; especificamente no

estudo das diferenças entre adjuntos, conjunções e preposições concessivas como even

so, even though e despite.

Este projecto foi inspirado nas ideias de Tim Johns (1994) sobre ‘Data Driven

Learning’ e na sua crença de que o uso interactivo dos corpora permite que o aluno

interaja com as concordâncias e que desta forma descubra por si mesmo regras e

padrões na gramática da língua em aprendizagem.

A experiência levada a cabo com 56 alunos do 3º ano de Inglês da FLUP, teve

como objectivo perceber como as preferências na forma de estudo da gramática por

alunos mais avançados – empregando um estilo dedutivo ou um estilo indutivo (Felder,

1995) - iria influenciar o seu sucesso na assimilação da gramática em aprendizagem

usando como base uma aproximação indutiva guiada.

Após completar um questionário para determinação do seu estilo de

aprendizagem, os alunos foram divididos em dois grupos; os que favoreciam o estilo

tradicional de aprendizagem, baseado numa aproximação baseada em regras, e os que

mostraram uma maior tendência para induzir as regras a partir do seu contexto natural.

Cada um destes grupos foi então subdividido de forma a que metade estudasse a

gramática com a vertente corpórea e fossem encorajados a identificar as regras por si

mesmos pela análise de concordâncias, enquanto que a outra metade estudava os

mesmos itens gramaticais pela via puramente dedutiva.

Foi providenciado um site para fornecimento de todos os prefácios e exercícios

práticos adicionais para ambos os grupos. O grupo que usou a via indutiva, teve ainda

acesso a um programa com ligação à BNC.

9

Todos os alunos foram submetidos aos mesmos exames, anteriores e posteriores

à experiência, bem como ao mesmo questionário após a experiência, de forma a serem

obtidos resultados com informação quantitativa e qualificativa.

Após ser efectuada uma análise estatística, os resultados revelaram uma ligação

significativa entre o estilo de estudo dos alunos e a via de aprendizagem por eles

seguida. Nos grupos onde o estilo de estudo se completava com a via seguida, os alunos

na sua generalidade tiveram melhores resultados no teste apresentado após a experiência

do que no teste previamente apresentado, enquanto que os grupos onde não havia

concordância entre o estilo e a via o resultado foi inverso.

Desta forma, o estudo levanta questões sobre a eficácia para muitos alunos da

vertente interactiva usando os corpora e revela a necessidade dos professores talharem

métodos-base computorizados de instruções, de acordo com as preferências da forma da

aprendizagem dos seus alunos, bem como providenciar a formação necessária se o uso

de concordâncias for para ser utilizada como ferramenta para facilitação do estudo

gramatical.

10

Chapter 1: Introduction

This chapter serves to introduce the two key areas of study in this thesis,

namely concordancing as a tool to aid language learning and learning style

as a key factor in determining how best to adapt methodology to suit

individual learner preferences. This leads to a definition of the problem

followed by a brief description of the structure and contents of the thesis.

11

Here now in the 21st century, everyday language use is so tied to technology that

language learning through technology has become a fact of life with important

implications for language teachers.

Writing about communicative competence in the new century, Rassool (quoted in

Chapelle, 2001) points out:

In a world increasingly driven by (a) the need for innovation through research and development, (b) the multileveled changes brought about in our everyday lives as a result of the nature and speed of technological developments, (c) the volume and range of information available, and its open accessibility, (d) the multimodal features of electronic text as well as (e) its interactive nature, we require significantly more than just the ability to read and write in a functional way. (Rassool, 1999)

Therefore, language teachers are becoming increasingly aware of the need to grasp the

nature of the evolving multimedia-based tasks which learners can engage in to aid the

language learning process. Such activities range from the many Computer Assisted

Language Learning (CALL) exercises available both on and offline, to the harnessing of

the Internet itself as a means of practising the skills of reading and listening; from the

application of email and weblogs for the purpose of developing writing skills, to the use

of specific electronic tools such as online dictionaries and translators.

One electronic tool in particular which has seen considerable development over the past

two decades is the concordancer. Put simply, it enables the user to access a corpus of

texts (plural, corpora) that is, a database of texts held in electronic form on a computer

or on the web in order to carry out acts of linguistic research. Using a concordancer, one

can search for individual instances of the use of a specific word or phrase, by keying in

the desired word/expression in much the same way as with a regular search engine.

Most concordancing programs will then search the corpus and any instances of the

selected word found therein will be presented as a list of lines, known as concordances,

often with the keyword in the middle, as shown in the screenshot below.

12

Fig. 1 Screenshot of concordance lines drawn from the Compleat Lexical Tutor’s Online Concordancer.

Concordancing tools were first developed in the field of corpus linguistics, where they

were used to conduct research into how language actually works in reality in order to

facilitate the compiling of new dictionaries and grammars with examples taken directly

from the corpora.

However, in the early 1980s, with the advent of cheap microcomputers, language

teachers, many of whom were already involved in the world of CALL, realised how

concordancing could benefit language learners, and began developing computer-assisted

concordancer activities. It was argued, by Higgins and Johns in particular (1984), that

such activities empowered the learner to investigate questions of vocabulary use and

grammatical collocation on their own. For example, the screenshot above shows a

selection of concordances for depend revealing that this verb frequently collocates with

the preposition on and upon (and not in as many students might have supposed). A

student studying this data can also see which adverbs are commonly associated with the

pattern depend...on e.g. heavily, importantly and largely.

13

Nowadays, concordancers are a common tool of the linguist’s trade, especially used by

translators. Language teachers and learners are also using them, though many have yet

to appreciate fully the benefits offered by the interactive, student-centred nature of

corpus-based study.

There are various reasons that explain this reluctance on the part of teachers to use

corpora. Firstly, until recently the technology available in schools and universities

lacked the computer power to handle any sizeable corpus. Secondly, teachers have felt

that corpora analysis could be a source of overload and confusion for learners rather

than enlightenment, given the rather unfriendly appearance of a page of concordance

lines. Lastly, the academic nature of the language found in many of the texts held within

corpora has been seen as something that might put students off analysing the data to

work out patterns of use.

However, the tide appears to be turning. Concordancing interfaces have become more

user-friendly allowing the user to refine the search more effectively and select a

particular corpus reflecting a specific style or genre of language. Materials writers are

also beginning to compile the kind of language corpora which are more appropriate for

students’ needs, with highly academic texts removed.

The idea of hands-on concordance activities for learners is a natural extension of trends

in applied linguistics, with increased emphasis on task-based learning (Prabhu, 1987).

Moreover, student concordancing seems a clear application of constructivist ideas about

learning (Bednar, Cunningham, Duffy, and Perry, 1991), whereby learners are

characterized as "scientists," in this case linguists or lexicographers, using modified

scientific research tools to handle raw rather than pre-encoded data. In addition, the

analogy between searching a corpus with a concordance and browsing the Internet with

a search engine is of topical relevance to students.

In the literature, corpus and concordance are regularly described as the most promising current idea, for example Leech and Candlin (1986); Clarke (1992); Hanson-Smith (1993). Leech (1997) announces that at the turn of the century it seems evident that

14

something like a ‘corpus revolution’ is changing dramatically the way that language is taught, even if the effects of these changes are felt gradually. Concordancing is the centrefold idea in a paradigm-shift within CALL from computer as "magister" to computer as "pedagogue" (Higgins, 1988), from a tutor-dominated, process-control model of language instruction, to an "information resource" model, where the learner is free to explore the language for himself, leaving the instructor in the role of providing tools and resources for doing so.

English language teachers at the Faculdade de Letras, Universidade do Porto have been

using concordancers to help students study grammar, particularly in the 3rd and 4th year

of their course when they deal with text grammar and features of text cohesion.

The present study focuses on one such application of concordancing to the study of the differing grammatical properties of a variety of linkers, particularly those expressing concession e.g. however, although and despite.

The study looks specifically at how students’ post-test results following an online lesson using a concordancer may be attributed to their preferred learning styles, specifically whether they favour a deductive or inductive approach to grammar study.

One of the most significant developments in second language acquisition theory has

come from research on individual differences, notably the contributions made by Brown

(1994) and Skehan (1989, 1998). A large number of learning style models can be found

in the literature (detailed in Chapter 2.3), most of which fall into general categories such

as information processing, personality patterns, and social interaction.

All learners have individual attributes relating to their learning processes. Some people

may rely heavily on visual presentation; others may prefer spoken language; still others

may respond better to hands-on activities. It is evident that people learn differently and

at different paces because of their biological and psychological differences (Reiff 1992).

Language teachers, just like all other educators, have to bear in mind that people differ

consistently from each other in their preferences, whether emotional or environmental,

15

and in the way they process information. This awareness should necessarily impact on

their own teaching methods.

According to Brown (1994), when students’ learning styles are matched with

appropriate approaches in teaching, then their motivation, performances, and

achievements will increase and be enhanced. Bedford (2004) likewise points out that

matching or mismatching students' learning styles with instructional techniques affects

learning significantly. Felder and Henriques (1995) also state that ‘mismatches often

occur between the learning styles of students in a language class and the teaching style

of the instructor with unfortunate effects on the quality of the students’ learning and

their attitude toward the class and the subject.’

In his own model of learning style, Felder (1988) classifies students into various

categories (detailed in Chapter 2.3.9), including inductive learners (who prefer

presentations that proceed from the specific to the general) and deductive learners (who

prefer presentations that go from the general to the specific).

This distinction is particularly important for language teachers to be aware of, especially

in their consideration of how to present grammar. Traditionally, grammar teaching has

followed a deductive approach with rules being given explicitly, followed by practice

activities. However, in recent years, following the ideas of Krashen (1982) about the

difference between ‘acquisition’ and ‘learning’ (detailed in Chapter 2.1), and no doubt

influenced also by the research into learner differences, more inductive means of

presenting grammar have gained favour. Students are now often encouraged to explore

the language via pre-selected texts or lists of examples demonstrating a particular

language pattern, and discover the rules by themselves.

This brings us back to the issue of concordancers in language learning. The use of

concordancing is very much an inductive approach to language study since corpus data

provide the opportunity for the learner to infer rules and work out patterns of use.

As CALL developed in the 1980s, researchers looked at the role of individual learner

differences on the effectiveness of different instructional approaches. Investigating

learning style and task variables in CALL, Abraham (1985) found that field-

independent (i.e. more analytic) learners performed better on post-tests when they had

16

used a deductive, rule-based approach to presentation and field-dependent (i.e. more

holistic) learners performed better after using a more inductive approach. However,

similar studies to measure the impact of learning style on the effectiveness of corpus-

based approaches, as opposed to more traditional deductive approaches, are lacking in

the literature.

The present study is therefore an attempt to shed some light on this area. The problem it

addresses can be defined as: to what extent is the effectiveness of an online corpus-

based approach to grammar learning dependent on whether students prefer to learn

grammar deductively or inductively?

Chapter 2 presents theoretical perspectives on the various topic areas involved in this

thesis so as to provide an appropriate background framework for the experiment.

This review of the literature begins with a summary of the developments in notions of

grammar and approaches to grammar learning and how these have been influenced by

current theories about language and language learning. In particular, this section serves

to demonstrate the extent to which different language learning methodologies have been

deductive or inductive in their approaches to grammar.

There follows a summary of developments in corpora and concordancing, beginning

with a general overview of the uses of corpora in linguistic research and then focusing

on how concordancing has been applied to language learning. Advantages and

disadvantages of corpus-based EFL activities are examined and applications for both

teachers and learners are reviewed. Finally, there is a look at current online applications.

Chapter 2 concludes with a summary of developments in theories regarding individual

differences, beginning with a brief look at language aptitude before then focusing on

learning style theory. Various models of learning style are reviewed and their

application to language learning examined.

Chapter 3 outlines the experiment itself, providing first an account of the background

circumstances that led to the project before presenting its aims and a description of the

methodology used as well as a profile of the subjects. The target language area is

17

described. Information is given about the learning style questionnaire and the pre-test

and results of each are presented. There then follows a detailed description of the

inductive, corpus-based approach, incorporating screenshots to illustrate the various

stages of the online lesson.

Chapter 4 presents the results of the post-test with accompanying statistical analyis.

Results of the post-project questionnaire are also given.

Chapter 5 offers a conclusion based on the results of the experiment and discusses

lessons to be drawn and plans for future research.

The thesis concludes with a bibliography and appendices, which include the various

tests and questionnaires.

18

Chapter 2: Theoretical Perspectives:A Review of the Literature

This chapter is divided into three sections. Section 1 provides an overviewof developments in linguistics and how these have impacted on grammar

teaching/learning. Section 2 describes the growth of corpora in appliedlinguistics and applications of concordancing to language teaching/learning.

Section 3 traces developments in theories about individual differences,outlines various learning style models and discusses how learning style applies

to language teaching/learning, in particular to computer-assisted learning.

19

Chapter 2: Section 1 Developments in notions of grammar and approaches to grammar learning

2.1.1 Historical framework

Until the early 20th century language learning was dominated by Latin and Greek and

for centuries the study of a language meant chiefly the study of its grammar, with

written language regarded as the model for linguistic correctness. Typical language

lessons would begin with an explicit statement of the rule followed by exercises

involving translation in and out of the mother tongue. This so-called ‘grammar-

translation’ method was thus purely deductive with grammar treated in a wholly

prescriptive way.

Attitudes to grammar and grammar teaching were revolutionised by the development of

structural linguistics. This movement began with Ferdinand de Sausserre who sought to

highlight and analyse the systematic nature of language structure. He appreciated that

speech, rather than written language, must be the starting point for a valid description of

how people use language. He recognised that it is the combination and arrangement of

phonemes into systematic patterns that allows meaning to be conveyed between

speakers and referred to the shared understanding of the language system among a

language community as the “langue”. From this basis structural linguists such as

Leonard Bloomfield attempted to make linguistics into a true science, by studying the

observable patterns and forms of a language. With them was born the descriptive

approach, the idea of describing each language in its own terms rather than according to

a classical model. The reason for the name ‘structuralist’ was not because they tried to

find a general structure of language but because they tried to describe the whole of a

given language by simply making a comprehensive list of categories of patterns which

each had a particular structure. Structural linguists significantly developed the study of

phonology and morphology according to the principle of relationships between

meaningful units.

20

Language teaching methodology in the first half of the 20 th century deliberately turned

away from explicit grammar focus but, nevertheless, tended to follow a syllabus of

grammar structures, often labelled under functional labels.

In the 1950s and 60s, the Audio-Lingual Method was even stricter in its rejection of

explicit grammar teaching. Audiolingualism derived its theoretical base chiefly from the

work of Skinner, who found the lists of sentence patterns describing a particular

language convenient to his behaviourist view of language learning. Language was

considered as simply a form of behaviour, to be learned through the formation of correct

habits, a process in which rules played no part. The Audiolingual syllabus consisted of a

graded list of sentence patterns, which, although not necessarily labelled as such, were

grammatical in origin and formed the basis of practice drills. Brooks saw ‘control of the

structures of sound, form and order in the new language’ as the essential aim of such an

approach. (Brooks, 1964).

Where psychology and linguistics had been separate before, Noam Chomsky brought

them together by asserting that the nature of language and the nature of language

acquisition were two parts of the same question. This new perspective led to the birth of

psycholinguistics, a new science to study language acquisition. Chomsky reacted

against Skinner’s behaviouristic view of language learning, seeking to show that all

human language has a universal structure which may reflect the structure of the human

mind. Language is not a set of stimulus-response associations but the result of rules

used creatively to generate unique sentences. Chomsky devised the concept of

‘transformational grammar’, that is those rules which allow transformations of

declarative sentences into their negative, interrogative and especially passive forms. The

number of transformational rules in each language is finite but the number of sentences

that can be produced is infinite. According to Chomsky, a given language has two

distinct levels of organisation – deep structure and surface structure. One deep structure

can lead to many varying surface structures, and that while two surface structures may

appear similar, they are different at deep level.

It is certain that the Chomsky revolution in linguistics produced a significant change in

language teaching and a return to a more cognitive approach to language learning. No

complete methodology emerged as it had done with behaviourism, the nearest to one

21

being the cognitive code method whereby rules are taught explicitly and language

regarded most definitely as a system.

The work of M.A.K. Halliday (1973) has further revolutionised approaches to grammar

and grammar teaching. While Chomsky views language from ‘underneath’ and from

there into the mind, Halliday looks at it from ‘on top’ and from there into language as a

social system. He sees language as being made up of a system of grammatical choices

and a system of semantic choices. Hallidayan grammar has influenced sociolinguistics

through its concept of the three functions of language - ideational, interpersonal and

textual - according to which the features of every sentence can be analysed. Halliday’s

ideas have led to an increased awareness of the social dimension of language use,

especially in the area of cohesion in discourse.

By the 1970s descriptive approaches to grammar were in the ascendancy. Structural

linguistics had asserted the primacy of speech, leading sociolinguists such as Trudgill

(1974) to question the pre-eminent status of standard forms of pronunciation and of the

written language. Moreover, humanistic teaching emphasised self-expression and

discovery learning and in mother tongue teaching, traditional grammar lessons gave

way to creative writing and focus on fluency.

At the same time, the sociolinguist Hymes was developing the notion of

‘communicative competence’, a concept with not only a linguistic but also a

sociolinguistic dimension, seen especially in the ability to use and interpret language

appropriately in specific contexts (Hymes, 1972). This notion appealed to those, such as

Van Ek, Richterick and Wilkins, seeking to develop foreign language teaching

syllabuses that reflected language in use (Wilkins 1976). There followed a number of

functional/notional syllabuses with a division between ‘stuctures’ and ‘functions’ with

the former often influenced by the latter

Such progress in the field of sociolinguistics led also to the development in the 1970s of

Communicative Language Teaching (CLT), which aimed to replicate in the classroom

the purposes and contexts vital to real communication. In its purest form CLT rejected

grammar instruction since it was suggested that accurate grammar was not a necessary

22

requirement of successful communication. However, CLT-influenced syllabuses did not

fully reject grammar and it was often presented albeit under functional headings.

The movement away from explicit grammar teaching was hastened also by

developments in psycholinguistics. Chomsky’s claim (detailed in Lyons, 1970) in the

late 1950s that language ability is an innate human capacity and that we are ‘wired’ at

birth for language acquisition led to Krashen’s belief that formal language instruction

was unnecessary. His Natural Approach does away with both a grammar syllabus and

explicit rule-giving. Instead, learners are exposed to large doses of ‘comprehensible

input’ with the assumption that innate processes ‘in due course’ convert this into output.

Krashen’s Input Hypothesis (Krashen 1985) was influenced by studies showing that

second language acquisition follows similar regularities in the order in which

morphemes are acquired as does L1 acquisition. Krashen’s concept of ‘acquisition’ is

based on his idea of the ‘creative construction process’. A form of ‘hypotheses testing’

whereby learners acquire rules without consciously analysing them. Controversially, he

suggested that ‘learned language’ – that is rules which have been studied consciously –

cannot be absorbed into the learner’s ‘acquired competence’ – that body of language

that he can call upon automatically in rapid, fluent speech.

The vast growth of interest in English language teaching throughout the world in the

1970s and 80s, and the need for adequate theories to explain L2 acquisition, meant that

Krashen’s ideas were readily absorbed. Thus by the 1980s, grammar had lost the central

position that it once held, replaced by function-based or task-based syllabuses.

Grammar was learned inductively with learners often required to study examples and

work out rules for themselves. Chomsky (1987) himself says that grammatical

description belongs to the world of externalised (or E-) language, while what was more

important was the learner’s internalised (or I-) language, a language system that

developed ‘automatically’, uninfluenced by any well-intentioned parent or teacher.

