International landscape of corporate governance

Get Complete Project Material File(s) Now! »

CHAPTER 3 Research Methodology and Analytical Framework

Introduction

The previous chapter dealt with theoretical frameworks. This chapter outlines the research methods and processes that were followed in the current study so as to achieve the stated aims. The chapter is organised into six sections: this introduction followed by 3.2 which is the restatement of research questions and aims, while 3.3 describes the research design selected to achieve the aims of the study, 3.4 presents the methods and procedures used to collect data, 3.5 provides a step by step outline of the processes followed in compiling the English-isiXhosa Parallel Corpus and lastly 3.6 offers an analysis of corpus data retrieved from English-isiXhosa Parallel Corpus for the design of a bilingual dictionary of financial terms.

Research aims and objectives

As indicated in Chapter 1 (section 1.3), the main aim of this study is to explore how parallel corpora can be used in the compilation of specialised bilingual dictionaries of English and isiXhosa. In order to achieve the main aim, the following objectives were identified in order to direct this chapter:
• To create an English-isiXhosa Parallel Corpus of financial texts
• To extract bilingual information from the parallel corpus
• To illustrate how information extracted from parallel corpora can be used to address communicative and cognitive lexicographic functions that will meet the needs of different users of specialised dictionaries.
The study explores the use of electronic corpora as the basis for the compilation of specialised bilingual dictionaries. In order to answer the above questions, a specialised parallel corpus of English and isiXhosa comprising financial texts was designed. Because a corpus cannot be fully exploited without a corpus analysis tool, ParaConc was selected and used to explore the English-isiXhosa Parallel Corpus. It, and its value for the study, were described earlier. The following section describes and justifies the type of research design selected for this study.

Research design

Selecting a research methodology and design is a decision-making process that is preceded by a research question and aims: ‘A research design is a plan for a study which provides the overall framework’ (Leedy, 1997:195). In other words, it is a well-organised plan that includes several activities such as methodologies, research sites, instruments and data collection procedures with the aim of answering the research question. The ultimate goal of a design is to provide research outcomes that are valid and credible. Creswell (2014:41) states that ‘research designs are types of inquiry within qualitative, quantitative, and mixed methods approaches that provide specific direction for procedures in research design’. A research that combines a qualitative and quantitative approach is called a mixed methods approach. In addressing the research aims re-stated in section 3.2, a corpus-based approach was selected as a suitable choice for this study because the analysis of the parallel corpus requires a corpus analysis tool. ParaConc, the software selected for this purpose, possesses featues that combine the qualitative and quantintave approaches. Because this study is using a corpus-based method, it is employing a mixed method. Corpus–based translation studies is categorised as quantitative research because it relies on computers and computer analysis software to draw word counts, frequencies and statistics. (Conrad, 1999:3-4) explains further that:
1) Corpus-based studies use a principled collection of naturally occurring texts, i.e. corpus
2) They use computers for analyses
3) Corpus-based studies include both quantitative analyses and functional interpretations of language use.
The statistics on the frequency and quantification of collocations display the strengths of the quantitative approach in the current research. In this study, it is important to state that corpus-based methods do not solely rely on corpus data; other methods could complement further actions. A quantitative approach may include exploratory research and descriptive statistics, as this study does. Zanettin (2013:31) declares, ‘quantitative and qualitative approaches are radically intertwined in corpus-based translation studies and they are not mutually exclusive’. In this study, the researcher will be using a computer analysis tool in analysing corpus results and applying qualitative analysis where applicable. That is, word counts, frequencies and other forms of statistical information will be presented and interpreted in line with the aims of the study.
The following sections outline the methods and procedures that were followed in collecting data, creation of the EngXhPC and the method used to analyse data.

 Data collection procedures

This section presents the preliminary processes that were followed in the collection of data that was used to create the EngXhPC of financial texts. The preparation of a corpus involves a number of decisions and considerations such as the text types, languages, period covered. The following processes are involved in preparing the design of a specialised corpus.

