Research Methodologies in Translation
Studies
Gabriela Saldanha and Sharon O’Brien
First published 2013 by St Jerome Publishing
Published 2014 by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
711 Third Avenue, New York, NY, 10017, USA
Routledge is an imprint of the Taylor & Francis Group, an informa business
Copyright © Gabriela Saldanha and Sharon O’Brien 2013
Notices
Knowledge and best practice in this field are constantly changing. As new research and
experience broaden our understanding, changes in research methods, professional practices, or
medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described herein.
In using such information or methods they should be mindful of their own safety and the
safety of others, including parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors,
assume any liability for any injury and/or damage to persons or property as a matter of
products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions, or ideas contained in the material herein.
British Library Cataloguing in Publication Data
A catalogue record of this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
Saldanha, Gabriela, author.
Research methodologies in translation studies / Gabriela Saldanha and Sharon O’Brien.
pages cm
Includes bibliographical references and index.
ISBN 978-1-909485-00-6 (pbk : alk. paper)
1. Translating and interpreting--Research--Methodology. I. O’Brien, Sharon, 1969- author. II.
Title.
P306.S244 2013
418’.02--dc23
2013030989
ISBN: 978-1-909485-00-6 (pbk)
Delta Typesetters, Cairo, Egypt
For Fionn, Rebeca, Rián, Martina
This page intentionally left blank
Research Methodologies in Translation Studies
As an interdisciplinary area of research, translation studies attracts students and
scholars with a wide range of backgrounds, who then need to face the challenge
of accounting for a complex object of enquiry that does not adapt itself well to
traditional methods in other fields of investigation. This book addresses the needs
of such scholars – whether they are students doing research at postgraduate
level or more experienced researchers who want to familiarize themselves with
methods outside their current field of expertise. The book promotes a discerning and critical approach to scholarly investigation by providing the reader not
only with the know-how but also with new insights into how new questions can
be fruitfully explored through the coherent integration of different methods of
research. Understanding core principles of reliability, validity and ethics is essential for any researcher no matter what methodology they adopt, and a whole
chapter is therefore devoted to these issues.
While necessarily partial, the survey presented here focuses on methodologies that have been more frequently applied and therefore more thoroughly
tested. It is divided into four different chapters, according to whether the research
focuses on the translation product, the process of translation, the participants
involved or the context in which translation takes place. An introductory chapter
discusses issues of reliability, credibility, validity and ethics. The impact of our
research depends not only on its quality but also on successful dissemination,
and the final chapter therefore deals with what is also generally the final stage
of the research process: producing a research report.
Gabriela Saldanha is a Lecturer in Translation Studies at the Department of
English Language and Applied Linguistics, University of Birmingham, UK, where
she convenes both the distance and campus-based MA programmes in Translation Studies. Her research has focused on gender-related stylistic features in
translation and on translator style, using corpus linguistics as a methodology.
Her teaching focuses on translation theory, research methods and translation
technology. She is co-editor of the second, revised edition of the Routledge Encyclopedia of Translation Studies (2009). She is co-editor of Translation Studies
Abstracts and is on the editorial board of InTRAlinea.
Sharon O’Brien is a Senior Lecturer in Translation Studies at the School of Applied Language and Intercultural Studies, Dublin City University, Ireland, where
she teaches postgraduate and undergraduate courses in Translation Studies.
Her research has focused on translation technology, especially the post-editing
of machine translation output, translation processes, and controlled authoring
using keyboard logging, screen recording and eye tracking. Her teaching focuses
on translation technology, software localization, translation theory and research
methods. She is co-editor of St. Jerome’s Translation Practices Explained series
and a track editor for the journal Translation Spaces.