However, a reaction against Krashen’s theories and the emphasis on I-language has

contributed to the revival of explicit grammar study. It is now suggested that a lack of

conscious attention to form may lead to the fossilisation of a learner’s language and

according to Tonkyn (1994), it is now widely held that formal grammar study is a

highly beneficial complement to informal learning, especially at higher levels of

23

proficiency. Tonkyn explains how it is now recognised that the learner’s role is less

passive than Krashen implies and that acquisition involves conscious processes,

particularly that of attention. Pointing out features of the grammar system is a form of

consciousness-raising, an approach encouraged by many such as Rutherford (1987),

which may lead in time to accurate and appropriate production. The researcher Richard

Schmidt (1990) questioned Krashen’s suggestion that what had been formally ‘learned’

could not pass into the ‘acquired’ system. Schmidt concludes that noticing is a

prerequisite for acquisition. Research by Long (1988) (detailed in Tonkyn (1994) also

appears to show that learners develop more quickly and go on learning for longer if they

are supported by instruction. Modern syllabuses now tend to incorporate a focus on

form alongside task-based activities. There are now also a great variety of grammar

reference and practice books to supplement general courses.

The 1980s saw revised interest in an updated study of the language system and current

trends in linguistics clearly influenced the descriptive grammars of Greenbaum and

Quirk (1985) and the more Hallidayan-based Downing and Locke (1992). The

development of a grammar based on the massive COBUILD database (Sinclair, 1987)

also led to further growth of interest in modern grammatical description.

The potential of concordancing programs to reveal how the language works has led to

new developments in the way language is viewed. Insights from language corpora into

collocations and how words combine into patterns led linguists, most notably Nattinger

to talk about ‘lexical phrases’ (Nattinger, 1992). Sinclair talks of ‘semi-preconstructed

phrases that constitute single choices, even though they might appear to be analysable

into segments.’ (Sinclair, 1988) A lexical approach to grammar was popularised by

Michael Lewis who asserts that ‘language consists not of traditional grammar and

vocabulary but often of multi-word prefabricated chunks’ (Lewis, 1997). Such language

‘chunks’ also include ‘functional phrases’ (Baigent, 1999). Vocabulary has traditionally

been seen in terms of individual words but this awareness of language chunking urges

language teachers to view vocabulary as units of meaning or ‘lexemes’ (Crystal, 1995).

Notions of grammar have also been challenged with the development of ‘pattern

grammar’. Hunston and Francis (1999) in particular have developed a corpus-driven

approach to the lexical grammar of English. A fundamental observation arising from

24

corpora analysis is the link between pattern and meaning. Sinclair states that if a word

has several senses, each sense will tend to be associated most frequently with a different

set of patterns, or ‘phraseology’. (Sinclair, 1988) According to Sinclair these

phraseologies should replace the word as the unit of vocabulary teaching. Indeed, he

argues that a view of language as phraseology necessitates the rejection of lexis and

grammar as separate entities. Hunston (2002) states that such an approach to grammar

‘prioritises the behaviour of each individual lexical item and patterns can be given for

both lexical words and grammatical words.’ Sinclair also argues that much of what

appears in spoken or written texts follows what he calls the ‘idiom principle’- that is,

each word in the text is used in a common phraseology and meaning is attached to the

whole phrase rather than to its constituent parts such that the hearer or reader

understands the phrase as a whole unit rather than as a grammatical template comprising

separate lexical items. When a stretch of text cannot be interpreted by the ‘idiom

principle’, the ‘open-choice principle’ is used – that is, when word-choice is constrained

only by the general grammatical rules of English. Thus words are either used in lexical

chunks or inserted into slots dictated by grammar. Pinker sums this up by stating that

‘the mind analyses language as some mixture of memorised chunks and rule-governed

assemblies’ (Pinker, 1999).

As yet, such lexical approaches to grammar have not given rise to a fully-fledged

language teaching methodology. Indeed, Harmer points out that ‘there are doubts about

how the learning of fixed lexical phrases can be incorporated into the understanding of a

language system’ (Harmer, 2001) or about the order in which such phrases should be

presented to learners. Certainly those such as Tim Johns who advocate ‘data-driven,

corpus-based learning’ (Johns, 1984) are already providing solutions to the problem of

how to expose learners to large quantities of the right kind of input. Corpus-based

concordancing technology now provides learners with versatile tools for exploring the

patterns of language. As far as language teaching is concerned, this new research has

stimulated discussion about what students should study. Harmer (2001) points out that a

lexical approach would ‘steer us away from an over-concentration on syntax and tense

usage.’ However, Lewis (1997) suggests that exposure to enough suitable input, not

formal teaching, is the ‘key to increasing the learner’s lexicon.’ Finally, one of the

original advocates of the lexical approach, Dave Willis, has set out the theoretical basis

for a corpus-based language syllabus (Willis, 1990) and has recently written a book –

25

Rules, Patterns and Words that illustrates different teaching techniques geared towards

promoting pattern grammar in the classroom. Willis advocates that in order to help

learners acquire the patterns of English they should follow a process of recognition,

system-building and exploration (Willis, 2003).

Section 2 shows how concordance-based approaches to language learning are providing

alternatives to the traditional formal grammar instruction of the language classroom and

that learners are being encouraged to discover the whole picture of the language for

themselves via access to corpora.

2.1.2 Deductive versus Inductive Approaches to Grammar Learning

Essentially we acquire knowledge either by being shown (learning from explanation), or

by finding out for ourselves (learning by discovery, and problem-solving). This applies

to language learning as much as it does to other skills, but with specific reference to

grammar learning Thornbury provides the following definitions:

A deductive approach starts with the presentation of a rule and is followed by examples in which the rule is applied.

An inductive approach starts with some examples from which a rule is inferred. (Thornbury, 1999)

He suggests that it may be easier to use the terms ‘rule-driven’ learning and ‘discovery’

learning respectively.

The purest form of deductive approach was that used by the Grammar-Translation

method whereby the teacher was the fount of knowledge and the students would learn

best by being taught about the grammar first of all and then putting this into practice

through exercises. Mental effort on the part of the student was not seen to be

particularly important, except insofar as that effort was needed to put the rule into

operation in a very limited way.

However, in the modern classroom a certain degree of deductive-style learning can be

beneficial. It can save time to present grammar rules explicitly as well as confirming

26

many students’ expectations about classroom learning, particularly those with an

analytical learning style.

Felder & Henriques (1995) argue that (2nd) language learning is

a largely conscious process that involves formal exposure to rules of syntax and semantics followed by specific application of the rules, with corrective feedback reinforcing correct usage.

They characterize the flow of the learning process from general to specific as a

deductive process.

However, as Thornbury states, grammar explanation encourages ‘a teacher-fronted,

transmission-style classroom’ and teacher explanation is often at the expense of student

involvement and interaction.

The rationale behind the ‘pure’ inductive approach, as used by the Audio-Lingual

method was that students should develop a series of language habits and that if enough

examples of a structural pattern were repeated, the students would automatically

internalise the common features of that pattern and be able to apply it by analogy to

other situations and contexts.

Such a ‘behaviourist’ view of language-learning assumed that just as a child induces

language patterns from the language around him, so foreign language learners could

also induce language patterns, especially if they were presented in situational contexts

with lots of examples.

It was assumed the mind played little or no part in the process, though it is now

recognised that it is precisely the student’s mental effort to make sense of the patterns in

the language that is key to successfully assimilating new language items.

Developments in educational theory led to the promotion of ‘discovery learning’,

essentially a form of guided-inductive learning whereby the teacher provides the learner

with examples of the target language and helps him to work out the grammar rules for

himself by consciously focussing on the patterns. This is the approach now favoured by

a great many contemporary language learning courses and books.

27

Lewis (1986) argues that what we discover for ourselves is absorbed more effectively

than what we are taught. A guided-discovery approach is beneficial to students because

they are more actively involved in the learning process and become more autonomous.

Students are also more likely to remember rules they have worked out for themselves

(Skehan, 1998). Harmer (2001) argues that ‘the complex nature of language itself’ is a

strong reason for promoting discovery learning. He believes that encouraging language

learners to be more autonomous should be a priority, arguing that:

students who encounter real language outside the classroom will find it is considerably ‘messier’ than it may appear in a language lesson. Their response to this may well depend

on how prepared they are to observe this messy language and work out, for themselves, how it is put together. Any training in language analysis we have given them will make them more able to do so. (Harmer, 2001)

However, it is a time-consuming process with the risk that students will not hypothesize

the rule correctly. It may also frustrate those students who favour a more teacher-

centred approach or whose past learning experience causes them to prefer simply to be

given the rules. Harmer (2001) points out that discovery learning may be unsuitable for

students whose learning culture conflicts with the approach used. It is also doubtful

whether a more inductive approach works equally well with every type of grammar

item. Moreover, as Thornbury (1999) argues ‘successfully inferring patterns and rules

from the study of language data depends not only on how the data is presented, but on

the quantity and quality of the data itself.’

Two studies show how learners’ preferences about approaches to grammar can vary

considerably.

Fortune (1992) conducted a study comparing different types of self-study grammar

practice exercises, some inductive and others more deductive. He concludes that:

The experience of doing the exercises caused a significant number of learners to change their opinions about inductive and deductive practice. Many informants ended up preferring discovery activities to more traditional types of grammar exercise.

However, a similar study by Ranalli (2001) resulted in contradictory findings. In this

case, the subjects, who were from a Korean background, were given a questionnaire to

determine their preferred approach before then working through two different

28

treatments of the same grammar point (present perfect to describe experiences), one a

deductive treatment from a grammar practice book and the other an inductive treatment

using instances selected from the Bank of English corpus. Afterwards, the subjects

completed a second questionnaire which asked them to rate the approaches according to

their difficulty and usefulness and reassess their own preferences. Findings revealed that

though a majority initially favoured an inductive approach (surprising considering the

deductive approach is deeply ingrained in the Korean education system), many

subsequently said that they would prefer a more deductive approach.

With the advent of large computerised corpora, the possibilities of an inductive

approach have been vastly increased. The next section looks at how corpora can provide

teacher and student alike with a larger and more representative range of examples, such

that pattern-seeking can be done more reliably and more quickly.

29

Chapter 2: Section 2 Developments in Corpora and Concordancing

2.2.1 What is a corpus?

Traditionally the word corpus (plural corpora) was used by linguists to describe

collected examples of language (as it occurs naturally), whether a few sentences or a set

of texts, for the purpose of linguistic study.

In recent years the word has come to refer to ‘collections of texts (or parts of text) that

are stored and accessed electronically…and designed for some linguistic purpose’

(Hunston, 2000). Essentially, a corpus is an electronic ‘store of used language’. Due to

the capacity and processing power of modern computers, electronic corpora are

normally much larger than their paper-based forerunners. A corpus differs from an

electronic archive in that its aim is not to preserve the texts themselves, nor indeed even

to read them as a whole, but instead to study the body of texts in a nonlinear way using

both quantitative and qualitative means, in order to analyse the frequency and actual use

of linguistic items within the corpus. Thus, a corpus provides observational data about

how a language works in reality.

Corpus linguistics is now a well-established branch of linguistics. It first came to

prominence with the announcement of Randolph Quirk’s Survey of English Usage

Corpus in 1959 which led shortly afterwards to the Brown corpus, consisting of one

million words of written American English, and in 1961 the one million word LOB

corpus of written British English. In the 1980s the COBUILD project in the English

department of Birmingham University, led by John Sinclair, resulted in the development

of a 20 million word corpus of contemporary English as a basis for the preparation of a

range of reference and teaching materials for English as a Foreign Language, including

the Collins COBUILD English Dictionary (Sinclair, 1987). Since that time a number of

corpora have been collected and researchers are using them to work towards empirically

based description of languages and varieties of languages.

30

2.2.2 What is a concordancer?

Access to a corpus is usually through a concordancing program or concordancer. This

software enables the user to

recover from text all the contexts for a particular item (morpheme, word or phrase) and to print them out in a way which facilitates rapid scanning and comparison (Johns, 1994).

The most common program is the keyword-in-context (KWIC) concordance whereby,

much like a conventional search engine, the user keys in a given word and the computer

then displays a list of all instances of that keyword in the form of lines of text, known as

concordances, accessed from the corpus. Most commonly the keywords are arranged

one below the other down the centre of the screen with a fixed number of characters on

either side to provide a context. The contexts to the left or right of the keyword can be

sorted alphabetically so as to highlight certain patterns and regularities. Many

concordancers also provide a greater amount of context if required, displaying every

sentence in which the keyword appears in its entirety.

Flowerdew (1996) points out that ‘the ordering which has been found to be most useful

is from the first word to the right of the keyword, because it shows up phrases and

typical collocations which begin with the keyword.’ Hunston (2000) explains how the

concordancer’s ability to reveal patterns of phraseology is a useful tool for teachers

trying to explain differences between minimal pairs and suggests that the learner’s focus

should be drawn to the common phrases highlighted by the concordancer. A wild card

symbol can be keyed-in to stand for any letter or combination of letters (useful in

finding different forms of the same verb, for example).

Among other functions available on most concordancers is the inclusion of a list of all

the words in a corpus and their frequencies, ordered either alphabetically or by

frequency. This is particularly useful when comparing frequency lists for different text

genres and can provide information about differences between spoken and written

discourse and between different registers of language. For example, Biber uses software

which counts not only words but categories of linguistic item and accordingly calculates

the distribution of present and past tenses across four registers (revealing that the

31

present tense occurs more frequently than past tense in conversation and academic

corpora than in fiction and news corpora). (Biber et al, 1999)

Another way in which corpora data can be exploited is in calculating statistical

information about frequently co-occurring words, or collocates. Lists of collocations

and their frequencies can reveal otherwise undetectable information about how lexical

items combine and about the different shades of meaning that words may possess.

2.2.3 Concordancing software

Back in the early 1990s the best available software for language teaching applications

was the Longman Mini Concordancer. This had the advantage of being both user-

friendly and very fast, with almost instantaneous presentation of concordances.

However, it could only handle a maximum of about 45,000 words of text. The Mini

Concordancer has been superseded by the Oxford Mini-Concord which can deal with

much larger amounts of text, due to the fact that corpus size is limited only by the size

of the hard disk, while for the Longman program, it is limited by the amount of working

memory (RAM).

The two main types of concordancing software for use on personal computers can be

divided into streaming concordancers, which “read” a text line-by-line in real time and

produce concordanced text either to screen, printer or disk, and Text-indexing

concordancers which initially create an index of the text in one operation and then

permit a large variety of text retrieval activities.

Streaming concordancers are generally not limited to a particular size of text file – the

only real constraints being the amounts of hard-disk space and Random Access Memory

(RAM) available. Modern, commercially available concordancers such as MonoConc

for Windows and WordSmith run quickly enough to be suitable for both research

purposes and classroom use. Such applications offer a wide range of functions.

Text-indexing software has remained the preferred choice for many academic

researchers because once an index has been created it can be used many times and

access speed with a large indexed corpus is still much faster than with a streaming

32

concordancer. The most commonly used indexing program for language

teaching/learning is WordCruncher.

2.2.4 Types of corpora

The type of corpus depends on the purpose it was created for. A specialised corpus

consists of texts of a given text type, such as newspaper editorials or academic articles,

and is used to investigate a particular register. Such a corpus may be restricted to a

specific topic, time period or social scenario. Well known examples include the 5

million word Cambridge and Nottingham Corpus of Discourse in English (CANCODE)

which specialises in informal registers of British English, and the Michigan Corpus of

Acedemic Spoken English (MICASE).

A general corpus is a corpus of texts of many types, possibly including both written or

spoken language and perhaps form more than one country of origin. Such corpora aim

to include a wide cross-section of texts and thus tend to be very large in size. Examples

include the 100 million word British National Corpus (BNC) and the 400 million word

Bank Of English.

Comparable corpora is the name given to two, or more, corpora in different languages,

designed to contain the same spread of text types and allow learners and translators to

compare different languages or language varieties and identify differences and

equivalences in them. An example is the International Corpus of English (ICE) which

consists of several one million word corpora each of different varieties of English.

Parallel corpora consist of two, or more, corpora in different languages, each

comprising texts translated from one language into the other, and used by translators to

research into how the languages differ and to seek matching equivalent expressions.

A learner corpus is a corpus of texts made by language learners in order to see how the

language of individual learners differs from that of native speakers and other learners.

The most well known is the International Corpus of Learner English (ICLE) – a

collection of sub-corpora, each 20,000 words in size and containing essays written by

learners of English form a specific language background.

33

As well as using published corpora, users may wish to create their own corpora.

Purpose-built corpora are especially appropriate for English for Special Purposes (ESP)

as each corpus can suit the particular target learning group it is designed for.

2.2.5 Applications of corpora

So great is the effect that corpora have had on the production of reference books for

learners of English that nowadays almost all large publishers produce learner

dictionaries and grammars based on a corpus. (Examples include the Macmillan English

Dictionary for Advanced Learners, 2002 and the Longman Grammar of Spoken and

Written English, 1999.) Such books often now include information about the relative

frequencies of words and frequencies of the different uses of the same word. The

diversity of use shown by corpora has led to dictionaries now listing many more senses

of common words. There is also an emphasis now on collocation and phraseology, with

a tendency to define phrases rather than words, and on lexical/pattern grammar and

authenticity, definitions being illustrated with examples taken directly from corpora that

show the most typical usage of a given word.

Corpora can also be used to study the discourse of particular communities and to

examine the variation between regions, gender and social groups as well as linguistic

differences between periods in history. Corpora can throw light on the way society

thinks about issues of gender, race and ideology. Corpora of texts from different ages

can be compared in order to determine how literary styles have changed and developed.

Forensic linguistics use corpora in the comparison legal documents to verify that they

were written by the same person, so as to authenticate police statements, for example.

Translators use comparable corpora to compare the use of apparent translation

equivalents in two languages, and parallel corpora to see how words and phrases have

been translated in the past.

There are many implications for both language learners and teachers arising from

corpus-based analysis of language. The area of syllabus design has been influenced the

most but classroom teaching and individual learner strategies have been affected by the

34

increased availability of concordancing technology. The present study will focus on the

application of concordancing in language learning.

2.2.6 Advantages of using corpora in language learning

A native speaker obviously knows about his/her language but even the most

experienced teacher can have difficulty explaining this knowledge, which is often felt as

an intuitive assumption that a certain word is correct in a given context. Even a

conventional dictionary only provides a few instances of a word in use. Corpora allow a

non-native speaker to get a feel of how language items are actually used. Thus a corpus

can illustrate the reality of language use and learners can test their own hypotheses

against the corpus. A corpus can provide proof that a particular structure is not in fact

used when the learner (or even the teacher) may be convinced otherwise. Biber says that

as linguists, we often have strongly held intuitions, but those intuitions frequently prove to be incorrect when they are tested empirically against the actual patterns of use in large text corpora.’ (Biber et al, 1994)

2.2.7 Disadvantages of using corpora in language learning

However, a corpus will not give information about whether something is possible or

not, only about its relative frequency. Native speaker intuition ultimately has to answer

doubts about acceptability. Even the largest corpus is still only a sample of a language

as a whole and results are only extrapolations. A corpus provides evidence of language

patterns and collocations but does not actually give information about meaning. This

still has to be worked out intuitively, which can prove difficult for learners of English,

especially at low levels.