Identification and location of parallel texts

The first step taken by the researcher was to identify the sources of information that were relevant to the study. The researcher visited the websites of the Eastern Cape and the Western Cape Provinces with a view to identifying the translated financial documents in English and isiXhosa. The two provinces were selected because isiXhosa is predominantly spoken by 78,8% of the population in the Eastern Cape and by 24,7% in the Western Cape (Census 2011). The official languages of the Western Cape Province, for example, are English, isiXhosa and Afrikaans. The annual reports and budget documents published on the respective departmental websites for access by the public were identified, and these were available in the official languages of the Province in terms of the Provincial Language policies. The criteria used in the selection of these texts were:
• The type of financial information contained in the official written documents
• The diversity of texts translated by different translators
• The quality and reliability of translation in the target language, isiXhosa, and
• Accessibility in the electronic format.
The above selection criteria ensured that the texts would not be randomly selected, which would compromise reliability and authenticity of corpus data. After identifying the texts, the researcher was required to follow ethical procedures which are described in the following sections.

Ethical clearance, copyright and permissions

Research ethics focuses on what is morally proper and improper when engaged with participants or when accessing information or data. In consideration of ethical issues two important steps were taken, namely, application for ethical clearance and letters requesting permissions from copyright holders. In compliance with the UNISA policy on research ethics, an application for ethical clearance was submitted to the Research Review committee before undertaking the research. This was done in order to ensure that the ‘research is conducted with scholarly integrity, excellence, social responsibility and ethical behaviour’ (Unisa Research Policy). The application was approved and ethics approval was issued on 31 August 2015. The ethical clearance certificate is attached hereto as Appendix 1.
During data collection, written parallel texts consisting of original texts in English and their translations in isiXhosa were identified in order to create the EngXhPC. After the identification and reading of the texts on the relevant Websites, the researcher was obliged to consider ethical and copyright issues. Before retrieving the parallel texts from the websites and inclusion in the corpus, permission had to be granted by the authors. As noted, this was to ensure that the corpus was ‘in good legal health’ (Atkins and Rundell 2008:82). All the texts that were uploaded in EngXhPC are in line with research ethics and copyright procedures (cf. section 1.6.2). The copies of both request letters and permission letters are attached at the end of the dissertation as Appendixes 2 & 3 respectively.
Seeking permission from the publishers was not the straightforward process which had been envisaged. After letters were written to the various departments and institutions, it took time to obtain responses because some of the documents had already been made available on the Web. To seek permission for a document posted for public consumption did not make sense to some copyright holders; hence I had to make telephonic follow-ups to explain why the permission was being requested. On the Treasury website of the Western Cape, only the English source texts were posted; users were requested to write to the department if they wanted either Afrikaans or isiXhosa versions of the original text. This also took time as the Treasury personnel had to retrieve the requested texts and to ensure that they were quality assured before they were sent to the requester.
The data was subsequently collected from Provincial Departments and other institutions, namely, Financial Services Board (FSB) and Bona magazine, both located in the Gauteng Province, and Western Cape’s Language Committee and Treasury.