This page intentionally left blank
Contents
Acknowledgements
xiii
Chapter 1. Introduction
1.1 Motivation and Intended audience
1.2 Scope and limitations
1.3 Research model, structure and content of the book
1
2
5
Chapter 2. Principles and ethics in research
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
Introduction
Ontology and epistemology
Research terminology
Types of research
Research questions and hypotheses
The literature review
Data
Qualitative, quantitative and mixed-methods approaches
Research operationalization
2.9.1 Measurable variables
2.10 Research quality
2.10.1 Validity
2.10.2 Reliability
2.10.3 Generalizability
2.10.4 Qualitative methods, credibility and warrantability
2.11 Research ethics
2.11.1 Ethics in context
2.11.2 Ethics approval
2.11.3 Informed consent
2.11.4 Deception
2.11.5 Power relations
2.11.6 Protection from harm
2.11.7 Internet-mediated research
2.11.8 Plagiarism
2.12 Summary
10
10
12
14
16
19
20
22
23
25
27
27
35
36
38
41
42
43
43
45
45
46
47
48
49
Chapter 3. Product-oriented research
3.1 Introduction
3.2 A descriptive/explanatory approach to the analysis of language
50
50
3.3
3.4
3.5
3.6
3.7
3.2.1 Critical discourse analysis
3.2.2 Corpus linguistics
3.2.3 Strength and weaknesses of critical discourse analysis and
corpus linguistics
Designing studies in critical discourse analysis and corpus
linguistics
3.3.1 Corpus-driven, corpus-based, argument-centred and
problem-based designs
3.3.2 Selecting and delimiting texts as units of investigation
3.3.3 The need for comparative data
3.3.4 Corpus typology
Building corpora
3.4.1 Corpus design criteria
3.4.2 Annotation and alignment
Analyzing texts and corpora
3.5.1 The linguistic toolkit
3.5.2 Fairclough’s relational approach to critical discourse
analysis
3.5.3 The tools of corpus analysis
3.5.4 Addressing issues of quality in critical discourse analysis and
corpus linguistics
Research on translation quality assessment – Introduction
3.6.1 Strengths and weaknesses
3.6.2 Design
3.6.3 Which QA model(s)?
3.6.4 Data collection
3.6.5 Analysis
Summary
51
55
57
61
61
64
66
67
70
71
76
80
80
83
85
92
95
96
97
100
105
107
108
Chapter 4. Process-oriented research
4.1 Introduction
4.1.1 Common topics
4.2 General translation process research issues
4.2.1 Design
4.2.2 Data elicitation
4.2.3 Analysis
4.3 Introspection
4.3.1 Design
4.3.2 Data elicitation
4.3.3 Transcription
4.3.4 Analysis
4.4 Keystroke logging
4.4.1 Design
4.4.2 Data elicitation
4.4.3 Analysis
109
111
113
113
118
119
122
124
126
128
130
132
133
134
135
4.5 Eye tracking
4.5.1 Design
4.5.2 Data elicitation
4.5.3 Analysis
4.5.3.1 Analysis of temporal data
4.5.3.2 Analysis of attentional data
4.5.3.3 Analysis of data pertaining to cognitive effort
4.5.3.4 Analysis of linked data
4.6 Complementary methods
4.6.1 Contextual inquiry
4.6.2 Personality profiling
4.6.3 Physiological measurements
4.7 Summary
136
138
141
142
143
143
144
145
145
145
146
148
148
Chapter 5. Participant-oriented research
5.1 Introduction
5.2 Questionnaires
5.2.1 Overview
5.2.2 Strengths and weaknesses
5.3 Designing questionnaire surveys
5.3.1 Operationalization
5.3.2 Number and phrasing of questions
5.3.3. Open and closed questions
5.3.4 Likert scales
5.3.5 Pilot testing
5.3.6 Reliability and validity
5.3.7 Ethical considerations
5.4 Data collection using questionnaires
5.4.1 Sampling
5.4.2 Response rate
5.4.3 Internet-mediated collection methods
5.5 Interviews and focus groups
5.5.1 Overview
5.5.2 Strengths and weaknesses
5.6 Designing interviews and focus groups
5.6.1 Types of interviews and focus groups
5.6.2 Designing interview and focus group schedules