Moreover, concordance lines only provide the barest of contexts and this absence of the

wider textual context in which the word or phrase was originally placed may make it

difficult for the learner to work out the meaning of a given item. This is especially true

of spoken language when there is no information about paralinguistic features such as

intonation and gesture.

35

2.2.8 Applications of corpora in language learning: for the teacher

One important application of corpora for the teacher is to use the concordancer as a

resource tool to increase his/her own awareness of linguistic usage, and in turn use this

data to make decisions about what linguistic items to teach and in what order, based on

their relative frequency.

Johns stresses the importance of revising teaching materials in light of what corpora

reveal about the nature of language. He says that ‘the evidence thrown up by the data

has left no escape from the conclusion that the description of English underlying our

teaching, whether homemade or inherited from other teachers and linguists, needs major

assessment (Johns, 1991). It is his view that descriptions of grammar in traditional

grammar books are ‘more often based on the armchair intuitions of the grammarian than

on any close analysis of data’.

Biber’s research reveals that there are ‘important systematic differences among registers

at all linguistic levels’ which are not taken into account by most popular EFL grammars

(Biber et al, 1994). He advises teachers of advanced students to ‘focus on the English of

particular varieties in naturally-occurring discourse, rather than “general” patterns that

are culled from linguists’ intuitions and do not accurately reflect the grammar of any

variety.’

One example Biber gives of the tendency for grammar books to over-generalize is in the

area of post-modification in noun phrases. Far more emphasis is placed on relative

clauses than on prepositional phrases as noun modifiers, when in fact corpus-based

research shows that the actual patterns of use are very different, with far more instances

of the latter in post-modification than the former.

Tense usage is another area where corpora can reveal just to what extent reality differs

from expectation. According to Mindt, for example, nearly all the future time reference

in conversational English is indicated by will or other modals. The going to future

accounts for about 10% of future time reference and the present continuous less than

5%. (Mindt, 2000)

36

In deciding what grammatical structures to emphasize and how to sequence them, most

textbooks focus on how difficult given items are to teach. In contrast, Biber states that

an equally important consideration is whether beginning students will ever need to produce or comprehend the construction in question outside the classroom, and if so, how frequently that need will arise.

Biber thus believes that a teacher’s choice of material should depend equally on patterns

of actual use as on questions of teachability.

Flowerdew (1996) suggests that

native speakers who want the added support of an objective linguistic informant to reinforce and corroborate or refute their intuitions and their grammaticality and vocabulary usage judgements, as well as non-natives who lack the confidence to trust their linguistic intuitions, can thus turn to a concordancer to provide them with reliable, objective data on grammatical and lexical usage.

He suggests, for example, that in order to teach the use of sentence connectors to

students for science and technology, the teacher should first consult the vocabulary

frequency list of a scientific corpus to see which connectors are commonly used in

scientific writing and their relative frequency. Next, he/she should examine the

concordances generated from the most frequent connectors to find out what semantic

and syntactic differences there are between them.

As well as using a concordancer as an informant on usage, teachers can use it as a

source of input for teaching and can generate authentic instances of usage to present

students when teaching a particular language point. If the concordancer is kept in the

classroom then the teacher can run a search when a problem of usage arises during the

lesson.

The concordancer can also be used to prepare teaching materials in advance and lists of

concordance lines can be tailored to meet students’ needs or level, by removing those

instances that might be inappropriate for teaching purposes or be beyond the students’

language level.

Moreover, such edited lists of concordances can themselves be turned into exercise

material for students. Honeyfield (1989) has developed seven types of exercises based

on concordance material. These range from gap-filling tasks where keywords are

removed, to discourse-oriented exercises such as the use of discourse markers.

37

Honeyfield also advocates freely using a concordancing program to assist writing,

correction or comprehension and also to explore the emotional tone or style of certain

concordanced items between different genres.

More recently, Tribble and Jones have likewise produced a resource book for teachers

called Concordances In the Classroom (1997) which focuses on practical applications

of concordance data for language learners, including printout-based exercises and

interactive hands-on activities designed to raise students’ awareness of such areas as

position of adverbs, appropriate use of conjuncts in different types of writing and

analysis of different registers. By presenting learners with pre-selected off-line

concordance data the authors’ aim is to limit the information overload that can occur

when learners encounter huge amounts of raw data, and remove the need to operate a

concordancing interface.

Others who have recently advocated using concordancing to teach grammar include

Bernhard Kettemann of the University of Graz, who believes that a concordancer

makes hidden structures visible, enhances at the same time imagination and checks it by inductivity, thus making higher degrees of objectivity possible.’ Ketteman (2000)

He examines some possible applications of concordancing to the teaching of Reported

Speech, the difference between Past Simple and Present Perfect and contrasts between

since and for.

2.2.9 Applications of corpora in language learning: for the learner

Where learners use the concordancer themselves, as opposed to being exposed to pre-

edited printouts, the approach is inherently inductive. Johns writes of the inductive

nature of concordance-based learning as follows:

A concordancer occupies an intermediate position between the highly organised, graded and idealised language of the typical coursebook, and the potentially confusing but far richer and more revealing “full flood” of authentic communication. By concentrating and making it easier to compare the contexts within which a particular item occurs, it organises data in a way that encourages and facilitates inference and generalisation.’(Johns, 1991)

38

Johns in particular has developed an inductive approach to concordancing which he

calls ‘Data Driven Learning’ (Johns, 1991). DDL was developed in the English

Department of Birmingham University, home also to the COBUILD project (Sinclair,

1987). According to Johns, both share an approach which aims to ‘contextualise’ and

‘demythologise’ the language but differ in the use they make of the evidence.

The method used by COBUILD in working with its corpus is summarised by Sinclair as

follows:

The computer sorts the words in various ways and delivers information on each word to a team of editors and compilers. They study the words and build up an elaborate profile of their meanings and uses in a database back in the computer. The database is then the primary source of a family of books which will span many years of editorial work. (Sinclair, 1987)

Johns seeks to ‘cut out the middleman as far as possible’ – that is, the mediating role of

the editing team –

and to give direct access to the data so that the learner can take part in building up his/her own profiles of meaning and uses. (Johns, 1991)

Johns sees effective language learning as a form of ‘linguistic research’ and that the

learner’s needs are ‘driven by access to linguistic data’. In his view, the concordance

printout

offers a unique way of stimulating inductive learning strategies – in particular the strategies of perceiving similarities and differences and of hypothesis forming and testing.

Learners are like ‘language detectives’ discovering facts about the language from

authentic examples.

A DDL concordancer activity might begin spontaneously in class when a question of

usage or function arises such as ‘What is the difference between therefore and hence?’

or ‘Why aren’t all shoulds real shoulds?’ (Johns, 1994) Johns explains how, in the latter

case, the teacher would respond that he/she does not know and that they can find out

together. Students might then consult a corpus and ask the concordancer to return all

instances of should. The computer’s output would include sentences with should that

the learners would then attempt to classify according to their various meanings and

contexts. This kind of spontaneous study uses a raw corpus - that is, the learner and

teacher will examine the corpus together without either necessarily knowing what they

39

will find. Indeed, a student may well notice something that a teacher has overlooked or

that no textbook covers. Johns ascertains that it is ‘this element of challenge and of

discovery that gives DDL its special flavour and stimulus.’ (Johns, 1991)

The teacher’s role becomes that of ‘a director and coordinator of student-initiated

research’. (Johns, 1991) Johns himself acknowledges that it can be difficult for teachers

to adapt to this change in role. Hunston points out that the teacher has very little control

over what happens and may feel ‘a loss of expertise’ if no answers are forthcoming and

students pose further difficult queries. (Hunston, 2000) She details other drawbacks to

DDL such as the fact that one-to-one interaction between student and teacher is rarely

achievable and computer access is often insufficient to allow students to do their own

linguistic research. Moreover, in cases where the teacher prepares pre-selected printed

handouts, by choosing the language area for study, he/she may remove the high

motivation felt by students when they want to satisfy their own personal linguistic

doubts.

Chapelle (2001) points out that if a DDL activity results in considerable amounts of data

which cannot be resolved by learners nor teacher, the effect could be negative because

learners may see it as a waste of time. She stresses the need for learner training in both

the mechanics of concordancing and the sort of questions that can be asked of the

concordancer.

Flowerdew (1996) also describes a number of problems connected with the use of

concordance data. First, it is likely that many concordance lines will contain language

beyond the learners’ level of understanding even when the target language item may be

at their level. Second, if single-line concordances are used, not all of them may provide

enough context to make the meaning clear. Third, depending on the size of the corpus

and the frequency of the given item, the concordancer may provide too few examples of

a particular usage or, conversely, too many, either of which may cause learners to

become easily frustrated. Finally, if a certain item has a variety of usages, one or other

of these may be very thinly represented in the corpus and difficult to pick out amidst

more common usages.

40

Cobb (1997) claims that advocates of data-driven learning have over-estimated the

needs of learners to get their hands on raw linguistic data and indeed the amount of such

data they can cope with. Both Cobb and Goodfellow argue that, in the field of

vocabulary instruction, learners should not be left to deal with concordancing data

unaided, but require ‘tutorial support for the use of lexical resources.’ (Goodfellow,

1995)

Johns (1994) has since attempted to increase his students' interest in hands-on

concordancing by building a tutor, CONTEXTS, on top of his concordance program.

The online tutor poses questions for learners to answer by exploring the concordances

supplied onscreen. This data comes in the form of pre-selected citations, which have

been chosen to make specific points about grammar or lexis, points which learners

might never notice for themselves.

Cobb (1997) criticises such use of pre-selected concordance data:

Theoretically, any learning advantage that might accrue to a genuine process of "discovery" or "construction," in other words to making sense of raw data, is compromised if the data has already been made sense of by somebody else.

Cobb also points out that the process of pre-selecting concordance lines to illustrate

specific points is very time consuming and

effectively means that learners will be exposed to a limited number of concordancing experiences.

Cobb advocates bringing learner and corpus together in a way that neither compromises

the essential idea of concordancing – free exploration of language patterns - nor pre-

limits the amount of program use.

2.2.10 Lack of Research

It is a fact that student concordancing has generated a lot of enthusiasm but little

empirical research. Hunston (2000) makes the point that Tim Johns’ claims that Data

Driven Learning actually improves students’ general skills in deducing meaning from

context ‘have yet to be adequately tested’.

41

Of the several studies of student concordancing gathered into Johns and King (1991)

most simply described students in various concordance activities in guided lab sessions.

Only one presented any quantitative data about the learning effectiveness of

concordances, and that was an off-line study. Vance Stevens (1991) at Sultan Qaboos

University in Oman predicted that learners would be able to retrieve a word from

memory more successfully when cued by several pre-selected concordance lines with

the key word masked than by a single gapped sentence. His prediction was confirmed,

thus providing some proof that concordancing can facilitate language learning.

Tom Cobb, while at the City University of Hong Kong, conducted a study to see if there

is ‘any measurable learning from hands-on concordancing’. (Cobb, 1997) He proposes

that ‘a computer concordance might stimulate and potentially rationalize off-line

vocabulary acquisition by presenting new words in several contexts’. To test this idea,

an experimental lexical tutor was developed to introduce new words to subjects, either

through concordances or through other more conventional sources of lexical

information. In a series of tests involving transfer of word knowledge to novel contexts,

a small but consistent gain was found for words introduce through concordances.

Accounts of similar studies to measure the effectiveness of a corpus-based approach to

grammar learning are lacking in the literature.

In attempting to explain this lack of research, Cobb says that ‘it is doubtful that students

have ever used concordancing enough to generate a very large database so that even

initial pattern-perception can begin.’ (Cobb, 1997). Moreover, he goes on to say that ‘in

an open-exploration environment there is no learning task built in and no way of

knowing for certain what learners are trying to learn or whether they are succeeding.’ In

Cobb’s view ‘the way concordancing is typically introduced to students does not allow

variables to be isolated in any straightforward research design.’

This lack of hard research has led to an under-implementation of a potentially powerful idea. Cobb believes that ‘without the benefit of an instructional design process guided by research, the concordancing idea is now widely seen as running into trouble.’ (Cobb, 1997)

42

2.2.11 Online Applications of Corpora

Aside from the classroom-based uses of corpora mentioned above, there are now

available many online applications, providing the learner direct access to concordancing

tools.

Tom Cobb, now at the Université du Québec à Montréal, Canada, has developed the Compleat Lexical Tutor providing a full range of tools to enable data-driven learning on the web. These include online tutorials, text tools, and resource building materials for teachers.

Included also is a powerful online concordancer (created by Chris Greaves) which gives the user access to fourteen separate corpora (among them the Brown corpus and the BNC spoken and written corpora) with the option of searching within the 4 million words of all these corpora combined. The interface of the Lextutor concordancer is user-friendly and displays instances using the KWIC (Keyword in context) mode. The user can further click on any of the keywords to see the source of the instance and read the surrounding context. The interface includes controls for refining the search by selecting associated words, and their range from the keyword, particularly useful for exploring collocations and lexical phrases.

There are also concordance tests which provide samples of sentences

focusing on prepositions, singular-plural, gerund-infinitive, and simple past/present

perfect. Learners read simple sentences, click "OK" or "Not", and check to see if they

are correct by using the concordancer.

Another impressive online resource is the Hong-Kong-based Edict Virtual Language

Centre (VLC), which bills itself as ‘a resource-assisted eLearning website for studying

English’. It is designed to be used for both independent self-access learning and as a

resource for teachers. Of particular interest here is VLC’s inclusion of a web

concordancer which lacks some of the more sophisticated design features of the

Lextutor but provides effective KWIC concordancing, with a wide choice of corpora

43

which include the Bible and the Koran, several novels and issues of The Times as well

as the Brown and LOB corpora.

There is also a tutorial on studying grammar with the aid of the concordancer. It

includes exercises dealing with preposition collocations which get users to think first

and use the concordancer to check their answers.

The University of Michigan Corpus of Academic Spoken English (MICASE) is also available online, providing 152 transcripts (totalling

1,848,364 words). Its concordancer allows the user to refine a search by choosing from a selection of speech event attributes (such as the type of speech act or academic area) and speaker attributes (such as gender, age and academic position). MICASE is designed to study patterns of spoken English and focus on lexical phrases in speech.

In addition, Collins Cobuild and the BNC both have ‘sampler’ sites where users can use a simple concordancer to access a sample of their corpora. These are useful tools for teachers and learners but do not contain any specific tutorials or accompanying exercises.

44

Chapter 2: Section 3 Developments in Learning Style theory

2.3.1 Overview of factors involved in second language learning

Stern (1983) provides a useful framework for examination of second language learning.

↗ ↓

→ ↑↑

↑

↘

Fig 2 Diagram identifying the key factors or variables which are believed to be essential to second language learning and suggesting probable ways in which the sets of different variables may interact.

45

Social context

Sociolinguistic, sociocultural, and socioeconomic factors

Learner characteristics

AgeCognitive characteristicsAffective characteristicsPersonality characteristics

Learning process

Strategies, techniques, and mental operations

Learning conditions

Educational treatment:ObjectivesContentProceduresMaterialsEvaluation

Learning outcomes

L2 competence/proficiency

There are five sets of variables, three of which – social context, learner characteristics

and learning conditions – are seen as determining the learning process and through it

the learning outcome.

According to Stern (1983),

the learner process can be looked upon as consisting overtly of strategies and techniques

employed by the learner and, covertly, of conscious and unconscious mental operations.

The present study focuses on the set of variables here labelled learner characteristics,

and in particular cognitive characteristics. Attention is directed especially towards the

area of inductive learning.

2.3.2 Language Aptitude

Aptitude is usually regarded as a cognitive variable. Carroll defines it as follows:

Aptitude as a concept corresponds to the notion that in approaching a particular learning task or program, the individual may be thought of as possessing some current state of capability of learning that task – if the individual is motivated, and has the opportunity of doing so. That capability is presumed to depend on some combination of more or less enduring characteristics of the individual. (Carroll, 1981)

Larsen-Freeman & Long (1991) explain how the idea of an aptitude for language arose

from the common-sense notion of some students being better at learning languages than

others, and being specially talented in the same way that there are gifted learners who

demonstrate a high degree of musical aptitude, mathematical ability, manual dexterity

and so on.

Skehan (1989) describes how Carroll and Sapon (1959) devised a series of tests based

on several criteria related to language learning including ability to recognize phonemes,

and ability to develop meanings inductively. These tests were given to large numbers of

language learners and after analysis of the data Carroll put forward what has become the

standard ‘four component’ view of language aptitude.

1. Phonetic Coding Ability2. Grammatical Sensitivity3. Inductive Language Learning Ability4. Rote-Learning Ability for foreign language materials

46

Skehan (1989) explains that ‘grammatical sensitivity’ refers to the ability to recognize

the grammatical functions that words fulfill in sentences. Carroll (1973), quoted in

Skehan (1989), defines ‘inductive language learning ability’ as

The ability to examine language material and from this to notice and identify patterns of correspondence and relationships involving either meaning or syntactic form.

In other words, the ability to infer from limited evidence.

Larsen-Freeman & Long (1991) explain how language aptitude tests like the MLAT

have been useful primarily as practical instruments with which to conduct placement

testing at the beginning of language courses and to identify learners’ individual

strengths and weaknesses. Such tests are considered by some to be useful in providing a

better theoretical understanding of the nature of language aptitude. Stern (1983), for

example, says that language aptitude is

a group of characteristics which are relatively stable and should be considered as a given – as a learner factor to count with.

He contends that language aptitude is not simply a ‘gift’ that someone either has or does

not have but that it is

a composite of different characteristics which come into play in second language learning

However, there has been considerable criticism of the whole notion of language aptitude

and the usefulness of aptitude tests. Harmer (2001) points out that they are more suited

to people who have little difficulty doing grammar-focused tasks and also do not test the

non-analytical aspects of language learning nor its communicative and social features.

The MLAT tests were also very much the product of their times and mirrored the

audiolingual principles of the 1950s and 60s which were rooted in theories of

behaviourism. More recently language learning has become increasingly seen as a

universal human cognitive characteristic and less a capacity that individuals possess in

greater or lesser doses. Neufeld (1978) (quoted in Larsen-Freeman & Long, 1991)

suggests that all humans are equipped to master basic language skills but that an

individual’s intelligence determines the extent to which he can master the higher level

skills. For Neufeld there is no specific innate faculty called language aptitude.

However, Carroll (1981) counters this by saying that

47

Intelligence and aptitude may overlap but are not identical.

According to Skehan (1989), language aptitude is

consistently the most successful predictor of language learning success and at least as important, and usually more important, than any other variable investigated.

Skehan (1998) believes that language aptitude should focus much more on the aspect of

memory, as what distinguishes exceptional students from the rest is that they have

unusual memories, particularly for the retention of things that they hear. Unlike the

MLAT, more recent research by Skehan has focused not solely on associate memory,

but on a wider consideration of memory, including size of working memory.

2.3.3 Cognitive Style and Learning Style

Alongside interest in discovering language aptitude some attempts have also been made

to identify global cognitive and learning characteristics.

Cognitive style refers to the preferred way in which individuals process information or

approach a task. Witkin et al (1981) (quoted in Larsen-Freeman & Long, 1991) define it

as

a characteristic self-consistent mode of functioning which individuals show in their perceptual and intellectual activities.

The term learning style is used to refer to how such cognitive styles may apply to

learning. Skehan (1998) defines learning style as

the characteristic manner in which an individual chooses to approach a learning task.

Skehan also points out the importance of distinguishing learning style form language

aptitude since a particular style adopted may reflect personal preference and not innate

ability. Individuals may display a given learning style to varying degrees, so there is not

the same fixedness which applies to language aptitude.