Type of documents and selection criteria

The collected texts were written electronic texts in English as the source language and isiXhosa as the target language. The fact that parallel texts were machine-readable made the capturing and text conversion much faster and easier. The types of texts that were selected fall under the following categories: (i) annual reports, (ii) annual financial statements, (iii) information brochures and articles on financial matters, (iv) legislation. The annual reports, budget documents were retrieved as full texts whilst text extracts were taken from terminology booklets or lists. The following is a list of texts that were collected and used to create the EngXhPC:
Annual Reports (AR)
1. Annual Report Western Cape Language Committee 2011/12
2. Provincial Treasury Western Cape
3. Annual Report 2013/14
4. Western Cape Language Committee Annual Report 2013/14
Annual financial statements (AFS)
1. Pan South African Language Board’s Financial Statements 2002
2. Provincial Treasury, Western Cape Financial Statements 2011-13
Legislation & Budget Documents
1. Western Cape Adjustments Appropriation Bill 2014
2. Adjustment and Appropriation Bill 2015
3. The Constitution of the Republic of South Africa (chapter 13)
Articles on finance from Bona Magazine (2016/17 publications)
1. Master your credit card: April 2017
2. Secret savings: February 2017
3. Invest in yourself: March 2017
4. Clothing account debt trap: January 2017
5. Bonus benefits: December 2016
6. How to prepare for unpaid maternity leave: October 2016
7. Jet-setting on a budget: May 2017
Brochures from Financial Services Board: December 2012
1. Protecting consumers- What is the purpose of the FAIS Act?
2. Collective Investment Scheme Brochure
3. Investment in shares on an exchange
4. The Pension Fund Surplus
5. What is long-term insurance?
6. The Supervision of South African Markets
7. Retirement Funds
8. Why should I have a will? (cf. www.mylifemymoney.co.za)
Government annual reports were presented in five sections/parts covering different topics such as general information, chairperson report, accounting officer’s, auditor-general’s report and audited annual financial statements. The diversity in the structure pointed to different themes, which has helped in showing how financial language is said and written differently in the same document. The variety of information covered in a single text provides a representative corpus of financial terms written and used in the official documents. Other documents listed above are publications from the Financial Services Board, a public entity, which is subject to the Public Finance Management Act No.1 of 1999. The brochures selected cover diverse topics in the financial sector, such as: insurance, pensions, wills, capital markets, investment schemes, etcetera. The aim of these brochures is to educate the public.
Bona is described as the only South African publication available in four languages: English, isiZulu, seSotho and isiXhosa (www.bona.co.za). Bona publishes issues on how to improve people’s lives, from their wardrobes to their finances. The focus of this study is on issues related to money matters. Therefore, the articles selected for inclusion in the parallel corpus inform and equip the readership on how to spend, manage and save money. These articles cover diverse topics and therefore contain terminology used in numerous specific areas of finance management. The common denominator in all these texts is that they are for public interest and can be classified as informative texts. The main function of such texts according to Nord (1997:37) is to inform the reader about objects and phenomena in the real world. The composition of the texts indicates the communicative role they play for the target audience, that of providing information on various aspects of a single field: the domain of finance.
The researcher selected these texts because they are authentic documents that are prepared by recognised institutions and posted on the official websites which are reliable. The translations are always performed by professional translators and quality controls are conducted before publication. As Bowker and Pearson (2002:54) assert, ‘texts that are produced by authors with proven credentials are likely to contain more authentic examples of LSP use than texts by authors who are not proven experts’. In other words, as mentioned, authorship has an influence on the reliability and authenticity of the whole study. The element of authenticity is further strengthened by the fact that the selected texts were not developed for the purposes of this research but are documents that were published for genuine communicative reasons.

Corpus size

One of the important issues to be taken into consideration when designing a corpus is the size. The issue of corpus size is controversial because in the corpus design literature there is no consensus on the minimum and maximum size. However, the size of a corpus is guided by the corpus typology and the aim of its design. The size of a specialised corpus cannot be equal to the size of a general-purpose corpus. A special purpose corpus generally is smaller because it is specific and deals with a special subject field, which is restricted. The EngXhPC under investigation, covers a restricted specialised language written and used in financial documents published by government and public institutions. Therefore, it is not the intention of the researcher to compile a large corpus due to scarcity of translated texts in the chosen subject-field. Many scholars attest that specialised corpora are usually smaller in scale than general language corpora because of their narrow focus.
The EngXhPC design was informed by two reasons: the availability of translated financial texts and the ultimate goal of the study. The translation of financial texts from English into indigenous South African languages is still a new language practice that is posing challenges. Bowker and Pearson (2002:45) caution that ‘you may find that you can get more useful information from a corpus that is small but well-designed than from one that is larger but is not customised to meet your needs’. The researcher made a careful selection of texts that contained the kind of terminology which might be useful in compiling a representative parallel corpus that suits the purposes of this study. The EngXhPC can be described as a small specialised parallel corpus with a total of 154 029 items (English 88 101 and isiXhosa 65 928) as illustrated in Figure 3.5 below.