5.6.3 Language issues
5.6.4 Piloting
5.6.5 Ethical considerations
5.7 Eliciting data using interviews and focus groups
5.7.1 Sampling and recruiting participants
5.7.2 Interviewing and moderating: Basic principles and
key challenges
150
151
151
152
153
153
154
157
157
158
159
161
163
164
165
166
168
168
169
171
172
174
177
178
179
180
180
184
5.7.3 Face-to-face, telephone and Internet-mediated interviews
and focus groups
5.8 Analyzing qualitative data
5.9 Analyzing quantitative data
5.10 Data analysis in mixed methods research
5.11 Summary
186
188
194
201
204
Chapter 6: Context-oriented research: case studies
6.1
6.2
6.3
6.4
Introduction
Definition of case study
When to use case studies
Case study design
6.4.1. Types of case studies
6.4.2 Delimiting case studies
6.5 Collecting data
6.5.1 Written sources
6.5.2 Verbal reports
6.5.3 Observation
6.5.4 Physical artefacts
6.5.5 Quantitative data
6.5.6 Using a database to manage data
6.5.7 Ethical considerations
6.6 Analyzing case-study data
6.6.1 General principles
6.6.2 Practical suggestions
6.6.3 Computer-aided qualitative analysis
6.7 Summary
205
207
209
211
211
215
217
218
220
221
224
224
224
225
227
227
230
231
232
Chapter 7: Conclusion: The research report
7.1 Introduction
7.2 Structuring the report
7.3 Framing the report: introduction, literature review and
conclusion
7.4 Reporting methods
7.5 Reporting qualitative data
7.6 Reporting quantitative data
7.7 Reporting linguistic data
7.8 Summary
234
234
References
244
Index
270
236
237
240
241
242
242
List of Figures
Figure 2.1 Example of Research Terminology Applied to TS
Figure 2.2 A theoretical model expressing relationship between
time pressure (X axis) and translation quality (nominal scale
on the Y axis)
14
Figure 3.1 Design of a bidirectional parallel corpus
Figure 3.2 Sample header file from the Translational English Corpus
Figure 3.3 Example of tagged text taken from the British National
Corpus
Figure 3.4 Example of a parallel concordance obtained with
Paraconc
Figure 3.5 Fairclough’s relational model (adapted from Fairclough
2003:36)
Figure 3.6 A concordance of the node ‘source’ obtained using the
BYU_BNC interface for the British National Corpus
Figure 3.7 A sketch engine profile of the word ‘source’
68
77
Figure 4.1 Example of TAP transcription from the TransComp project
(http://gams.uni-graz.at/fedora/get/o:tc-095-201/bdef:
TEI/get), post-phase_2, participant: Professional AEF)
Source Text A1
Figure 4.2 Example of Translog linear data
Figure 4.3 Example of fixations during a reading task from a study by
Doherty and O’Brien (2012)
Figure 5.1 Example of bell curve in normal distribution
26
78
79
84
90
91
129
132
137
197
List of Tables
Table 2.1 Threats to validity identified by Frey et al. (1999)
30
Table 5.1 Examples of coding of semi-structured interview data
Table 5.2 Example of quartile data: from processing times
measurement in Guerberof (2008:38)
191
198
This page intentionally left blank
Acknowledgements
Many people assisted us in the writing of this book and we are enormously
grateful for that assistance, no matter how small or large the contribution. We
would especially like to thank Jenny Williams for providing us with insightful
and helpful feedback on a draft of this book. We drew on the expertise of many
others for advice and feedback on specific chapters and we express our sincere
gratitude to them: Fabio Alves, Andrew Chesterman, Claire Hewson, Dorothy
Kenny, Kaisa Koskinen, Ian Mason, Rebecca Tipton. Any errors are, of course, of
our own making. Finally, we are hugely grateful to our families for their patience
and support.