The main application of learning style research to language learning has been through

the concept of field dependence/independence (FD/I), developed in mainstream

psychology by Witkin (1962) (detailed in Larsen-Freeman & Long, 1991), who

48

proposed that a contrast can be made between analytic and holistic individuals. The

most widely-known means of measuring FD/I is via the Group Embedded Figures Test

(GEFT) whereby subjects have to pick out simple visual figures embedded in a complex

pattern. When faced with a decision making task, analytic, or field independent learners,

separate the problem into constituent parts and focus on these components,

manipulating them individually, while holistic, or field dependent learners perceive

situations as a whole.

According to Witkin, in language learning, an analytic learning style is shown by the

individual’s ability to analyse language material he/she is exposed to, identify

components and explore the relationship between them. Such a learner extracts what is

important from the flow of incoming speech or text, focusing on what is meaningful and

separating out irrelevance. This ability to channel selectively the essential from the non-

essential has its downside in that such learners tend to be more aloof and withdrawn,

shunning opportunities to communicate and work with others.

A holistic learning style, on the other hand, is shown by the individual’s ability to

interpret situations as wholes. Such learners are not such effective information

processors and depend on external reference points such as other people’s opinions.

However, they are excel in communication situations and seek out opportunities to

receive good quality, relevant input and are more sociable and people-oriented.

Witkin is not claiming that some people have greater ability than others, but that there

are differences in the way people interact with the world and perceive and organize

information. Neither type of learning style is better than the other but each is

advantageous for different tasks.

However, the main drawback to the concept of FD/I is that it is a polarized view. It is

easy to categorise people at extreme ends of the continuum but those in the middle are

harder to classify. Brown (1994) makes the point that people are flexible and may adapt

to different situations in different ways. People are thus not fixed at one end of the

continuum for all tasks and learning style may vary according to the situation.

49

In language learning FD/I has not been shown to be a significant influence on language

proficiency. Skehan (1989) suggests that investigators turn their attention to other issues

such as memory retention. He concludes that the outcome of FD/I tests depend a lot on

general intelligence factors. Griffiths and Sheen (1992) (quoted in Skehan, 1998) are

even more critical and claim that FD/I is a case of a construct from one discipline being

misapplied to another. The GEFT test was originally proposed as a visual/perceptual

construct and is not, in their view, applicable to language learning in any convincing

way.

However, Chappelle and Green (1992) defend FD/I as a learning style model and

discuss how the construct should be divided into three distinct aspects:

Reliance on internal or external frames of reference Cognitive restructuring ability Interpersonal competence

They argue that people are more or less likely to want to make judgments based on their

own interpretations and some will want to check against others’ opinions and may be

less confident. Chapelle and Green propose that their ‘cognitive restructuring ability’

correlates with general intelligence and other cognitive abilities including language

aptitude. They propose that the three component structure for aptitude should be

extended with a language analytic component divided into a ‘crystallised’ language

analytic ability (the capacity to draw upon relevant knowledge of language from the

learner’s L1) and a ‘fluid’ general analytic ability (capacity to solve language-learning

problems without recourse to existing language knowledge).

More recent research has tended to use two dimensions in measuring style preferences

and characteristic modes of behaviour.

Riding and Cheema (1991) state two basic problems with Witkin (1962). Firstly, that it

does not represent each of the poles of the dimension positively and that it implies that

‘holistic style’ is simply the absence of ‘analytic’. Under Witkin, ‘holists’ do not get

the chance to show their ability to apprehend integrated, unanalysed perceptions as

unified fields. Riding and Cheema add an extra dimension, namely a contrast between

visual/verbal representations, arguing that the original GEFT test was too visually

50

oriented. Riding and Cheema propose that an analytic/holistic approach should focus on

style of processing. Visual/verbal contrast thus targets subjects’ preferred style of

representation - that is it measures how some people prefer to think verbally and others

visually. When applied to L2 analysis of language, analytic learners prefer to search for

components of pattern in language and analyse a chunk of language into its component

parts. This may lead them to formulate rules, either implicitly or explicitly. Their mental

representation of the input they have processed will be either visual or verbal. Holistic

learners display a less flexible approach to patterns in language and prefer to deal with

language as ‘chunks’ (Riding & Cheema, 1991). They resist breaking down such

chunks into subsections, but prefer to retain the larger unit of language, and seek ways

of using it as it is. They, too, would then represent material verbally or visually.

Riding and Cheema (1981) have developed computer-based procedures to measure the

two dimensions of style; analytic/holistic and verbal/visual. In a similar way to the

GEFT, their test uses embedded figures to assess analytic style but also uses geometric

figures to assess holistic style, whereby subjects have to perceive a whole figure,

retaining its unity, rather than just locate a simple figure within a more complex figure.

They also use verbal judgement tests to assess verbal/visual preferences. Being

computer-based, it can record how fast subjects respond and as such, even though all the

subjects can answer the questions, the test can identify which are ‘easier’ and whether

an individual has more or less capacity to make verbal or visual judgements.

2.3.4 – 2.3.11 Learning Style Models

2.3.4 Gregorc

Gregorc (1979) also describes learning style using two dimensions, namely

abstract/concrete and serial/random. The latter is similar to Riding’s analytic/holist axis,

both focusing on how learners process information. According to Gregorc, ‘serialists’

favour methodical analysis and work through material sequentially, while ‘random’

learners take a more global view and expect to see patterns from larger amounts of

initially unstructured data. Gregorc’s second dimension, again like Riding, is concerned

51

with the information being represented. ‘Abstract’ learners deal well with

decontextualised material where as ‘concrete’ learners mistrust abstractions and cope

better with material in context that relates to their own personal experience.

Gregorc's Mind Styles Model identifies four major learning types (Gregorc, 1979):

Concrete Serial

These learners like order, logical sequence, following directions, predictability,

and getting facts. They learn best when they have a structured environment.

They can rely on others and can apply ideas in pragmatic ways. They find it hard

to deal with abstract ideas or questions with no right or wrong answer.

Abstract Random

These learners like listening to others, bringing harmony to group situations, and

establishing healthy relationships with others. They learn best when they are in a

personalized environment and given general guidelines. They find it hard to

work with dictatorial/authoritarian personalities, or in a restrictive environment

and dislike having to concentrate on one thing at a time.

Abstract Serial

These learners like to analyze a situation before making a decision or acting, and

applying logic. They learn best when they have access to experts or references,

they are placed in stimulating environments, and they are able to work alone and

have time to deal with a subject thoroughly. They find it hard having to repeat

the same tasks over and over and dislike lots of specific rules and regulations.

Concrete Random   

These learners like experimenting to find answers, taking risks, using their

intuition, and solving problems independently. They learn best when they are

able to use trial-and-error approaches, and are given the opportunity to work

52

through problems by themselves. They dislike having to show how they reached

an answer, and being restricted to choosing only one answer.

2.3.5 Kolb

Another view of learning style is taken by Kolb (1976), who characterises it in dynamic

and static terms. Kolb’s model does not focus on how information is processed but more

on how the individual interacts with the world, whether actively or passively. He

proposes that learning is a cyclical process involving four stages:

concrete experience reflection-observation abstract conceptulisation active experimentation

Effective learning would take in each of these stages with the cycle repeating itself as

the individual enters new and more complex levels of learning. Kolb also suggests that

some learners may be fixed at certain stages of the cycle. For example, those relying too

much on a concrete approach will be unable to relate to anything that does not arise

from their own experience.

Kolb’s Learning Styles inventory (Kolb, 1984) includes:

Diverging

People with ‘diverging’ styles are able to look at things from different

perspectives. They are sensitive, preferring to watch rather than do, and tend to

gather information and use imagination to solve problems. They are best at

viewing concrete situations from several different viewpoints. They tend to be

imaginative and emotional with broad cultural interests and like to gather

information. They are interested in people and prefer to work in groups, listen

with an open mind and receive personal feedback.

Assimilating

53

Those with an ‘assimilating’ learning style prefer a concise, logical approach

with more importance on ideas and concepts than people. Such learners require

good clear explanation rather than practical opportunity. They excel at

understanding wide-ranging information and organizing it in a clear logical

format. People with this style are more attracted to theories than practice and in

formal learning situations prefer readings, lectures, exploring analytical models,

and having time to think.

Converging

People with a ‘converging’ learning style use their learning to find solutions to

practical issues. They prefer technical tasks, and are less concerned with people.

They enjoy solving problems and making decisions as well as experimenting

with new ideas and working with practical applications.

Accommodating

The ‘accommodating’ learning style is 'hands-on', and relies on intuition rather

than logic. Such learners use other people's analyses, and prefer to take a

practical, experiential approach. They are attracted to new challenges and

experiences, and to carrying out plans. They commonly act on 'gut' instinct

rather than logical analysis while tending to rely on others for information.

People with this learning style prefer to work in teams to complete tasks. They

set targets and actively work in the field trying different ways to achieve an

objective.

2.3.6 Willing

Willing (1987) attempts to relate concepts of learning style from general psychology to

the field of language learning. Willing proposes his own two-dimensional framework of

learning style based on Kolb’s model. This consists of an analytic/holistic axis

(corresponding to field dependence/independence) and a passive/active axis which

54

focuses on the extent to which an individual is proactive or authority-oriented. Like

Kolb, Willing thus considers that an individual’s personality disposition is key to how

he/she processes information. On this basis Willing characterises four types of language

learner:

Convergers

These are ‘analytic’ learners who prefer to study alone and who are independent

and confident in their own abilities, able to impose their own structures on

learning. They like studying grammar, studying English books and reading

newspapers, finding their own mistakes and working on problems set by the

teacher.

Conformists

These are ‘authority-oriented’ learners who like the teacher to explain

everything and who prefer emphasis on learning ‘about language’ over learning

to use it. They like to have their own textbook, to write everything in a

notebook, to learn by reading and learn new words by seeing them.

Concrete learners

They are more interested in using language for communication rather than

studying it as a system. They like the social aspects of learning and like to learn

from direct experience, enjoying games, pictures, films, video, pair and group

work activities.

Communicative learners

They are confident at using the language, comfortable speaking out of class and

are more willing to take risks. They are much more interested in social

interaction with other speakers of the language than they are with analysing how

the language works. They are prefectly happy to work without the guidance of a

55

teacher. They like watching TV in English and learning new words by hearing

them.

In a major study of learning styles among adult learners of English as a second langauge

in Australia, Willing (1987) obtained data on the learning preferences of 517 learners.

The main aim of the research was to determine whether or not learning style differences

can be attributed to different learner biographical variables, such as ethnic background,

age, level of previous education, and speaking proficiency level. Data was collected by

means of a questionnaire which learners completed during the course of an interview.

Surprisingly, the research showed that there was no significant correlation between the

biographical variables and any of the learner preferences.

None of the learning differences as related to personal variables were of a magnitude to permit a blanket generalization about the learning preference of a particular biographical sub-group. Thus, any statement to the effect that ‘Chinese are X’ or ‘South Americans prefer Y’, or ‘Younger learners like Z’ or ‘High-school graduates prefer Q’, is certain to be inaccurate. The most important single finding of the study was that for any given learning issue, the typical spectrum of opinions on that issue were represented, in virtually the same ratios, within any biographical sub-group. (Willing, 1989)

Willing also reports that 10% of the subjects tested were convergers, 30% conformists,

10% concrete and 40% communicative learners. Skehan (1998) points out that the

results could be very different if the study were conducted in other circumstances with

people from different backgrounds or in a country where English was not the native

language.

2.3.7 Skehan

Skehan (1998) proposes a system of characterization of individual learning styles using

two related dimensions, namely ‘degree of analysis’ and ‘amount of memory’. He

suggests that high analysis language learners would develop differentiated, organized

and rule-based representations of language and would value accuracy. Low analysis

learners, on the other hand, would not be so able to organize the language input they are

exposed to and would be less able to articulate rules. High memory learners would be

able to draw on a wide range of lexical elements, internally represented in a variety of

56

ways, and recall them easily in communicative situations. Low memory learners would

not have recourse to such a variety of lexical exemplars or mental representations.

2.3.8 Myers-Briggs

The models of learning style highlighted thus far have dealt mainly with the way

learners sense, think, solve problems, and remember information. However, another

category of learning style models are those that focus on personality patterns.

The Myers-Briggs Type Indicator measures preferences on four scales derived from

Jung's Theory of Psychological Types (Myers & McCaulley, 1985).  People are

classified according to their preference for:

Introversion (I) - interest is drawn mainly to the inner world of concepts and

ideas.

Extroversion (E) - interest is drawn mainly to the outer world of actions,

objects, and persons.

Sensing (S) - tendency to perceive immediate, real, practical facts of experience

and life

Intuition (I) - tendency to perceive possibilities, relationships, and meanings of

experiences.

Thinking (T) - tendency to make judgments or decisions objectively and

impersonally.

Feeling (F) - tendency to make judgments subjectively and personally.

Judging (J) - tendency to act in a planned and decisive way.

Perceiving (P) - tendency to act in a spontaneous and flexible way.

On this basis, it is claimed by Felder that an individual learning type can be made out of

sixteen possible combinations of these preferences (Felder, Felder, and Dietz, 2002).

For example, an ENTP would have a preference for extroversion, intuition, thinking,

and perception.  A preference for one or the other category of a dimension may be mild

or strong. Students with different type preferences tend to respond differently to

different teaching styles. ‘Extroverts’ like activity and group work while ‘introverts’

57

prefer working alone.  ‘Sensors’ like concrete learning experiences and clearly defined

expectations and dislike instruction heavy in abstractions such as theories and

mathematical models. On the other hand, ‘intuitors’ like instruction based on

understanding concepts rather than on memorization of facts, rote substitution, and

repetitive calculations. ‘Thinkers’ like logically organized presentations of course

material and feedback related to their work. ‘Feelers’ like those teachers who establish a

personal rapport with them and show appreciation of their efforts. Judgers like well-

structured instruction with clearly defined assignments, goals, and milestones.

‘Perceivers’ like to have choice and flexibility in their assignments and dislike rigid

timelines (Felder et al, 2002)

2.3.9 Felder-Silverman

Based on the Myers-Briggs Type Indicator, at North Carolina State University Felder

and Silverman developed their own Learning Style Model (Felder & Silverman, 1988),

with engineering students in mind. It classifies students using five dimensions:

Sensing / Intuitive

Sensing learners are concrete, practical and oriented toward facts and procedures

while intuitive learners are conceptual, innovative and oriented toward theories

and meanings.

Visual / Verbal

Visual learners prefer visual representations of presented material e.g. pictures,

diagrams, flow charts while verbal learners prefer written and spoken

explanations.

Inductive / Deductive

Inductive learners prefer presentations that proceed from the specific to the

general while deductive learners prefer presentations that go from the general to

the specific.

58

Active / Reflective

Active learners learn by trying things out and prefer to work with others while

reflective learners learn by thinking things through, and prefer to work alone.

Sequential / Global

Sequential learners are linear, orderly, and learn in small incremental steps

while global learners are holistic, systems thinkers, and learn in large leaps.

2.3.10 McCarthy

Among other models that could be considered as personality-based is that built by B.

McCarthy (1990), who identified four learning styles:. 

Innovative learners

They look for personal meaning while learning, draw on their values while

learning, enjoy social interaction, cooperate and want to make the world a better

place.

Analytic learners

They want to develop intellectually while learning, draw on facts while learning,

they are patient and reflective, they want to know " important things" and to add

to the world's knowledge.

Common sense learners

They want to find solutions, they value things if they are useful, they are

kinesthetic, they are practical and straightforward, they want to make things

happen.

Dynamic learners

They look for hidden possibilities, judge things by gut reactions, synthesize

information from different sources, and are enthusiastic and adventurous.

59

2.3.11 Gardner

H. Gardner's (1985) concept of multiple intelligences' is commonly viewed as, in fact, a

model of learning styles. According to this point of view, the following types of

learning styles can be identified (Gardner, 1985):   

Visual Learners

These learners need to see the teacher's body language and facial expression to

fully understand the content of a lesson. They tend to prefer sitting at the front of

the classroom to avoid visual obstructions. They may think in pictures and learn

best from visual displays. They often prefer to take detailed notes to absorb the

information.

Auditory learners

They learn best through verbal lectures, discussions, talking things through and

listening to what others have to say. Auditory learners interpret the underlying

meanings of speech through listening to tone of voice, pitch, speed and other

nuances. Written information may have little meaning. These learners often

benefit from reading text aloud and using a tape recorder.

Tactile/Kinesthetic learners

They learn best through a hands-on approach, actively exploring the physical

world around them. They may find it hard to sit still for long periods and may

become distracted.

2.3.12 Practical Application of Learning Style Theory

60

With such a large number of different learning style models, the task of teachers to

apply the theory to their own students in specific learning environments is clearly a very

complex one.

Harmer (2001) cautions against ‘pigeonholing students with fixed descriptions so that

we assume they are always going to behave in the same way.

Felder too, in defence of his own Index Of Learning Styles (an online instrument

designed to assess preferences on the dimensions of the Felder-Silverman model),

recognises the need to guard against possible misuse of learning style models. He

reminds us that learning style dimensions should be viewed as continua, and not as

either/or categories.

A student’s preference for one or other pole of a given dimension may be mild, moderate or strong.

Felder further argues that learning style profiles suggest behavioural tendencies but are

not infallible predictors of behaviour just as they do not provide a reliable measure of

students’ actual learning strengths and weaknesses and can also be affected by previous

educational experiences.

Williams and Burden (1997), writing from a social constructivist angle, are particularly

averse to current developments in the whole area of individual differences.

In dealing with averages and statistics we appear somehow to have lost track of the individual. This kind of approach does not help us to deal effectively with such issues as how individuals make their own sense of the process of learning a language, or how we as teachers can best help our learners, given that they are all different. (Williams and Burden, 1997)

They call for a different approach which will focus on

the unique contribution that each individual brings to the learning situation, and on how the teacher can assist the learner in learning more effectively.

This is a view echoed by Harmer (2001) who says we should ‘recognise students as

individuals as well as members of a group’. He states the importance of providing

activities ‘which offer maximal advantage to the different people in the class.’ This

means in practice that some activities will be more useful for some students than others

but that ‘most of the class will be engaged with the learning process most of the time’.

61

Felder and Spurlin (2005) also argue that

The point of identifying learning styles is not to label individual students and modify instruction to fit their labels. While studies have shown that greater learning may occur when teaching styles match learning styles than when they are mismatched, a strong case can be made against teaching exclusively to accommodate learning style preferences.

They advise teachers to try to provide a balanced approach to teaching such that all students are sometimes taught in a manner that matches their learning style preferences, and sometimes in the opposite manner

so they are forced to stretch and grow in directions they might be inclined to avoid if given the option.

2.3.13 Learning Style Theory Applied to CALL

To what extent do learners’ preferred styles of learning influence their use of CALL as a

means of studying grammar? Chapelle (2001) states that this is a ‘thorny question’

which ‘remains an important research issue for the future’.

She does refer, however, to a study carried out by Abraham (1985) on cognitive style

and task variables in materials teaching participial phrases. Findings showed that field-

independent (i.e. analytic and independent) language learners performed better on post-

tests when they had used a rule-presentation (deductive) approach and field-dependent

(i.e. holistic and dependent on others) learners performed better after using software

presenting examples of the structure (inductive).

In another study, Chapelle and Jamieson (1986) found field-independent language

students tended to have a more negative attitude to CALL-based instruction, while field-

dependent students had more positive attitudes.