Text types, period coverage and authorship

The types of texts that were selected for inclusion in a corpus are always determined by the type of corpus to be designed. The EngXhPC as a specialised parallel corpus was designed for purposes of extracting technical terminology for dictionary making purposes. Technical terms can be found in different genres such as legal, medical, science and technology, etcetera. The parallel corpus in this study focuses on the domain of finance. In other words, it is a single-field corpus restricted to one domain rather than a wide range of technical or specialised domains. The texts that were included in the EngXhPC cover a period between 2002 up to 2017 and were translated by a variety of translators contracted by different departments or organisations. The recent publications were selected in order to capture the latest terminology in the field under investigation. Language for specific purposes is very dynamic; therefore capturing old terminology would not be beneficial to users of specialised dictionaries.
The table below displays the characteristic features of the EngXhPC compiled and its suitability in being a source of lexicographic data required of a specialised bilingual dictionary of finance.
The EngXhPC can be described as a unidirectional parallel corpus of financial terms, which is contemporary. A well-constructed corpus is a recipe for successful analysis.

Challenges encountered during data collection

This section outlines the challenges that were encountered during the collection and capturing of corpus material. As discussed in section 3.4.2, requesting permission to use data for research purposes led to its own difficulties. Obtaining permission from financial institutions was problematic due to their red tape; as such, data from the private sector was not included in this corpus. One of the constraints during the selection of parallel texts was choosing ‘financial information or annual financial statements’ only to discover they had not been translated, because according to the publishers translating financial information is cumbersome and costly. Documents that had no translations were not captured since this is a parallel corpus.
Initially, the provinces that were selected to collect parallel texts were the Eastern Cape and the Western Cape because isiXhosa is one of the official languages spoken by the majority of citizens. However, when searching various websites, no Xhosa versions of the annual report or budget documents were posted on the Eastern Cape Provincial official websites, and this was confirmed telephonically. This limitation does not affect the corpus balance and representativeness. The three categories of texts retrieved for the creation of a parallel corpus focusing on financial terms provide adequate information to achieve the aims of the current study. The translations of these documents by different professional translators in various locations enhance the reliability of financial terminology. The subsequent section outlines the type of corpus data extracted from EngXhPC.

Compiling and analysing the English-ISIXhosa Parallel Corpus using ParaConc

The analysis of a parallel corpus is preceded by various steps ranging from text selection to corpus alignment which is a precondition of a parallel corpus. This section, therefore, describes the steps followed in the compilation of the said parallel corpus.

Pre-processing of corpus data

The parallel texts that were collected were in different formats, which necessitated text conversion and clean-up. Most texts were posted on the website in a pdf format, a format that is not editable. All the texts were then converted from pdf to MSWord documents in order to allow text normalisation. Text normalisation involves basic processes such as cleaning of data, correction of spelling mistakes, and removal of characters that emerged during conversion of texts from pdf to Word. The text conversion was automated by a Microsoft office pdf converter while the clean-up was done manually by the researcher. This process was the initial cleaning of documents in MSword. It should be noted that there were no changes made to the original content of the texts except the removal of tables, pictures and graphs, as these are not readable in the parallel corpus which is word based.
After the completion of these basic processes, the texts were then converted from Word to plain text (.txt). Besides the fact that, technologically, plain text uses a lesser storage space than rich text formats, all corpora according to Bowker (2003) are text-based resources, just like the EngXhPC. The plain text is also editable and the researcher was able to perform quality checks before the files were loaded onto the corpus. Cleaning of plain text files was done manually in order to eliminate ‘noise’ in the corpus. ‘Noise’ in a corpus simply means unnecessary information such as misprints, or strange characters that might affect corpus analysis. It should be noted that ParaConc does not permit cleaning up or correction of any mistakes, so all the errors should be removed from the plain text. Figure 3.1 and 3.2 below illustrates bilingual plain texts in the source and target language.
Figure 3.1 is an illustration of a plain text format in the source language. It was mentioned above that txt. format occupies a lesser space than a rich text. The format runs horizontally from left to the far right of the screen. Figure 3.2 below illustrates how the plain text of the target text, isiXhosa, looks after normalisation.