This page intentionally left blank
Chapter 1. Introduction
1.1
Motivation and Intended audience
Recent years have witnessed an increase in the number of translation training
programmes across the world, with a resulting explosion in the number of masters and doctoral students and, as reported in Mason (2009a), a concomitant
move towards explicit forms of research training in translation studies. The book
entitled The Map: A Beginner’s Guide to Doing Research in Translation Studies,
co-authored by Jenny Williams and Andrew Chesterman and published in 2002,
was given a very warm welcome by the translation studies community and is
still highly regarded by established and novice researchers alike. Clearly there was,
and still is, a thirst for a book that was specifically focused on research within the
domain of translation studies. Since the publication of The Map, there have been
some methodological developments in the field with, for example, the application
of methods such as keystroke logging, eye tracking, Internet-mediated research,
as well as an increased focus on sociological and ethnographic approaches to
research and on research ethics. We feel it is now time to build on the excellent
foundation set by Williams and Chesterman. The Map establishes the foundations of translation studies research and is particularly useful for those who
are starting to think about doing research in this area, and who need to decide
between different areas of research, theoretical models, types of research, and
so on. The focus of this book is on specific methodologies. We describe in detail
when and how to apply different methodologies and we provide examples from
translation studies research. There are, already, many excellent publications
that describe how these methodogies are applied in related domains such as
applied linguistics, social science, psychology and cultural studies. These books
are, of course, valuable to the translation researcher. However, it is our experience that even in related disciplines the books fail to answer all our questions
about doing research in translation. Often the examples feel distant or even irrelevant, thus failing to inspire translation studies researchers. We are convinced
that discussing methodologies within the translation studies context and offering examples of current best practice has a value above and beyond standard,
generic textbooks.
The Map is a beginner’s guide, as stated in the title, and is mostly directed
at PhD students. This book will also hopefully be useful to PhD, Masters and
Undergraduate students. Research students are expected to develop core research skills, such as understanding what counts as creativity, originality, and
the exercise of academic judgement. We have kept these needs in mind during
the writing process. However, we feel that a need exists beyond this readership
too. As discussed below, translation studies is interdisciplinary by nature. While
the professionalization of translation and the recognition of translation as an
academic discipline have resulted in translation-specific educational pathways
all the way from the undergraduate to the doctoral level, the field of translation
Gabriela Saldanha and Sharon O’Brien
studies continues to attract researchers from many different backgrounds who
may not be familiar with the wide range of methodological practices in the field.
By bringing together in one publication methodologies originating in different
disciplines and discussing how they can be fruitfully combined for the study of
translation we aim to contribute to the cross-fertilization of the different research
practices that inform translation studies.
1.2
Scope and limitations
Guba and Lincoln (2005:191) argue that “[m]ethodology is inevitably interwoven with and emerges from the nature of particular disciplines”. Linguistics and
literary criticism were for a long time the main source of theories and methods
in translation research, which was based on comparative text analysis carried
out with varying levels of linguistic or literary insight. Much of the research on
literary translation is still embedded within a comparative literature framework
and linguistic approaches are still widely used, although rarely with the same
narrow focus they initially adopted. During the 1980s, translation scholars began
to draw more heavily on methodologies borrowed from other disciplines, including psychology, communication theory, anthropology, philosophy and cultural
studies (Baker 1998:278). More recently, the importation of theories and models
from social theory has again widened the range of methodologies applied within
translation studies. In 1998, Baker suggested that:
Although some scholars see translation studies as interdisciplinary by
nature (Snell-Hornby 1988), this does not mean that the discipline is
not developing or cannot develop a coherent research methodology of
its own. Indeed, the various methodologies and theoretical frameworks
borrowed from different disciplines are increasingly being adapted and
reassessed to meet the specific needs of translation scholars. (Baker
1998:279)
The picture emerging from the current book is not of a single coherent methodology that could be described as specific to translation studies. However, there
is indeed evidence of adaptation and reassessment, which is perhaps most clear
in the dynamism with which different theoretical frameworks and methodologies
are being combined for the purpose of addressing translation studies’ concerns,
since none of the methodological apparatuses of related disciplines can, on their
own, fully account for translation phenomena (see, for example, the 2013 special
issue of Target on interdisciplinarity in translation process research).
In their overview of the main paradigms in contemporary qualitative research, Guba and Lincoln (2005:191) note that “[i]ndeed, the various paradigms
are beginning to ‘interbreed’ such that two theorists previously thought to be
in irreconcilable conflict may now appear, under a different theoretical rubric,
to be informing one another’s arguments”. We believe that translation studies
has now successfully moved beyond the paradigm conflicts of the 1990s (Baker
Introduction
1998) and has succeeded not only in celebrating and accepting a diversity of
approaches but in ‘interbreeding’ for the benefit of a more comprehensive and
nuanced account of the object of study.