The literature reveals a lack of research into the area of learning style as applied to

concordancing and corpus-based approaches to language learning.

62

Chapter 3: The Experiment

This chapter outlines the procedure of the experiment central to this thesis.It first sets out the background and aims of the study, before detailing the

target language, the subjects as well as the methods and materials used.It then documents the inductive and deductive modes of the online lesson,

providing screenshots of relevant webpages.

63

3.1 Background

A number of computer-based initiatives for grammar teaching preceded the present

project, a brief description of which will provide a picture of the context in which the

experiment was run.

Since 2000 FLUP has liaised with the Reitoria of the University of Porto in the setting

up and maintenance of a website designated for English 3 students, specifically for

presentation and practice of the material included in their grammar third year grammar

syllabus. This self-contained grammar course is accessed via the WEB-CT learning

platform. The course has been modified and developed over the years.

The material is divided into a number of modules subdivided into units, each dealing

with a different area of the grammar syllabus. Module 1 presents the Noun Phrase and

Module 2 the Complex Sentence, with units on nominal clauses and adverbial clauses.

The grammar is presented in a variety of ways and students can to some extent choose

whether to follow a purely deductive approach or a more inductive ‘guided discovery’

type approach. For example, they can first read up on the grammar in their

accompanying Study Guide (also accessed online) and then do practice exercises on the

computer, or they can find out about the grammar and be guided to an understanding of

the main principles by following a series of online ‘lessons’ which encourage them to

think and work out the rules for themselves.

This online course did not use concordancing as a means of presenting grammar.

In 2002 the teachers of the English 3 course became involved in project ITEGO, a joint

venture with the University of Chemnitz, Germany to collaborate and share each other’s

know-how regarding online grammar teaching. Chemnitz had developed their own

Internet Grammar (Schmied, 2002) – a site designed to help students find out more

about a range of grammar areas. What is distinctive about this online grammar site is the

64

way that it offers the user a dual approach to learning as the following from the

homepage explains:

This web-based self-learning environment is a tool to explore English Grammar in a new way. You can learn at your own pace and work through various grammar topics as often as you like. The grammar is intended for learners at an intermediate or advanced level of English. Throughout the Grammar we offer you two approaches and accompanying exercises: Discovery Approach Challenges you to discover the rules on your own, with the help of various examples and hints. Explanation Approach Will help you to learn about English Grammar by presenting rules and relating examples in detail.

The ‘examples’ referred to in the Discovery Approach are in fact concordances

carefully selected for the most part from the BNC (British National Corpus).

Materials from FLUP’s Online Grammar were included in the Explanation section of

the Chemnitz Internet Grammar and FLUP students were encouraged to access the

Chemnitz site for further opportunities to practise.

However, it was felt that the Discovery section was only a partially inductive

experience, since the examples to illustrate the grammar were selected from the corpus

and messy or potentially confusing concordances were deliberately avoided. Attention

turned to the possibility of building a concordancer into the website so that students

could actually consult the original concordance lists and access raw language content

with all the irregularity intact.

3.2 Aims

The present project is thus a small scale attempt at providing a dual approach to learning

about a particular grammar area.

65

The main aim of the project is to try to discover whether or not an individual’s learning

style, specifically his/her ability to process information inductively, will influence how

successfully he/she deals with using a concordancer as a means of studying grammar as

opposed to a more conventional deductive presentation of grammar.

3.3 Initial Experiment

The method and materials here described constitute a second version of the experiment.

An initial experiment was run during the second semester of the academic year

2003/2004. A number of factors rendered the results of that experiment most unreliable.

These included the fact that the post-test was misguidedly included as part of the

students’ end of year grammar exam and, therefore, many of the subjects who had

followed an inductive approach were highly motivated to revise the grammar points in

question via reference books and other purely deductive means, which severely

distorted the findings as it was impossible to determine to what extent their use of the

concordancer had influenced their post-test result.

Furthermore, the pre and post-tests covered too wide an area of grammar, focusing

overmuch on low frequency linking expressions, such as lest and albeit, which it was

felt the students would be unlikely to have met but which they could ‘learn’ via the

concordancer. Correspondingly, the hands-on corpus-based lesson, on which the

research was based, was itself too lengthy and many students did not finish all the

activities, which again impacted negatively on their test results as they had not had

sufficient time to study some of the items included in the post-test.

It was also felt that students should have had more contact with hands-on concordancing

prior to the experiment, as their unfamiliarity with the tool hampered their progress

through the activities.

For all of these reasons the experiment was re-run using a different set of subjects at the

end of the first semester of the academic year 2004/2005.

66

3.4 The Target Language

The area of grammar studied in the experiment was that of connectors, specifically

linking adverbials (conjuncts) e.g. nevertheless, subordinators e.g. although and

prepositions e.g. despite.

This topic was chosen because it fitted in to the students’ third year grammar syllabus,

which focused on the complex sentence and the grammar of adverbials. Moreover,

connectors lend themselves to a corpus-based treatment since their different semantic

roles make them more like lexical items, and it is quite easy to produce concordance

lists to reveal their use and meaning.

The virtual lesson (in both modes of instruction) deals with the following selection of

linkers of each type, chosen because they share similar semantic roles:

Linking Adverbials Subordinators Prepositions (conjuncts)

Time (progressive) meanwhile while during

Cause / Reason for this reason because / since due to

Consequence therefore so as a result of

Time (sequential) afterwards once following

Concession nevertheless although despite nonetheless though in spite of

even so even though notwithstanding though whereas

still

67

Table 1. Linkers studied in the virtual lesson.

3.4.1 Linking Adverbials

Biber et al (1999) describe the primary function of linking adverbials as ‘to state the

speaker/writer’s perception of the relationship between two units of discourse.’ They

can be seen as signaling devices, highlighting the links between passages of text, and

are thus fundamental in providing text cohesion. Greenbaum and Quirk (1985) refer to

them as ‘conjuncts’, the term used in the materials here described.

Biber et al identify the following semantic meanings of linking adverbials: addition,

summation, apposition, result/inference, contrast/concession and transition.

Biber et al also reveal interesting corpus findings regarding the distribution of these

semantic categories across the four registers of academic prose, conversation, fiction

and news.

Linking adverbials are far more frequent in academic prose and conversation than in

news and fiction. A large number of the linking adverbials used in academic prose and

conversation express result/inference. Of particular pertinence to the present study, there

is a similar level of frequency of contrast/concession adverbials in conversation, fiction

and academic prose but far fewer in news registers.

As the above table shows, linking adverbials are syntactically realized in various ways.

The most common form is the single adverb e.g. nevertheless, but adverb phrases e.g.

even so and prepositional phrases e.g. for this reason are also found. Biber et al detail

how corpus findings reveal that ‘in conversation almost all linking adverbials are single

adverbs’ while in academic prose ‘prepositional phrases are also relatively common.’

Of interest to this study, the linking adverbials so and though are shown to be far more

common in conversation than in academic prose. So is often used in narrative accounts

to move a story along. Though is used in oral discourse as speakers mark contrasts

between one statement and another, as shown by the following example, taken from

Biber et al p.888 (punctuation unchanged):

68

So it should have everything, I still think it’s a bit expensive though.

The corpus findings also reveal something about the most common positions of linking

adverbials. In both conversation and academic prose they are most commonly found in

initial position. Indeed, So cannot occur in any other position and Still almost always

appears at the front of a sentence. Final position is the second most common position

that linking adverbials take in conversation, though in particular being very common at

the end of a sentence, as in the above example. (In fact though cannot be used as a

linking adverbial in initial position, only as a subordinator.) Medial positions are very

rare in conversation but are much more common in academic prose, where final position

is rare. Biber et al suggest that though as a linking adverbial is sometimes used in

writing that is informal or intended to resemble speech.

3.4.2 Subordinators

Also known as subordinating conjunctions, these introduce (mainly finite) subordinate

clauses. Most subordinators introduce adverbial clauses, and it is on these that the

present study focuses. Adverbial clauses express a wide range of semantic relationships,

including time, manner, reason/cause, result, concession and condition. The

subordinator signals the semantic relationship between the sub- and superordinate

clauses in a sentence.

e.g. He was screaming because he had to go home. (taken from Biber et al p. 134)

Most subordinators are single words, but there are various complex subordinators made

up of two or more words which can introduce an adverbial clause e.g. even though.

Some subordinators can have more than one semantic meaning. For example, while is

used for both time and concession/contrast. Its use to mark a temporal relationship

occurs much more often in conversation, while its use to signal a concessional meaning

is far more common in academic prose (as this very sentence demonstrates). Similarly,

since can express both temporal and causal relationships, the former much more

frequently in conversation and the latter in academic prose.

69

Biber et al explain that there is an overlap between subordinators and other word

classes. Indeed, the virtual lesson looks at three such linkers which can each function in

more than one way, namely since (preposition and subordinator), besides (preposition

and linking adverbial) and though (subordinator and linking adverbial).

Across all four registers surveyed by Biber et al most non-finite adverbial clauses are

not introduced by subordinators. The register with the largest proportion of non-finite

clauses is academic prose. In this type of discourse, and in news, non-finite clauses

express time relationships more overtly than in conversation, due to the presence of

subordinators which remove any ambiguity.

e.g. He was accused of molesting a 14-year-old boy whom he had been counseling while

working as a school chaplain (from Biber et al p. 840)

Without the subordinator the two clauses in this sentence could be construed as having a

causal relationship.

Subordinators of purpose, time and similarity/comparison are the most common types to

signal non-finite clauses.

Biber et al point out that

many semantic categories have multiple subordinators; reason clauses can be introduced by because, since and as, for example.

Similarly, clauses of concession can be signaled by although, though and even though.

Of particular interest to the present study, corpus findings in Biber et al reveal that

regarding these concessive subordinators, they occur in all four registers, albeit with

different preferences of use. Though is used more in conversation and fiction, while in

academic prose although is about three times as frequent. Biber et al justify suggest this

may result from an attempt to distinguish this subordinator from the common use of though as a linking adverbial in conversation.

70

3.4.3 Prepositions

Prepositions are linkers which commonly connect with a noun phrase to form a

prepositional phrase, although they can link to a non-finite -ing clause, e.g.

he said Noreiga had called two or three times since turning himself over to U.S. forces.

(taken from Biber et al p.77)

In this case there is some ambiguity over whether the preposition is in fact a

subordinator. Another case in point is despite which frequently introduces non-finite

-ing clauses.

All the instances in this study are free prepositions; that is they are not dependent on any

specific words in the sentence.

Though most are single words, some are complex prepositions made up of two or more

words e.g. as a result of, in spite of and due to.

Biber et al point out that there is an overlap between prepositions and other word

classes. For example, with regard to prepositions in the present study, since can also

function as a subordinator and following is more commonly a verb form.

Finally, the preposition notwithstanding is included in the study – as a deliberately

challenging item to fully test the capacity of the inductive materials to help students

understand its use and meaning.

Notwithstanding is irregular in that it can appear in end position following a noun

phrase, especially in formal academic prose.

e.g. Teachers get irritated, of course they do, their elaborate and expensive training courses

notwithstanding. (taken from the BNC)

71

3.5 Method

Essentially, all the subjects were first given a questionnaire to identify their learning

style in respect to whether they favour a more or less inductive approach to learning

grammar. On this basis, the subjects were divided into two groups; inductive learners

and deductive learners. All the students then did the same pre-test.

Two approaches to the study of the same grammar area (linking devices – see section)

were devised, one an inductive approach using hands-on concordancing and the other a

more conventional deductive approach. Both were supplemented by a series of online

practice exercises designed using Hot Potatoes. Both approaches to the grammar were

computer-based with written study guides accompanying them.

All the subjects were asked to work through an online grammar lesson accessed via a

website set up specially for the experiment. The total number of 56 students were

divided into four groups. Half of the ‘inductive’ learners followed the inductive

approach while the other half followed the deductive approach. Similarly, the

‘deductive’ learners were divided into two groups each following one of the two

approaches.

Two weeks after completing the online lesson, all the subjects did the same post-test. By

comparison of the results between the two tests, conclusions could be drawn about

whether there was indeed any identifiable link between learning style and the type of

approach followed by each individual.

3.6 The Concordancer

The concordancer used for this project forms parts of the Corpógrafo ‘an integrated

web-based environment for corpus linguistics and knowledge engineering’, developed

at FLUP as part of Linguateca, ‘a distributed resource centre for Portuguese whose main

72

aim is to foster R&D in the processing of the Portuguese language.’ (MAIA, SANTOS

& SARMENTO, 2004). The Corpógrafo makes freely available on the web a

comprehensive set of text and language tools including a concordancer linked to the

BNC (British National Corpus). As a consequence, it employs the BNC’s own system

of grammatical tagging, which proved most useful in enabling more complex, specific

searches.

3.7 The Subjects

The experiment was run using all 56 available students following the English 3 course

at FLUP. These students are taking a degree in Línguas e Literaturas Modernas (LLM)

and as part of this wider course study four years of English language. The students are

divided into three variants: English/Portuguese, English/French and English/German.

Additionally, the students belong to one of two ramos or branches, namely educação or

cientifico. Essentially students of the former are hoping to eventually become teachers

whilst the latter hope to work in the field of translation.

The English language component of their course follows a broad approach, comprising

reading and writing skills, development of oral fluency and grammar learning. In

English 1 students revise the pedagogic grammar that they studied at school. Thereafter,

a more systematic approach is taken, based on A Student’s Grammar of English (Quirk

and Greenbaum, 1985) In English 2 the students focus on the elements of the simple

sentence, with particular emphasis on the verb phrase. In English 3 focus shifts to the

noun phrase and the complex sentence, with particular emphasis on adverbial clauses

and the role of conjunctions. English 4 goes on to examine text grammar and emphatic

structures such as fronting and inversion.

The students engaged in this project were therefore familiar with the simple sentence

and its constituent elements. Revision of the year 2 material was conducted at the

beginning of the semester such that by the time the students participated in this project

they were familiar with the terminology used. This was the first time that they would

examine in detail the differences between conjuncts and conjunctions.

73

As English 3 students they were also used to working with computers and using CALL

to study grammar, as the whole grammar syllabus is presented online. They were not so

familiar with using concordancers, though some had used them extensively as part of

their translation studies. Some effort was made to introduce students to concordancing

prior to the experiment, chiefly with activities focused on vocabulary and idioms.

Of the 56 subjects, 48 were female and 8 male.

3.8 The Questionnaire

The same questionnaire was given to all 56 subjects (see Appendix I).

Questions focused on the inductive/deductive axis of the Felder-Silverman Learning

Style Model. The main source was Rebecca Oxford’s Style Analysis Survey (1995) but

some questions were also drawn from How to Identify your Best Learning Styles

(Chapter 13 from How to Be a Successful Student) by Don Martin at College of Marin

(1991);

Other sources, though not focusing directly on deductive/inductive learning style, did

provide useful advice about building a questionnaire. These included The Translator as

Learner (Chapter 3 of Becoming a Translator) by Douglas Robinson (1997); Ellis and

Sinclair’s Learning to Learn English (1989); and the Index of Learning Styles (ILS)

Questionnaire developed by Soloman and Felder at North Carolina State University

(1995).

The latter, though based on the Felder-Silverman Learning Style Model, deliberately

chooses not to include questions to assess the inductive/deductive dimension. Felder

explains why in the FAQ section of his site:

Barbara Soloman and I don't want instructors to be able to give our instrument to students, find that the students prefer deductive presentation, and use that result to justify continuing to use the traditional deductive instructional paradigm in their courses and curricula. (Felder, 1995)

74

22%

78%

Inductive LearnersDeductive Learners

Results of Learning Style questionnaire

12 Inductive Learners44 Deductive Learners

Fig. 3 Results of Learning Style questionnaire

The subjects were then divided into the following four groups:

Group 1 (6 students) ‘Inductive’ Learners / Inductive (Corpus-based) approach

Group 2 (22 students) ‘Inductive’ Learners / Deductive (traditional rule-based) approach

Group 3 (6 students) ‘Deductive’ Learners / Inductive (Corpus-based) approach

Group 4 (22 students) ‘Deductive’ Learners / Deductive (traditional rule-based) approach

3.9 The Pre-Test

All the subjects were given the same pre-test (see Appendix II). This consisted of

twenty sentences, taken from the BNC, each with a linker removed and a list of four

options given with which to fill in the blank space.

75

The mean average score obtained by the group of six inductive learners who were to

follow the inductive approach was 14.1. That obtained by the group of six other

inductive learners who followed the deductive approach was 13.8. The average score

obtained by the group of 22 deductive learners who were to follow the inductive

approach was also 13.8. Finally, the average score of the group of 22 other deductive

learners who followed the deductive approach was 11.3.

Pre-test Results

Ind/ Ind 14,1

Ind/Ded 13,8 Ded/ Ind

13,8Ded/Ded

11,3

02468

101214161820

Pre-test

Fig. 4 Average totals scored in the pre-test by each of the four groups (learning style / approach).

3.10 The Inductive Lesson

The 28 students who followed the ‘inductive’ approach were divided into two batches

each of which spent a two-hour session in a computer room, with each student working

individually at a separate terminal. They were given a printed handout (see Appendix

III) which guided them through the lesson, though onscreen instructions were also given

and students were encouraged to follow these (since they would at certain times need to

copy/paste specific tagged search codes into the concordancer).

76

After accessing the website http://web.letras.up.pt/jlewis , the students selected

Inductive mode and then logged into the Corpógrafo, using a specially created

universal username and password, namely ‘english3’. They were then able to access the

BNC Concordancer. It was explained that the Concordancer could thereafter be

accessed either by clicking on ‘concordancer’ in the navigation bar at the top of the

screen or by clicking on ‘Concordância BNC’ in the window bar at the bottom of the

screen.

There followed a brief explanation of what a concordancer is and how to use it. See

screenshot below.

Fig. 5 Screenshot of ‘Set Up’ page from Inductive Mode

In a few simple steps students were taken through the process of how to enter words

into the concordancer and view the outcome of their search. They were encouraged to

begin by selecting words of their own choice before then viewing the results obtained

for the word “while”. The study guide informed students that none of the instances

77

show while at the beginning of a sentence and that in order to produce these they would

have to type in the code “W|while”. It was explained that in order to save time, it was

sufficient to simply copy/paste the code from the onscreen instructions into the

concordancer’s search box. See screenshot below.

Fig. 6 Screenshot of BNC Concordancer Interface

Instructions were supplied for those students who perhaps did not know how to

copy/paste using the right mouse key. This need to copy/paste the search codes was

considered very important so as both to save time and also to ensure that the codes were

entered correctly, seeing that many of those used later in the lesson were particularly

lengthy.

It should be noted in the screenshot above that the Corpógrafo divides the corpus into

10 sections labeled A to K. Students could, therefore, at any time select different

sections of the corpus and view the concordances there offered. In practice, however,

the concordancer was defaulted to section A and this provided sufficient input for

grammar analysis to be made.

78

After this brief setting up procedure, the students began the online grammar lesson

proper.

The first exercise asked them to look at the 15 linking expressions in the following box.

during therefore for this reason

so because as a result of although

afterwards despite meanwhile nevertheless

following due to once afterwards

Fig. 7 Box of linkers to classify, from Exercise 1 of the Inductive Approach

They then had to fill in the chart below so that the items in each row shared a similar

meaning while those in each column shared the same grammatical function. The chart

has been completed to show the answers.

Group A Group B Group C

time (progressive) meanwhile while during

cause/reason for this reason Because due to

consequence therefore so as a result of

time (sequential) afterwards once following

concession* nevertheless although despite

* (concessive) linkers join one idea or fact to another which is opposed to it.