Loading of corpus files

To upload the .txt corpus files, the software program was opened by clicking on ParaConc.
After opening ParaConc, an initial screen with two menu items: FILE and INFO appeared (see Figure 3.3). The diagram shows a blank window. This is confirmed by the information field in the left lower corner of the window which reads:
‘No files loaded’.
Before loading the files, a file menu bar on the left window was double-clicked to open ParaConc; then a dialogue box appeared, enabling the selected parallel texts to be loaded. A load corpus command was selected amongst others as shown in Figure 3.4.
After a load corpus file command was shown, a dialogue box appeared to enable the selection of parallel texts. Before adding the files, the source language and the target language were selected.
The source language in the corpus under compilation is English; therefore, English was chosen while the target text was isiXhosa. When running the down drop box, isiXhosa was not listed; fortunately, ParaConc allows one to choose any language or font that can represent unlisted target languages. In this instance, French (Canadian) was selected to represent isiXhosa as shown in Figure 3.5 below.

TABLE OF CONTENTS
DECLARATION
ACKNOWLEDGEMENTS
1. OVERVIEW AND RATIONALE
1.1 Introduction
1.2 Background to the study (problem in context)
1.3 Problem Statement
1.4 Research objectives
1.5 Research Questions
1.6 The Hypothesis
1.7 Empirical investigation
1.8 The importance and contribution of the study
1.9 Structure of the dissertation
1.10 Conclusion
CHAPTER 2: THEORETICAL FOUNDATION
2.1 Introduction
2.2. Decision process
2.3 Board Process Studies
2.4 Board Context
2.5 Individual and collective inputs, outputs and outcomes
2.6 Group Dynamics
2.8 Board decision quality (BDQ)
2.9 Process Variables
2.10 Board activism
2.11 Board Independence
2.12 Effort norms
2.13 Use of knowledge/skills
2.14 Cognitive Conflict
2.15 Information Quality
2.16 Cohesiveness
2.17 Improving board process ad board effectiveness
2.18. Conclusion
CHAPTER 3: LITERATURE REVIEW
3.1 Introduction and Overview
3.2 Defining corporate governance
3.3 Overview of Governance Theories
3.4 International landscape of corporate governance: codes of practice
3.5 Evolution of thinking of governance
3.6 Governance Theories
3.7 Board structure
3.8 History of Corporate Governance
3.9 Importance of good corporate governance
3.10 Governance regimes and reforms in South Africa
3.11 The Governance Environment of South African Companies
3.12 International Corporate Governance Frameworks and Protocols
3.13 Governance practices and its impact on global ratings
3.14 The Public Sector South Africa: Schedule 1, 2 and 3
3.15 Board Failure
3.16 Board Functioning
3.17 Measures for the board
3.18 Financial Measures and non-financial measures of Performance
3.19 Chapter Summary
CHAPTER 4: RESEARCH METHODOLOGY – PHASE 1 & PHASE 2
4.1 Introduction
4.2 Assumption of the Research Paradigms
4.3 Research Design
4.4 Research Strategy
4.5. Study Sample
4.6 Unit of analysis
4.7 Data Collection Instruments
4.8 Instrument Pre-test
4.9 Data Analysis Process
4.10 Ethical considerations
4.11 Quantitative data analysis
4.12 Validity and Reliability Issues
CHAPTER 5
5.1Introduction
5.2 Qualitative research: results, discussion and interpretation of findings
5.3 Results, discussion and interpretation of quantitative findings
5.4 Results of Regression Analyses and Tests
5.5 Interpretation of the findings
5.7 Factor Analysis
5.9 CONCLUDING SUMMARY
6.1 INTRODUCTION
6.2 CONCLUSIONS ABOUT HYPOTHESES
6.3 CONCLUSION ABOUT THE RESEARCH QUESTIONS
6.4 IMPLICATIONS FOR PRACTICE – RECOMMENDATIONS
6.5 LIMITATIONS OF THE RESEARCH
5.5 Limitations and recommendations for future studies
6.6 Chapter summary
REFERENCE LIST
GET THE COMPLETE PROJECT