In terms of this particular book, while we have made an effort to reflect the
interdisciplinarity of translation studies and have tried our best to represent
the different epistemological and ontological positions within the discipline,
our account is necessarily partial, reflecting our own academic backgrounds
and experiences. We have aimed to remain aware, insofar as our in-built biases
allow us, that our way of seeing and thinking about research methods may not
necessarily be in agreement with the way others see or think about them. In
what follows we justify our choices and acknowledge their limitations regarding
the contents of the book and the views they reflect.
Translation studies is interdisciplinary not only because it borrows from a
wide range of disciplines but also because it covers a wide range of practices.
While we have made an attempt to reflect this diversity in the examples we
have selected for discussion, there are areas of translation research that are not
adequately covered by the methodologies described here, such as interpreting
and translation history.
In her reflections on the periods of fragmentation experienced by translation studies while fighting to establish itself as an academic discipline in its own
right, Baker (1998:279) mentions the fact that theoretical models in translation
studies have tended to ignore interpreting and produced research that is of no
relevance to those interested in that field. While our impression is that there has
been progress in this regard, our experience is mainly within translation studies
and we may not be the best people to judge. We see interpreting studies at the
forefront of many of the methodological advances in the discipline in recent years,
and this view is reflected here in recurrent examples from interpreting studies,
particularly in the discussion of critical discourse analysis, interviews and focus
groups. However, we also acknowledge that the nature of interpreting as spoken
interaction presents certain challenges in terms of research methodology which
we are not in a position to discuss in detail.
The same could be said about translation history. The methodology described
in Chapter 6, case studies, has often been used in historical translation research
and two of the examples used in that chapter deal with historical phenomena.
However, the specificities of researching the past are not the focus of the
chapter. It is worth noting that translation history is the one area of translation
studies research where there is a book-length publication devoted to questions
of methodology (Pym 1998).
One topic that has dominated the literature in translation studies in the past
few years is the question of centre and periphery, dominant and subservient,
Western and non-Western perspectives, and we feel it is important to reflect
on these matters in relation to our approach. A question we have often asked
ourselves while writing this book is: how ‘universal’ are the research methods
described here? Susam-Sarajeva (2002) helpfully rules out the use of the terms
‘Western/non-Western’. She argues that “[b]eing ‘non-Western’ has apparently
Gabriela Saldanha and Sharon O’Brien
become the only common denominator behind otherwise vastly different languages and cultures” (ibid.:193). Equally, “the same dichotomy renders ‘the
West’ more homogeneous than it actually is” (ibid.). She argues instead for the
terms ‘centre’ and ‘periphery’ but acknowledges that these are also problematic.
Susam-Sarajeva highlights the danger that those operating in ‘the periphery’ will
regard their own concepts and ways of thinking as inferior; they will be “’educated
away’ from their own culture and society” (ibid.:199). This, she says, is inevitable,
because for research to be considered ‘useful’, ‘publishable’ and ‘quotable’ it
must refer to the established (central) frameworks. In order to be rated highly as
a researcher, one needs to publish in specific journals, most of which use English
as the language of publication and prioritize their own research agendas, with
their concomitant limitations in terms of research models and methodologies.
One of the authors of this book originates from South America and left behind
a country and a language to pursue an academic career. The other originates
from and lives in a former colony on the periphery of Europe. Therefore, these
issues are close to our hearts as individuals. Nevertheless, there is no denying
that our academic perspective is ‘central’ in Susam-Sarajeva’s terms, even if
this is by (de)fault rather than choice, since it reflects the environment in which
we have been immersed during our academic careers and the one in which we
operate more comfortably. We can think of many good reasons for academics
to start operating outside their comfort zones, but we take the view that a book
on research methodologies is not the best place to do that. However, in the
writing of this book we do not intend to present specific frameworks as the only
relevant ones and we hope that they are relevant as one way to do things no
matter where the research is being conducted or where the researcher or the
researched come from.