Table. 2 Completed chart from Exercise 1 of the Inductive Approach

At this stage the students were not given any terminology. They simply had to try to

work out how the linkers are used by studying the concordance lines. The teacher went

around the classroom helping students to recognize patterns and by looking closely at

what type of structures precede or follow the linkers, work out which of them

functioned in similar ways. They were encouraged to look at the concordances for the

79

three examples already filled in, namely ‘meanwhile’, ‘while’ and ‘during’. For

example, attention was drawn to the fact that despite is commonly followed by a noun

phrase and that although is commonly followed by a subject and a verb (i.e. a clause).

This extra help given by the teacher was in fact very important and many students

working on their own in isolation would probably not have filled in the chart correctly.

They needed to be nudged in the right direction and may not have otherwise noticed the

patterns – at least not in this initial exercise.

In order to save the students from having to type in the codes every time, they were

given an onscreen chart from which they could copy/paste the relevant codes. See

screenshot below.

Fig. 8 Screenshot of chart from Ex.1 of the Inductive Approach, containing codes (in red) of grammar items to enter into the concordancer.

It should be noted that some of the codes are especially lengthy, so as to focus on

particular instances of the words in question and exclude other uses which would be

potentially confusing. For example, ‘following’ is most commonly used as a verb, and

80

so to access its prepositional use a complex code had to be used. Similarly, the codes for

‘so’ and ‘once’ reveal only their use as conjunctions.

The screenshot below shows an example of concordance lines for ‘during’.

Fig. 9 Screenshot of concordances for ‘during’.

After completing the chart in their handout, the students clicked on Go to Ex.2 Part i).

This took them to a page showing the completed chart for them to check and with

instructions to then use the concordancer, with the same onscreen codes, to answer a

series of multiple choice questions in their handout.

The following shows the first of these questions (for complete list see Appendix III).

1. Which group of linkers can usually be omitted without affecting the grammar/syntax of the sentence?

Group C ٱ Group B ٱ Group A ٱ

81

The aim of this exercise was to focus students’ conscious attention on the grammatical

differences between the linkers, looking again at such things as punctuation, word order

and what type of structures typically follow the linkers.

The last three of these questions asked students to identify what each type of linker is

called – conjuncts, conjunctions or prepositions. This information they, of course, were

unlikely to have known, so they were encouraged to make a guess.

Having completed this multiple choice exercise on their handouts, the students were

then told to do online Ex.2.ii). This was essentially the same exercise but completed

online using one of Hot Potatoes’ exercise templates. See below.

Fig 10. Screenshot of Interface for Exercise 2 of the Inductive approach, made using Hot Potatoes software

A smiley face appeared when the correct option was chosen. Additional feedback

appeared in the space at the top of the screen as well as the student’s cumulative

percentage score. The students had already done the necessary thinking, so used this

exercise as a means of checking their answers. It would have been much too easy if they

82

had done this online version of the multiple choice exercise first, as they would have

been tempted merely to click on box A, B or C until the smiley face appeared, without

necessarily thinking about the question, much less actively looking at the concordance

lines as they did so when completing the exercise on their handout.

Exercise 3 focused on three linkers – since, besides and though – which can function in

more than one way. By consulting the corpus students had to determine what two

functions each linker can perform and in the case of since how the meaning changes as

well. As before, the students were supplied with the necessary codes and feedback was

conducted in the same way via a Hot Potatoes exercise.

Exercise 4 gave the students the chance to manipulate some of the linkers they had been

studying by joining two short sentences together into one using a given linker. This was

done using another of Hot Potatoes’ exercise templates (see screenshot below).

Fig 11. Screenshot of Interface for Ex. 4 of the Inductive approach, made using Hot Potatoes

83

The extent to which the computer appears ‘intelligent’ in such an exercise relies on the

extent to which the teacher predicts the students’ responses. Effort was made to

anticipate several possible correct answers.

The remainder of the lesson focused on more linkers with a concessive meaning. As in

the first exercise, students were given a box containing seven linkers.

whereas notwithstanding still nonetheless

even so even though in spite of

Fig 12. Box of linkers from Exercise 5 of the Inductive Approach

They had to put the linkers into a chart, this time labeled with the terms conjuncts,

conjunctions and prepositions and already containing the concessive linkers

encountered in the lesson thus far.

Group A

Conjuncts

Group B

Conjunctions (subordinators)

Group C

Prepositions

nevertheless

though

although

though

despite

Table 3. Chart to be completed in Exercise 5 of the Inductive Approach

84

As before, an onscreen chart with the appropriate codes was supplied. See below.

Fig. 13 Screenshot of chart from Ex.5 of the Inductive Approach, containing codes (in red) of grammar items to enter into the concordancer

This was especially important in the case of still so as to reveal its function as a

conjunct. It was thus decided to limit instances of still to those where it appears in initial

position. Even so, many of the resulting concordances also show how still can be a

disjunct or an adjective. It was felt, however, that following the understanding they had

built up thus far, students would be able to pick out the relevant instances, namely those

where still is followed by a comma See screenshot below.

85

Fig. 14 Screenshot of concordances for ‘Still’ as a conjunct.

The screenshot below shows the completed chart.

Fig. 15 Screenshot of completed chart for Ex.5 and table of codes for Ex. 6

86

Exercise 6 got students to look more closely at though, notwithstanding and even so by

working through a multiple choice exercise on their handout. This exercise aimed to

show students how when though is used as a conjunct it appears at the end of a

sentence, especially in more informal contexts. The focus on notwithstanding was

aimed at showing how this linker is used in very formal writing and can appear at both

the beginning at end of a noun phrase, something that sets it apart from despite and

which students would not be familiar with.

The code for though (see highlighted in blue in the screenshot above) was again

designed so as to highlight its use as a conjunct. A selection of corresponding

concordances appear below.

Fig. 16 Screenshot of concordances for ‘though’ as a conjunct

The final two exercises (Ex. 7 and 8) used Hot Potatoes templates to provide students

with further practice opportunities at manipulating the forms. The first of these was a

straightforward gap-fill exercise using the linkers studied (see below).

87

Fig. 17 Screenshot of interface for Ex. 7, made using Hot Potatoes software

The final exercise provided students with an opportunity to join together sentences using the linkers given.

Fig. 18 Screenshot of interface for Ex. 8, made using Hot Potatoes software

88

3.11 The ‘Deductive’ Lesson

The 28 students who followed a more traditional rule-based approach were given a

handout (see Appendix IV) containing an explanation of the same grammar points

covered by the ‘inductive’ lesson. No attempt was made here to guide the learner to an

understanding of the grammar and there was no recourse to the concordancer. These

students were expected simply to read about the three types of linkers – conjuncts,

conjunctions and prepositions – and apply the knowledge thereby gained to the same set

of online ‘Hot Potatoes’ exercises as encountered in the ‘inductive’ lesson. Both

approaches thus used the same practice exercises but differed considerably in the way

the grammar was presented.

This group of students likewise took part in an online lesson, accessing the same

website but selecting ‘deductive mode’. The information could also be read onscreen –

see screenshot below.

Fig. 19 Screenshot of information about conjuncts from Ex. 1 of the deductive mode

3.12 The Post-Test

Two weeks after the online lesson, all 56 students did the same post-test (see Appendix

5). This was essentially identical to the pre-test with only the contexts changed.

89

Chapter 4: Results

This chapter presents the findings of the post-test, with accompanyingstatistical analysis, and results of the post-project questionnaire.

90

4.1 Post-test Results

14,115,113,8

11,513,8 12,9

11,3 11,7

02468

101214161820

Ind/ Ind Ind/Ded Ded/ Ind Ded/Ded

Pre-testPost-test

Fig. 20 Chart showing the average scores obtained by each group in the post-test compared to those in the pre-test. (Ind/Ind = Inductive learners with inductive approach.)

'Inductive' Learners / Inductive (Corpus) Approach

6 students

50%

50%

highersame

'Inductive' Learners / Deductive Approach

6 students 17%

0%

83%

highersamelower

'Deductive' Learners / Inductive (corpus) Approach

22 students27%

13%60%

highersamelower

'Deductive' Learners / Deductive Approach

22 students45%

23%

32%highersamelower

Fig. 21 Charts showing to what extent the post-test scores obtained by students in each group were higher than the pre-test.

91

The average score obtained in the post-test by the group of 6 ‘inductive’ learners who

followed the inductive approach was 15.1, an increase of 7.8% on the pre-test. Of these

students, 50% had the same score in the two tests while the other 50% scored higher.

The average score obtained by the 6 ‘inductive’ learners who followed the deductive

approach was 11.5, a 20% decrease from the pre-test. Of these, 83% scored lower in the

post-test and only student scored higher.

The group of 22 ‘deductive’ learners who followed the inductive approach had an

average score in the post-test of 12.9, a 7% decrease from the pre-test. 60% of the

scores in this group were lower in the post-test, 32% were the same and 45% higher.

The group of 22 ‘deductive’ learners who followed the deductive approach had an

average score in the post-test of 11.7, an increase of 3.5% on the pre-test. 60% of the

scores in this group were lower in the post-test, 32% were the same and 45% higher.

92

4.2 Statistical Analysis

4.2.1 Correlations

In order to create one variable for statistical analysis, the 56 students were divided into

two groups: those where learner style matches the approach followed and those where

there is no match. A second variable was created by calculating the differences between

the pre and the post test. Using SPSS, this information was entered and a correlation

analysis was performed (see table below) to measure the degree or strength of a linear

association between the two variables.

Such analysis reveals the Pearson correlation coefficient, which is a numerical measure

of linear association between two variables that takes the values between -1 (perfectly

strong and indirect relationship) to +1 (perfectly strong and direct relationship). Values

near zero indicate a lack of linear relationship. In this case the correlation value is 0.315,

indicating a ‘significant’ degree of relationship between pre/post test differences and

matching learner style/approach. As the table below shows, the correlation value is

significant at the 0.05 level. Thus it 95% certain that the correlation is not accidental.

The value of 0.018 is the p-value associated with the correlation; that is the probability

of getting a result as extreme or more extreme than the one observed if the proposed

null hypothesis is correct. A p-value of .018 indicates that there is only a 1.8% chance

of drawing the sample being tested if the null hypothesis was actually true. Thus, it

confirms the statistical significance of the correlations.

93

Correlations

1 ,315*, ,018

56 56,315* 1,018 ,

56 56

Pearson CorrelationSig. (2-tailed)NPearson CorrelationSig. (2-tailed)N

Difference betweenPost and Pre Test

Learner Style

Differencebetween Postand Pre Test Learner Style

Correlation is significant at the 0.05 level (2-tailed).*.

Table 4: SPSS Correlations table showing strength of relationship between pre/post test results and matching learner style/approach4.2.2 Regression

However, correlations do not show direction of causality i.e. that matching learning

style/approach was responsible for the difference in test results. Therefore, a simple

linear regression analysis was performed, confirming that matching learner

style/approach is a significant predictor of the difference between the pre and post-test

results.

The Model Summary table below contains the necessary summary statistics for

assessing the accuracy of the estimated sample regression line (SRL). R is the sample

correlation coefficient. The meaning of r = .351 is the same as the Pearson correlation

value given earlier – that the relationship between pre/post test result differences and

matching learner style/approach is significant.

R-square is the sample Coefficient of Determination, which measures the goodness-of-

fit of the estimated SRL in terms of the proportion of the variation in the dependent

variable explained by the fitted sample regression equation or SRL. Thus, the value

of .099 means that 9.9% of the variation in pre/post test results is accounted for by the

predictor of difference, in this case matching learner style/approach.

Adjusted R-Square is the sample Coefficient of Determination after adjusting for the

degrees of freedom lost in the process of estimating the regression parameters.

Standard Error of the Estimate is a summary statistic measuring the overall accuracy or

quality of the estimated SRL in terms of the average unexplained variation in the

94

dependent variable that may be due to possible errors that could originate from chance

sampling errors, or possible variation in the parameter.

Model Summaryb

,315a ,099 ,083 2,59229Model1

R R SquareAdjustedR Square

Std. Error ofthe Estimate

Predictors: (Constant), Learner Stylea.

Dependent Variable: Difference between Post and PreTest

b.

Table 5: SPSS Model Summary table showing results of simple linear regression analysis.

An ANOVA (Analysis of Variation) test (or F-test) was also run (see table below). It

tests the difference in means between two groups, in this case those students whose

learner style matched the learning approach and those where there was no match. It

calculates the total sum of squares in the dependent variable i.e. pre/post test

differences. The term Regression refers to the explained sum of squares, while Residual

refers to the remaining sum of squares which is unexplained and attributable to errors.

‘df’ refers to the degrees of freedom associated with the sources of variance. F here

refers to the variance or difference between the means of the two groups and the figure

of 5.949 shows significant variance. .018 is the p-value as explained above.

ANOVAb

39,978 1 39,978 5,949 ,018a

362,879 54 6,720402,857 55

RegressionResidualTotal

Model1

Sum ofSquares df Mean Square F Sig.

Predictors: (Constant), Learner Stylea.

Dependent Variable: Difference between Post and Pre Testb.

Table 6: SMSS ANOVA, or F-Test, table, showing analysis of variation between the means in each group.

The table below contains the estimated regression coefficients.

The figure of 1.691 represents the marginal effect of matching learner style/approach on

the mean difference between pre and post test results. The figure of .693 is a measure of

95

the precision of the estimated value of 1.691. The Standardized Coefficient, or beta

coefficient, is the same as the correlation coefficient R = .315. This confirms once again

that matching learner style/approach is an significant determinant of mean test result.

Coefficientsa

-2,863 1,084 -2,641 ,0111,691 ,693 ,315 2,439 ,018

(Constant)Learner Style

Model1

B Std. Error

UnstandardizedCoefficients

Beta

StandardizedCoefficients

t Sig.

Dependent Variable: Difference between Post and Pre Testa.

Table 7: SPSS Coefficients table showing estimated regression coefficients.

4.2.3 T-Test

An independent samples T-Test was also run which shows that the difference between

the means of the test results is statistically significant.

Group Statistics

Learner Style N Mean Std. Deviation

Std. Error

Mean

Difference between Learner Style Does

Post and Pre Test Not Match Approach

Learner Style

Matches Approach

28

28

-1,1724

,5185

2,42117

2,76476

,44960

,53208

Table 8: SPSS T-Test table showing the descriptive statistics for the two groups, that where learner style matches approach and that where it does not.

The Independent Samples Test table below shows the Levene's Test for Equality

of Variances. This indicates whether the two groups have approximately equal variance

on the dependent variable. The figure of .957 in the Sig. Column shows that the two

variances are not significantly different and indeed display almost equal variance.

Since the significance for Levene's test is above 0.05, then the "Equal Variances Assumed" test is used (the one on the top).

96

The T value is -2.439 with 54 degrees of freedom. The p-value of.018 shows that there is a significant difference between the two groups.

Independent Samples Test

,003 ,957 -2,439 54 ,018 -1,6909 ,69326 -3,08084 -,30103

-2,427 51,842 ,019 -1,6909 ,69660 -3,08886 -,29301

Equal variancesassumedEqual variancesnot assumed

Difference betweenPost and Pre Test

F Sig.

Levene's Test forEquality of Variances

t df Sig. (2-tailed)Mean

DifferenceStd. ErrorDifference Lower Upper

95% ConfidenceInterval of the

Difference

t-test for Equality of Means

Table 9: SPSS Independent Samples Test

4.3 Post-project Questionnaire

The students in the two groups that followed the inductive, corpus-based approach were

given a questionnaire to obtain feedback on their learning experience. (See Appendix 6)

Questions focused on what they thought about studying grammar using the

concordancer.

In response to the question ‘How easy was the Concordancer tool to use?’ 66% of

students in each of the groups responded that it was ‘easy’ and 17% said it was ‘very

easy’. 17% of students in each of the groups said that it was ‘difficult’.

When asked about how useful they found the Concordancer tool as a means of studying

grammar, two out of the six ‘inductive’ learners who had used it (33% of the group)

responded that it was ‘very useful’ and the other four students in this group (67% of the

total) said that it was ‘quite useful’. Of the 22 deductive learners who used it, 82%

thought that it was ‘quite useful’, 9% that it was ‘very useful’ and 9% that it was ‘not

very useful’.

All of the ‘inductive’ learners thought that it was ‘easy’ to identify patterns of grammar

and work out rules using the Concordancer tool. 57% of the ‘deductive’ learners

thought likewise while 43% of this group thought that it was ‘difficult’.

97

67% of the ‘inductive’ learners would also have liked to have the grammar rules given

in a more traditional way whilst 33% said they would not. All but 5% of the ‘deductive’

learners responded that they would have preferred a more traditional approach.

The ‘inductive’ learners were divided when it came to the question of whether they

would like to study grammar in this way again in the future, with three of them (50%)

saying they would like to and the other three saying they would not. 64% of the

‘deductive’ learners said they would like to use the tool to study grammar in the future

with 36% saying they would not.

The following charts serve to illustrate these findings

Fig. 22

How easy was the Concordancer tool to use?

Inductive Learners / Inductive Approach

17%66%

17%

Very EasyEasyDifficultVery Difficult

Deductive Learners / Inductive Approach

17%

17%

66%

Very EasyEasyDifficultVeryDifficult

98

Fig. 23

How useful did you find the Concordancer tool as a means ofstudying grammar?

Inductive Learners / Inductive Approach

33%

67%

VeryQuiteNot veryNot at all

Deductive Learners / Inductive Approach

9%9%

82%

VeryQuiteNot veryNot at all

Fig. 24

How easy was it for you to identify patterns of grammar and work out ‘rules’ using the Concordancer tool?

Inductive Learners / Inductive Approach

100%

Very Easy

Easy

Dificult

VeryDifficult

Deductive Learners / Inductive Approach

43%

57%

Very Easy

Easy

Difficult

VeryDifficult

99

Fig. 25

Would you also have liked to have the grammar rules givento you in a more traditional and straight forward way?

Inductive Learners / Inductive Approach

67%

33%YesNo

Deductive Learners / Inductive Approach

95%

5%

YesNo

Fig.26

Would you like to study grammar in this way again in the future?

Inductive Learners / Inductive Approach

50%

50%YesNo

Deductive Learners / Inductive Approach

36%

64%

YesNo

100

101

Chapter 5: Conclusions

This chapter offers a conclusion based on the results of the experiment and discusses lessons to be drawn and plans for future research.

5.1 Conclusions

The findings suggest strongly that learning style, more specifically the learner’s

preferred way of processing new information – whether tending more towards an

inductive style or a deductive style – does indeed have an influence on the effectiveness

of using a corpus-based approach to present grammatical concepts.

In the context of third year language students at FLUP, it would appear that a guided-

discovery approach involving learners in the process of inducing rules from analysis of

examples (in this case concordance lines) strongly favours those with a more inductive

learning style.

Perhaps the more telling conclusion, however, is that the converse is also true and that

such a hands-on approach using onscreen concordances (albeit filtered to some extent

using pre-selected codes) does not, in fact, favour the majority of students, who prefer a

more deductive means of learning. Indeed, the results suggest that many students were

102

confused by the inductive online ‘lesson’ and did not learn how the linkers are used or

what the differences in their grammatical properties are.

Interestingly, the same proportion (two-thirds) of students in the two groups who used

the corpus-based approach said they found the concordancer tool easy to use. This

suggests that the reason for the disparity in what the two groups actually learned was

not due primarily to any perceived difficulty that the students had in using the

concordancer. The reason, it would seem, lies much more in their comparative cognitive

ability to process information, and the necessity to work out rules from raw data proved

in many cases to hamper the learning process and cause more confusion, whilst at the

same time appearing to aid those who readily recognize and formulate rules for

themselves. Indeed, 43% of the ‘deductive’ learners studying the corpora data did

express that they found it difficult to identify patterns and work out rules using the

concordancer.

Of course, it is likely that some students were unable to define clearly their preferred

learning style. Indeed, there were several cases where students’ responses to the initial

questionnaire showed discrepancies. It is important to stress that the 12 students who

were categorized as ‘inductive’ learners were only those who came out strongly

favouring an inductive approach on the basis of the questionnaire. A number of students

in the ‘deductive/deductive’ group may have actually been more inductively inclined

without necessarily knowing it, being predisposed towards a deductive approach to

learning because of their previous educational experience. Such students could well

have improved in the post-test as a result of using the concordancer.

It is interesting that a significant number (64%) of the ‘deductive’ students using the

tool said they would like to use it in the future, though the overwhelming majority of

this group said they would also like to have grammar rules presented in a more

traditional way. It would seem to be the case, therefore, that many students would

benefit from a mixture of the two approaches – a combination of explicit rules and the

opportunity to further analyse and confirm these rules through studying concordance

lines.

103

It is almost certain that the results would have been different if the students had had

much more experience using concordancers prior to the experiment. With this extra

training students may then have become more used to how to look for and identify

patterns. Then again, considering that all the students were in the same position

regarding lack of previous exposure to concordancers, the snapshot view offered by this

experiment does, nevertheless, suggest that a pure hands-on approach is especially

beneficial to those with a more inductive learning style.

In hindsight there are a number of things that could have been done differently in order

to clarify the grammar concepts and aid learning. Chief among these was the way in

which the concordance lines were arranged on the screen. It would have been better to

use a concordancer that displayed the lines with the keyword in the middle (a good

example being the Lextutor Online Concordancer detailed in Chapter 2). The main

reason why such a means of presentation would have been preferable is that many of the

connectors were adverbial conjuncts and as such link back to the preceding sentence, or

indeed paragraph. Using the BNC Interface belonging to Linguateca it was not possible

for students to look back at previous statements since the concordancer was

programmed only to show the sentences in which the keyword was contained. Thus the

students were unable to use the overall context to understand the meaning and use of the

linkers, something which is fundamental to an understanding of the differences between

‘even so’ and ‘even though’, for example. Another advantage of concordancers like the

Lextutor is that the user can click on the keyword in each concordance line and open a

separate window showing the section of the original text where the concordance was

drawn from.

Though it is an accepted prescriptive rule that conjuncts are separated by commas, this

is not always the case and native speakers often neglect to add this punctuation.

Therefore, many instances of linking adverbials in initial position were not followed by

a comma and many students were confused by this. Indeed, this raises another issue

arising from getting students to try to induce rules from raw corpora data. It was evident

that students were often drawn to the irregularities in a list of concordances and failed to

perceive that the majority of instances listed onscreen were actually following a

particular pattern. Again, this tendency not to see the wood for the trees, as it were,

could be partially rectified by giving students more pre-training in using concordancers.

104

In hindsight, some of the linkers did produce potentially confusing lists of

concordances, especially while, which is clearly a subordinator but often introduces

non-finite -ing clauses – not something the lesson ‘guide’ mentioned. Similarly once

caused some confusion as the data included a mixture of finite and non-finite clauses.

More sophisticated codes could perhaps be devised to combat this problem and filter out

the -ing clauses. Then again, it is important that students are aware that while and once

function in this way.

The key problem with this approach could have been in the central idea of getting

students to classify the linkers according to tight parameters and rigid patterns.

Exceptions are always then going to stick out like a sore thumb and potentially distract

students from recognizing the underlying regularities and patterns.

Finally, it would seem clear that if students struggled to induce rules about the use of

linkers – a comparatively straightforward area of grammar – then how much harder

would they find it to work out the patterns of use involved in more complex areas such

as instances of ‘if’ or modal verbs? An exclusively hands-on data-driven approach to

grammar study is probably only to be recommended for students who are clearly

inductive learners.

5.2 Future research

As mentioned above, this experiment offers very much only a snapshot view of how the

effectiveness of concordancing may be affected by students’ preferred learning style. In

order to more accurately determine a student’s optimum learning style, more than a

simple questionnaire is required. Students could be asked to reflect on how they study

over a longer period of time, and then answer a subsequent questionnaire.

In future, a wider sample of subjects should be tested, as here there were only a small

number of inductive learners. Clearly, other variables also influence students’

performance in tests like those administered here, including previous educational

background and facility for using computers. However, with more extensive learner

training in using concordancers, interference from these other factors can be reduced.

105

The experiment carried out here, was part of a wider project aimed at providing a fully-

fledged online grammar course for third year English students at FLUP. The results are

helpful in determining how best to incorporate corpora-based exercises of this nature

into a program of online grammar activities.

For the study of those grammar items, such as connectors, which lend themselves to a

corpus approach, one solution would be to provide two modes of study in much the

same way as demonstrated here. Students could be asked to do a questionnaire online,

which would automatically suggest that they were either inductive or deductive learners.

Students would then choose their mode of study accordingly with inductive learners

doing exercises involving the analysis of corpus-data using a concordancer, albeit with

guidance from an online tutorial, vital for providing feedback to their hypotheses about

grammar rules.

Another solution would be not to give students choice over whether they studied a given

grammar item deductively or inductively, but instead to vary the approach used so that

some virtual ‘lessons’ would be presented more traditionally and others using hands-on

concordancing. That way all the students would follow their preferred approach some of

the time, and would likely benefit from being offered a mix of lesson treatments.

From the point of view of teaching grammar, or at least facilitating its study, this

experiment has shown that even though a teacher may be very enthusiastic about a

particular approach to learning, he must pay attention to the needs and preferences of

his individual students and try to tailor the materials he produces (whether on paper or

online) in order to accommodate them. It is important, therefore that he does not expect

all students to respond equally to hands-on concordancing but should encourage them to

see it as a tool to aid learning. Otherwise he runs the risk of alienating students and

actively harming the learning process.

106

Bibliography

Abraham, R.G. (1985). Field independence-dependence and the teaching of grammar. TESOL Quarterly, 19; pp 689-702.

Baigent, M. (1999). Teaching in chunks: integrating a lexical approach. In Modern English Teacher 8 / 2.

Bedford, T. A. (2004). Learning styles: a review of literature. Toowoomba,  OPACS, The University of Southern Queensland.

Bednar, A.K., Cunningham, D., Duffy, T.M., & Perry, J.D. (1991). Theory into practice: How do we link? In G.L. Anglin (Ed.), Instructional technology: Past, present, & future (pp. 88-101). Englewood, CO: Libraries Unlimited.

Chomsky, N. (1987). The Chomsky Reader. New York: Pantheon

Biber, D., Conrad, S., & Reppen, R., (1994) Corpus-based approaches to issues in applied linguistics. In Applied Linguistics, 15 (2), pp 169-189.

107

Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). Longman Grammar of Spoken and Written English. London. Longman.

Brown, H.D. (1994) Principles of Language Learning and Teaching. Englewood Cliffs, NJ: Prentice-Hall.

Carroll, J. (1981). Twenty-five years of research on foreign language aptitude. In Diller, K. (ed), Individual differences and universals in language learning aptitude. pp. 83-118. Newbury House, Rowley, Massachusetts.

Chapelle, C. (2001). Computer Applications in Second Language Acquisition. Cambridge : Cambridge University Press.

Chapelle, C., & Jamieson, J. (1986). Computer-assisted language learning as a predictor of success in acquiring English as a second language. In TESOL Quarterly, 20 pp. 27-46.

Chapelle, C., & Green, P. (1992). Field independence/dependence in second language acquisition research. In Language Learning, 42 pp. 47-83.

Clarke, M (1992). Vocabulary learning with and without computers: Some thoughts on a way forward. Computer Assisted Language Learning, 5 (3), 139-146.

Cobb, T. (1997). From Concord to Lexicon: Development and Test of a Corpus-Based Lexical Tutor. PhD Thesis, Department of Educational Technology, Concordia University, Montreal, Quebec, Canada.

Cobb, T. (1997). Is there any measurable learning from hands-on concordancing? In System, 25 (3), pp.301-315. Division of Language Studies, City University of Hong Kong.

Crystal, D. (1995). The Cambridge Encyclopedia of the English Language. Cambridge: Cambridge University Press.

Downing, A., & Locke, P. (1992). A University Course in English Grammar. Hemel Hempstead: Prentice Hall.

Ellis, G., & Sinclair, B. (1989). Learning to Learn English. Cambridge: Cambridge University Press.

Felder, R.M., & Silverman, L. (1988). Learning and Teaching Styles in Engineering Education. In Engr. Education, 78(7); pp 674-681 University of North Carolina.

Felder, R.M., & Henriques, E.R.(1995). Learning and Teaching Styles in Foreign and Second Language Education. In Foreign Language Annals, Spring 1995: University of North Carolina.

108

Felder, R.M., & Soloman, B.A. (1995) Index of Learning Styles (ILS) Online Questionnaire. North Carolina State University.

Felder, R.M., Felder, G.N. & Dietz, E.J. (2002). The Effects of Personality Types on Engineering Student Performance and Attitudes. In Journal of Engineering Education, 91(1), 3-17.

Felder, R.M., & Spurlin, J. (2005). Applications, Reliability and Validity of the Index of Learning Styles. Pdf University of North Carolina.

Flowerdew, J. (1996). Concordancing in Language Teaching. In Pennington, M.C. (ed.), The Power of CALL. Athelstan.

Fortune, A. (1992). Self study grammar practice: learners' views and preferences. In ELT Journal 46 (2): pp 160-171.

Gardner, H. (1985). Frames of mind: The theory of  multiple intelligences. New York: Basic  Books Inc.

Goodfellow, R. (1995). The language learner and the computer - what's really going on in there? Interpreted results from a vocabulary-learning program. Pdf - Paper presented at Eurocall95, Valencia.

Greenbaum, S., & Quirk, R. (1985). A Comprehensive Grammar of the English Language. Harlow: Longman.

Gregorc, A. (1979). An Adult’s Guide to Style. Columbia: Gregorc Associates.

Halliday, M.A.K. (1973) Explorations in the Functions of Language. London: Edward Arnold.

Hanson-Smith, E. (1993). Dancing with concordances. CÆLL Journal, 4 (2), 40.

Harmer, J. (2001) The Practice of English Language Teaching. Harlow: Longman.

Higgins, J. (1988). Language, learners and computers: Human intelligence and artificial unintelligence. London: Longman.

Higgins, J., & Johns, T. (1984). Computers in Language Learning. Reading: Addison- Wesley.

Honeyfield, J. (1989). A typology of exercises based on computer generated concordance material. In Guidelines, 11(1), pp. 42-50.

109

Hymes, D.H. (1972) On Communicative Competence. In: Brumfit, C.J. & Johnson, K. The Communicative Approach to Language Teaching. Oxford: Oxford University Press,

Hunston, S. (2002) Corpora in Applied Linguistics. Cambridge: Cambridge University Press.

Hunston, S., & Francis, G. (1999) Pattern Grammar: A corpus-driven approach tot eh lexical grammar of English. Amsterdam: Benjamins.

Johns, T. (1991). Should you be persuaded: two samples of data-driven learning materials. In Johns, T., & King, P.(eds), Classroom Concordancing. English Language Research Journal, 4. (pp. 1-16). University of Birmingham: Centre for English Language Studies.

Johns, T. (1994). From printout to handout: grammar and vocabulary teaching in the context of data-driven learning. In Odlin, T (ed.), Perspectives on Pedagogical Grammar (pp. 293-313). Cambridge: Cambridge University Press.

Kettemann, B. (2000). Concordancing in English Language Teaching. Pdf Universtiy of Graz, Austria.

Kolb, D.A. (1976). Learning Styles Inventory. Boston: McBer.

Kolb, D. A. (1984). Experiential learning: Experience as the source of learning and development.  Englewood Cliffs, N.J.: Prentice-Hall.

Krashen, S. (1982) Principles and Practice in Second Language Acquisition. Oxford: Pergamon

Krashen, S. (1985). The Input Hypothesis: Issues and Implications. Longman

Larsen-Freeman, D., & Long, M.H. (1991) Explanations for differential success among second language learners. Ch.6 of An Introduction to Second Language Acquisition Research. London: Longman.

Leech, G., & Candlin, C.N. (1986). Computers in English language teaching and research. London: Longman.

Leech, G. (1997) Teaching and language corpora: a convergence. In Wichmann, A., Fligelstone, S., McEnery, T, & Knowles, G. (eds), Teaching and Language Corpora (pp. 1-24). New York: Addison Wesley Longman.

Lewis, M. (1996). The English Verb. Hove: Language Teaching Publications.

110

Lewis, M. (1997). Implementing the Lexical Approach. Hove: Language Teaching Publications.

Lyons, J. (1970) Chomsky. Fontana

Martin, D. (1991) How to Identify your Best Learning Styles. In How to be a Successful Student. San Anselmo, CA: Martin Press LLC.

McCarthy, B. (1990). Using the 4MAT system to bring learning styles to schools. In Educational  Leadership, 48(2), pp. 31-36.

Mindt, D.(2000). An Empirical Grammar of the English Verb System. Berlin: Cornelsen.

Myers, I.B. & McCaulley, M.H. (1985). Manual: A Guide to the Development and Use of the Myers-Briggs Type Indicator, Consulting Psychologists Press, Palo Alto, CA.

Nattinger, J., & DeCarrico, J. (1992) Lexical Phrases and Language Teaching. Oxford: Oxford University Press.

Oxford, R. (1995) Style Analysis Survey. In Reid, J. (ed), Learning Styles in the EFL / ESL classroom pp. 208-215. Boston: Thomson International.

Pinker, S. (1999) Words and Rules. Weidenfeld and Nicholson.

Prabhu, N. (1987). Second Language Pedagogy. Oxford: Oxford University Press.

Ranalli, J. M. (2001). Consciousness Raising versus deductive approaches to language learning: a study of learner preferences. Pdf file from Centre for English Language Studies, University of Birmingham, UK

Reiff, J. (1992). What research says to the teacher: Learning styles. Washington, DC: National Education Association.

Riding, R., & Cheema, I. (1991). Cognitive styles – an overview and integration. In Educational Psychology, 11 pp. 193-215.

Robinson, D. (1997) The Translator as Learner Ch. 3 of Becoming a Translator. London: Routledge.

Rutherford, W.E. (1987). Second Language Grammar: Learning and Teaching. Harlow: Longman.

Schmied, J. (2002) Exploring the Chemnitz Internet Grammar. Pdf Chemnitz University of Technology, Germany.

Schmidt, R. (1990). The Role of Consciousness in Second Language Learning. In Applied Linguistics, Vol. 11, no.2, pp 17-46.

111

Sinclair, J.M. (ed.) (1987). Looking Up: An Account of the COBUILD Project. London: Harper Collins.

Sinclair, J.M., & Renouf, A. (1988). ‘A lexical syllabus for language learning’ in Carter, R., & McCarthy, M. (eds) Vocabulary and Language Teaching. London: Longman.

Skehan, P. (1989). Individual Differences in Second-Language Learning. London: Edward Arnold.

Skehan, P. (1998). A Cognitive Approach to Language Learning. Oxford: Oxford University Press.

Stern, H.H. (1983). Fundamental concepts of language teaching. London: Oxford University Press.

Stevens, V. (1991). Concordance-based vocabulary exercises: A viable alternative to gap-fillers. In T. Johns., & P. King. (eds.), Classroom concordancing: English Language Research Journal, 4 (pp. 47-63). University of Birmingham: Centre for English Language Studies.

Thornbury, S. (1999). How to Teach Grammar. Harlow: Longman.

Tonkyn, A. (1994). Introduction: In Bygate, M., Tonkyn, A. & Williams, E. Grammar and the Language Teacher. Hemel Hempstead: Prentice Hall.

Tribble, C., & Jones, G. (1997). Concordances in the Classroom. Houston: Athelstan.

Trudgill, P. (1974) Sociolinguistics. Harmondsworth: Cambridge University Press.

Wilkins, D.A. (1976) Notional Syllabuses. Oxford. Oxford University Press.

Williams, M., & Burden, R.L. (1997) Psychology for Language Teachers. Cambridge. Cambridge University Press.

Willing, K. (1987). Learning Styles in Adult Migrant Education. Adelaide, Australia: Adult Migrant Education Programme.

Willing, K. (1989) Teaching how to learn: Learning strategies in ESL. Adelaide, Australia: Adult Migrant Education Programme.

Willis, D. (1990). The Lexical Syllabus. Glasgow: Collins Cobuild.

Willis, D. (2003). Rules, Patterns and Words: Grammar and Lexis in English Language Teaching. Cambridge: Cambridge University Press.

Witkin, H., & Goodenough, D. (1981). Cognitive styles; essences and origins. New York: International Universities Press.

112

Appendix 1Learner Style Questionnaire

NAME ______________________________ Teacher__________

113

1. When faced with a new learning situation ... ___a) I jump in feet first without thinking about it.

___b) I hang back to figure things out first.

2. When learning a new game...

___a) I prefer to learn the rules ‘as we go along’.

___b) I like to know all the rules before starting.

3. When learning new grammar...

___a) I prefer to look at some examples (e.g. some sentences or a text) and try to discover a grammar rule for myself.

___b) I prefer to read or listen to a grammar explanation first and then do some exercises.

4. In dealing with grammar rules…

___a) I like to learn rules of grammar indirectly by being exposed to examples of grammatical structures.

___b) I like to start with rules and theories rather than specific examples.

5. How important is it for you to study explicit grammar rules in order to understand how a language works?

___ Very important ___ Important ___ Useful, but not very important ___ Unimportant

6. When learning a new grammar point... ___a) I prefer to be given simplified examples that show clearly how the

114

grammar structure is used.

___b) I prefer to study authentic examples of the grammar as used by native speakers.

7. To what extent do you enjoy dealing with the ‘messiness’ of real language use?

___Very much ___Quite a lot ___To some extent ___Not very much ___Not at all

8. How easy is it for you to identify patterns in a foreign language and work out grammar rules from them?

___Very easy ___Easy ___Not sure ___Difficult ___Very difficult

Appendix 2

Grammar Project Pre-Test

Name_______________________________________________________________

115

Indicate which is the correct linking expression to fit each space by writing a, b, c or d on the lines provided.

1. For this job you don't need any previous work experience. Two qualifications are needed, ___________. a) although b) notwithstanding c) though d) even though

2. First and foremost is that almost every facility has an OHP projector ___________ not all have slide projectors .

a) despite b) whereas c) even so d) notwithstanding

3. Industrial sabotage has been a neglected area of research, ____________ its being one of the commonest forms of worker resistance .

a) although b) though c) even though d) despite

4. Teachers get irritated , of course they do , their elaborate and expensive training courses ____________. a) nonetheless b) notwithstanding c) even so d) nevertheless

5. Vapour from a heated pool in the centre of the floor obscured Folly 's vision with a steamy , exotically scented mist , but , _________, she could see that she was not alone . a) even so b) even though c) though d) notwithstanding

6. ___________ the Conservatives domination of the area , we can expect some exciting contests in this election and perhaps even some upsets . a) although b) whereas c) because d) in spite of

7. Reform of the company's computer system has yet to begin. ____________, practically everything else we set out to do has been achieved , including the next magazine and the annual report . a) despite b) nonetheless c) though d) even though

8. For some art historians, the Renaissance is a new beginning, ____________ medievalists can point out convincingly that no one event divides the Middle Ages from the Renaissance.

a) though b) even so c) despite d) nevertheless

9. ____________ you have left drama school your attention will be on the immediate problems of survival and making progress with your work.

a) afterwards b) while c) once d) following

10. An astigmatism has hampered her recovery. ____________ , today she’s having more laser treatment to improve the sight in her left eye.

a) despite b) notwithstanding c) though d) nevertheless

11. Gary Cockett, the NRA’s assistant district engineer for North Essex, said every time there was a high tide the River Hyde at Colchester flooded ____________ the sea wall was too low.

a) in spite of b) due to c) because d) as a result of

116

12. ____________ the ultimate prize had eluded King Edward, his prestige, and the military reputation of his countrymen, was at its height in 1360.

a) despite b) even though c) in spite of d) even so

13. The bank provides many services for their customers ____________ keeping their money safe.

a) besides b) although c) in addition d) nevertheless

14. The UK observatory at Lerwick is now operating very well ____________ continued developments in instrumentation, communications and data-processing software.

a) with the result that b) so as to c) for this reason d) as a result of

15. Football is like a religion in Scotland and, ____________ , it has a very strong social aspect attached to it.

a) as well as b) besides c) also d) too

16. I hadn’t seen her for 15 years. ____________, I recognised her immediately when I saw her. a) still b) whereas c) despite d) even though

17. ___________ arriving in London, Sally has had over 10 job interviews.

a) once b) afterwards c) since d) meanwhile

18. We used to call him ‘Skinny’ ____________ he was really fat.

a) even so b) despite c) nevetheless d) although

19 This is a binding contract. ______________, we recommend that you review it with a lawyer.

a) therefore b) as a result of c) due to d) although

20. _____________ Tuesday is a bank holiday (feriado), I suggest that we don’t have a lesson on Monday.

a) due to b) once c) for this reason d) since

Appendix 3: Inductive lesson guide: Printed Handout given to ‘Inductive’ learners

117

Online Grammar Project Inductive Approach

Welcome to the project. First, access the website: http://web.letras.up.pt/jlewis

IMPORTANT!!! Select Inductive mode

Follow the Set Up Instructions onscreen.

Go through the 11 online exercises following the onscreen instructions.

Note: Some exercises are to be done on the printed handout which you will be given.

Grammar focus - linkers

Exercise 1

Look at the 15 items in the box below. They are all linking expressions.

118

during therefore for this reason

so because as a result of although

afterwards despite meanwhile nevertheless

following due to once afterwards

Your task is to fill in the chart below with the linkers from the box so that the items in each row share a similar meaning and those in each column share the same grammatical function. Look at the row that has been filled in by way of example. All the linkers in Group A should, therefore, function like meanwhile, all those in Group B like while and all those in Group C like during.

IMPORTANT! Consult online Exercise 1 and use the Concordancer to study the grammar of each linker. Use the codes in the table onscreen.

Group A Group B Group C

time (progressive)

meanwhile while during

cause/reason

consequence

time (sequential)

concession*

* (concessive) linkers join one idea or fact to another which is opposed to it.

Exercise 2

Look again at the chart you completed in Exercise 1. Answer the following questions in order to understand more clearly how each group of linkers function.

IMPORTANT! Again use the Concordancer with the onscreen codes for Exercise 2.i.

119

1. Which group of linkers can usually be omitted without affecting the grammar/syntax of the sentence? Group C ٱ Group B ٱ Group A ٱ

2. Which group of linkers join together two parts of a sentence thus providing cohesion?Group C ٱ Group B ٱ Group A ٱ

3. Which group of linkers are most commonly followed by a finite subordinate clause?Group C ٱ Group B ٱ Group A ٱ

4. Which group of linkers commonly introduce nonfinite subordinate clauses (e.g. -ing clauses)?Group C ٱ Group B ٱ Group A ٱ

5. Which group of linkers are usually separated from the rest of the sentence by a comma or commas? Group C ٱ Group B ٱ Group A ٱ

6. Which group of linkers commonly join together an idea from one sentence or paragraph to the next?Group C ٱ Group B ٱ Group A ٱ

7. Which group of linkers are very often followed by a noun phrase?Group C ٱ Group B ٱ Group A ٱ

8. Which group of linkers are flexible in their position and can appear at the beginning, middle or end of a sentence?Group C ٱ Group B ٱ Group A ٱ

9. Which group of linkers are in fact a type of adverbial?Group C ٱ Group B ٱ Group A ٱ

10. What are the linkers in Group A called?Prepositions ٱ Conjunctions ٱ Conjuncts ٱ

11. What are the linkers in Group B called?Prepositions ٱ Conjunctions ٱ Conjuncts ٱ

12. What are the linkers in Group C called?

120

Prepositions ٱ Conjunctions ٱ Conjuncts ٱ

Now do online Ex.2.ii

Exercise 3

Some linkers can function in more than one way. Let’s have a brief look at three such examples: since, besides and though

Answer the following questions by looking the words up on the Concordancer with the onscreen codes for Exercise 3.i. in order to see how they function.

1. 'since' can function as both a...

conjunct and preposition ٱ conjunct and preposition ٱ conjunct and conjunction ٱ

2. Before a finite clause (indicating person and tense) the meaning of 'since' can refer to...time and cause ٱ cause ٱ time ٱ 

3. Before a (non-finite) -ing clause the meaning of 'since' can only refer to...cause ٱ time ٱ result ٱ4. 'besides' can function as both a...conjunct and preposition ٱ conjunct and preposition ٱ conjunct and conjunction ٱ

5. 'though' can function as both a...conjunct and preposition ٱ conjunct and preposition ٱ conjunct and conjunction ٱ

Now do online exercises 3.ii and 4

Exercise 5

We’re now going to look at some more linkers with a concessive meaning (i.e.linkers joining one idea or fact to another which is opposed to it).

whereas notwithstanding still nonetheless

even so even though in spite of

Put the linkers from the box above into the chart below. The chart contains the concessive linkers we have already looked at.

IMPORTANT!

121

In order to see how the linkers function, use the Concordancer with the onscreen codes for Exercise 5.

Group A

Conjuncts

Group B

Conjunctions (subordinators)

Group C

Prepositions

nevertheless

though

although

though

despite

Exercise 6

Let’s look more closely at though, notwithstanding and even so

IMPORTANT! Again use the Concordancer with the onscreen codes for Exercise 6.i.

1. When 'though' is used as a conjunct, it can be positioned...

At the beginning, middle or end of the sentence ٱIn the middle of the sentence, separated by commas ٱIn the middle and at the end of a sentence ٱ

2. 'though' is most commonly used at the end of a sentence in what kind of context?

Formal, written contexts ٱ Informal contexts, especially in speech ٱ Formal speech ٱ

3. Which structure is 'notwithstanding' linked to?

Noun phrase ٱ Nonfinite -ing clause ٱ Finite clause ٱ

4. 'notwithstanding' can be positioned...

at the beginning of a noun phrase ٱ at the end of a noun phrase ٱ at both the beginning and end of a noun phrase ٱ

122

5. When 'notwithstanding' comes after a noun phrase, the effect is...

formal ٱ extremely formal ٱ informal ٱ

6. 'even so' is often used after which word?

but ٱ and ٱ yet ٱ

Now do online exercises 6.ii, 7 and 8

Appendix 4: Printed Handout given to ‘deductive’ learners

Online Grammar Project - Deductive ApproachGrammar Focus   -    Linkers

There are various types of linking words and expressions which join ideas together and provide cohesion between separate sentences and paragraphs as well as within a single sentence.

The most common kind of linkers are:   conjuncts, conjunctions and prepositions

Conjuncts            

Examples of conjuncts are:   meanwhile, therefore, afterwards, nevertheless

Typically, these linkers join together an idea from a previous sentence (or paragraph) to another idea in the following sentence (or paragraph). 

They can express a variety of meanings. Here are just a few examples.    

time (progressive)

The council consistently fail to offer any leisure facilities for the city's youth. Meanwhile, vandalism and street fighting continue to plague the community. 

consequence:     

123

By seven o'clock it will be dark. Therefore, the last race will, , be at 6 o'clock .

time (subsequent)

At the meeting Jeff raised the issue of job-sharing. Afterwards, he telephoned Lizzie to ask what she thought .

concession (i.e. conceding that the last point was true but he next opposes it in some way)

The Samsung monitor and video card don't work brilliantly together, but, nevertheless, the screen is good .  Conjuncts...

are usually separated from the rest of the sentence by commas. can be omitted without affecting the syntax of the sentence. often appear in initial position, linking back to the previous sentence. can link back to a clause within the same sentence - see the example with

'nevertheless' above - and in this case are often preceded by a coordinator e.g. and or but.

are flexible in their position and can usually appear at the beginning, middle or end of a sentence.

are a type of adverbial - sometimes known as sentence adverbs

  Conjunctions   Broadly speaking there are two types of conjunctions; coordinating conjunctions and subordinating conjunctions. Coordinating conjunctions - or coordinators - join together two units of the same status e.g. two noun phrases:  I watched a film and a documentary        two adjectives:  He felt cold and hungry.       two clauses:  David went to the cinema but Cathy stayed at home.

The most common coordinators are and, but and or 

Subordinating conjunctions - or subordinators - introduce a subordinate clause and link it to a main clause in the same sentence.

Examples of subordinators are:  while, because, so, once, although

They are like the cement that glues two grammatical blocks together and are fundamental to the sentence structure. If removed the syntax would be broken.

As with conjuncts there is a wide range of subordinator meanings. e.g.

time (progressive) I noted him working on the machinery while we were waiting to ascend .

cause She was pardoned because she was an internationally famous actress.

124

consequence Acting is very much a doing thing , so the emphasis is always on practical work .

time (subsequent) Once you 've taken the drug , your next decision could be influenced by that drug .

concession (i.e. conceding that the last point was true but he next opposes it in some way) York does not have a permanent site for travellers although one is planned .

Note: the subordinator can appear in initial position in the sentence (when the main

clause follows the subordinate clause) and in such cases a comma usually appears at the end of the subordinate clause (see examples above)

it is more common for the subordinate clause to follow the main clause and in such cases a comma is usually not required. 

'so' is a special case because the consequence it introduces necessarily has to follow the main clause. It is usually preceded by a comma (see above)

  Prepositions

Prepositions are usually single words e.g. in, on, at, for, during, despite These are known as simple prepositions.

Complex prepostions are composed of two or more words e.g  due to, as a result of

Prepositions do not usually appear by themselves. They are almost always followed by an accompanying structure, most commonly a noun phrase e.g.  in the early morning    at my old school    for many years   since the beginning of May   This accompanying structure is known as a prepositional complement and together they make up a prepositional phrase.

Other structures can also act as prepositional complements including finite and nonfinite clauses e.g.

(1)  (a letter) about what he planned to do in the future (2)  On arriving home, he had a strong drink.

Prepositional phrases most commonly function as postmodifiers in a noun phrase - see (1) above or as adverbials - see (2) above. 

The main semantic role of a preposition is to express a relationship of meaning between its complement (usually a noun phrase) and another part of the sentence. e.g.

time (progressive)    She was disappointed that he hadn't arranged to meet her during the afternoon. 

125

cause    Sadly , due to refurbishment and a desire to create more shopping space , the kennels have been closed .

consequence   There has been an increase in reports of deaths as a result of torture in Turkish police stations .

time (subsequent)  Local authority children 's departments were established following the report of the Curtis Committee in 1946.   (Note: the meaning here is also consequential.) 

concession (i.e. conceding that the last point was true but he next opposes it in some way) Despite being the most powerful man in the world, there seemed to be nothing he could do . 

The table below summarises what we have studied and includes a sample of linker meanings. 

Conjuncts Conjunctions(subordinators)

Prepositions

time (progressive)meanwhile while during

cause- because due to

consequencetherefore so as a result of

time (subsequent)afterwards once following

concessionnevertheless although despite

Some words can function as more than one type of linker. Common examples include since, besides and though.

since functions as a preposition, expressing a  time relation e.g.

(1)  Since 6th April 1990 , married couples have been taxed independently

(2)  Since coming to this country, I've experienced a fair degree of culture shock.

126

since also functions as a conjunction, expressing a causal relation (3) and a time relation (4) e.g. 

(3)  Since almost all those on death row come from impoverished backgrounds they lack the funds to hire their own attorney and will be appointed counsel by the Court .

(4)  Since the new boss arrived, motivation has increased greatly. 

Note:  Since can only precede a non-finite -ing clause as a preposition - see (2) above.

 

besides functions as a preposition e.g. 

Besides its industrial uses , the metal has been readily adopted for tableware .

The public order legislation will mean that the police , besides being charged with the duty to keep the peace , will be enabled to mediate the already constrained ability to

engage in protest and dissent on a much wider front.

besides also functions as a conjunct e.g.

"How dare you suggest that I'm out of touch! Besides , it 's nothing to do with you."

 

though functions as a conjunction (a synonym of although) e.g.

Tone and colour are not synonymous , though the two terms are often wrongly interchanged .

Though she has had little education , her vocabulary is excellent

though also functions as a conjunct e.g.

(1) I'll be going to Italy next week. I'll be back for your birthday, though.

(2) Corpus linguistics is a difficult subject. It does, though, give us a fascinating insight into how language works.

Notice that as a conjunct though cannot appear in initial position. It often appears in end position - as in (1) above - in speech and informal

writing.

127

The chart below shows a selection of linkers, all with a concessive meaning.

Conjuncts Conjunctions

(subordinators)Prepositions

  nevertheless

  nonetheless

  even so

  still

  though

  although

  though

 even though

  whereas

despite

in spite of

notwithstanding

Look at the examples below:

concessive conjuncts:

It's a difficult race. Nevertheless, about 1,000 runners participate every year.

In the 19th century education was the preserve of the male dominated upper classes. There were , nonetheless , women who managed to become well-educated .

The lighting is considerably brighter than before but, even so , the Committee reported , is at the margins of acceptability for the production of pictures of broadcastable quality .

Diana was annoyed with Jim for disagreeing with her. Still , she was pleased he had sought her opinion .

Note:  even so often follows 'but'

concessive conjunctions

Although he's got a good job now, he still complains.

The remains have always been visible , even though the temple was erected some 2,500 years ago .

Whereas knowledge can be aquired from books, skills must be learned through practice.

128

concessive prepositions

Solicitors are increasingly becoming the dominant branch of the profession despite the historical recognition that they are the junior part of the legal profession .

In spite of being a trained lawyer, he was unable to give her any legal advice.

The employment contract made between employer and employee still forms the basis of modern employment law , notwithstanding the detailed legislation of recent years .

He is an extremely conceited , very lecherous , nearly alcoholic bore , his excellent work notwithstanding .

Note: notwithstanding is unique in that it can appear after its noun phrase (as in the last example above), but he effect is extremely formal.

Appendix 5

Grammar Project Post-Test

Name___________________________________________________________

Indicate which is the correct linking expression to fit each space by writing a, b, c or d on the lines provided.

1. None of those photos I took from the top of the hill came out. I did get some nice pictures of the castle, ____________ .  a) although b) notwithstanding c) though d) even though

2. ____________ wildlife conservation lends itself to a science-based approach, this is not really possible for landscape conservation.

a) despite b) whereas c) even so d) notwithstanding

3. ____________ his having been based in Southampton for many years, he had never lost his Lancashire accent .

a) although b) though c) even though d) despite

4. It’s taken a fortnight to get the oil out of my fingers, nightly doses of Ecover washing up liquid ____________ . 

a) nonetheless b) notwithstanding c) even so d) nevertheless

5. Two defenders had raced back from midfield but, ____________ , Parker slid the ball through them

129

and scored a magnificent goal. 

a) even so b) even though c) though d) notwithstanding

6. During Rita’s presentation the projector broke down, but she calmly carried on ____________ this rather inconvenient technical hitch.  a) although b) whereas c) because d) in spite of

7. In recent months staff shortages and discipline problems have plagued Banbury Secondary School. ____________ , parents have given the school a vote of confidence.  a) despite b) nonetheless c) though d) even though

8. Attempting to capture the spirit of a period is another planning strategy for a survey, ____________ this philosophical notion of a Zeitgeist is notoriously elusive.

a) though b) even so c) despite d) nevertheless

9. Many teachers make a place in their classes for work of this nature, since ____________ the benefits have been experienced, most class members enjoy it and want more.

a) afterwards b) while c) once d) following

10. As with so many of Stockman’s ambitions, this one did not come close to full realization. ____________ , in the early months of Reagan’s first term, the legislature was repeatedly upstaged and out-manoeuvred by the White House.

a) despite b) notwithstanding c) though d) nevertheless

11. One soldier was trapped for 20 minutes ____________ the impact of the crash forced the lorry onto his body.

a) in spite of b) due to c) because d) as a result of

12. He stayed out there shovelling snow, confounding the pessimists ____________ his task was clearly hopeless.

a) despite b) even though c) in spite of d) even so

13. Bacteria, ____________ being the simplest form of life we know, are also among the oldest fossils we have discovered.

a) besides b) although c) in addition d) nevertheless

14. A child may be said to acquire the rules of grammar, although it is difficult to see how such rules can be established ____________ numerous learning experiences.

a) with the result that b) so as to c) for this reason d) as a result of

15. He wants to score a goal urgently to show he’s still on top form, and, ____________, he has a point to prove – he wants to show United they were right to invest in him!

a) as well as b) besides c) also d) too

16. It rained constantly every day we were there. __________ , it was great just to get out of the city for a few days.

130

a) still b) whereas c) despite d) even though

17. ___________ leaving university, I have been doing voluntary work in the Sudan.

a) once b) afterwards c) since d) meanwhile

18. She could feel him trembling with fever and cold, ___________ he was close to the fire which burned merrily now .

a) even so b) despite c) nevertheless d) although

19. The difference between Peak and RMS switching is that , with Peak , the compressor monitors the true signal peaks which may be much higher than the average level. For drums , ___________ , Peak makes most sense .

a) therefore b) as a result of c) due to d) although

20. ___________ the session normally ends in July, a Bill which is not on its way by May or June may stand little chance of passing into law .

a) due to b) once c) for this reason d) since

Appendix 6

Post-Project Questionnaire

Name:___________________________________________

The following questions refer ONLY to the online unit you studied recently about the grammar of linking expressions. Thanks you for your cooperation.

1. How clear were the instructions?Very unclearٱ Rather unclearٱ Reasonably clearٱ Very clearٱ

2. How easy was the Concordancer tool to use?Very difficultٱ Difficultٱ Easyٱ Very easyٱ

3. How useful did you find the Concordancer tool as a means of studying grammar?Not at all usefulٱ Not very usefulٱ Quite usefulٱ Very usefulٱ

4. How easy was it for you to identify patterns of grammar and work out ‘rules’ using the Concordancer tool?Very difficultٱ Difficultٱ Easyٱ Very easyٱ

5. How useful did you find the online practice exercises?Not at all usefulٱ Not very usefulٱ Quite usefulٱ Very usefulٱ

6. How difficult were the exercises on the whole?Very easyٱ Quite easyٱ Quite difficultٱ Very difficultٱ

131

7. How easy was it to follow the ‘path’ of the online ‘lesson’?Very difficultٱ Difficultٱ Easyٱ Very easyٱ

8. Would you also have liked to have the grammar rules given to you in a more traditional and straight forward way?Noٱ Yesٱ

9. Would you like to study grammar in this way again in the future?Noٱ Yesٱ

132