Our expertise is limited mainly to empirical research, which not only has
implications for the scope of the book, focusing on empirical methods and
methodologies, but probably also permeates the content in a more pervasive
and subtle manner in terms of our assumptions as to what constitutes good
academic practice, with its emphasis on evidence, hypotheses and operationalization. Despite acknowledging our limited focus, we have chosen not to call the
book ‘empirical research methodologies’ because we do not believe in a clear-cut
distinction between conceptual and empirical research. Good empirical research
needs to be based on conceptual research and conceptual research, to be useful,
needs to be supplemented by evidence. Evidence and theory are crucial to all researchers: “[y]ou need the ‘facts’ – imperfect though they may be; and you need
to be able to understand or explain them (theory)” (Gillham 2000:12). Although
we generally talk about theories as the basis on which we build our empirical
studies, we should not forget that theory is also what researchers create, the way
they account for the data, particularly in inductive approaches to research (see
Chapter 2). In other words, research can be seen as theory building as well as
theory testing; as providing answers (for example, in hypothesis-testing research)
as well as framing questions (in hypothesis-generating research).
A further clarification to be made in relation to our understanding of empirical
Introduction
research is that we do not believe that empirical research is necessarily descriptive or incompatible with critical-interpretive approaches. There has been a
tendency in translation studies to equate empiricism with descriptivism and the
latter with a-historical and uncritical methods that aim to produce generalizations about patterns of translational behaviour with predictive power (Crisafulli
2002). While there is a need for non-prescriptive research that establishes what
translators normally do and why, as opposed to telling translators what to do, we
also agree with Crisafulli that this does not mean that description must, or can,
be non-evaluative: “value judgements influence the selection of data as well as
the descriptive categories of analysis and the explanatory theories into which
these are organized” (2002:32).
When describing a phenomenon we inevitably foreground certain relationships at the expense of others and thus prioritize certain explanations over others.
For example, in a corpus-based study of explicitation, the design of the corpus
(whether it is comparable or parallel, whether it includes translations from more
than one language or from more than one translator) will necessarily limit the
otherwise extremely wide range of factors that could be seen as having an impact
on the frequency of instances of explicitation to be found in the corpus. A selfreflective approach to research should acknowledge this inherent bias while at
the same time highlighting the benefits of exploring certain variables in depth at
the expense of excluding others. It should also look for potentially contradictory
evidence as well as seek to back up any results with relevant data from other
sources. We consider methodological triangulation to be the backbone of solid,
high quality research and so it is implicitly suggested throughout each chapter.
1.3
Research model, structure and content of the book
Empirical research involves gathering observations (in naturalistic or experimental settings) about the world of our experience. Generally, the choice of what
aspect to observe will impose certain restrictions in terms of the methods we
use. Therefore, we have chosen to divide the chapters of this book according
to the focus of our observations: the texts that are the product of translation
(Chapter 3), the translation process (Chapter 4), the participants involved in
that process (Chapter 5) and the context in which translations are produced and
received (Chapter 6). It is important to stress, however, that (1) whether a piece of
research is process-, product-, participant- or context-oriented is not determined
by the methodology itself or even the source of data but by the ultimate aims of
the researcher, and (2) when investigating any of these aspects of translation it
is impossible to exclude from view all the others; there is inevitable overlap.
We are aware that, in adopting this division of translation phenomena, we
are offering the outline of yet another model of translation studies research,
rather than drawing on those already proposed by, for example, Marco (2009)
or Chesterman (2000). Our model of translation research is by no means flawless or complete; it reflects the perspectives from which translation has been
viewed rather than those from which we could possibly view it. In what follows
Gabriela Saldanha and Sharon O’Brien
we explain how this model compares to Chesterman’s and Marco’s.
Chesterman distinguishes three types of models: comparative models, which
aim to discover language-pair translation rules, language-system contrasts, or translation product universals (also known as features of translation); process models,
which represent change (from state A to state B) over a time interval (although
the process is not necessarily linear) and allow us to understand decision-making
in translation and cognitive factors influencing this process; and causal models,
which aim to explain why translations are the way they are by reference to three
dimensions of causation: the translator’s cognition (translator’s knowledge, attitude, identity, skills), the translation event (translator’s brief, payment, deadlines)
and the socio-cultural factors (ideology, censorship, cultural traditions, audience).
Chesterman (2000:21) argues that the causal model is “the richest and most
powerful” because it contains the other two models – linguistic and cognitive
factors are taken as causal conditions, and effects at the linguistic and cognitive
levels are recorded – and it encourages explanatory and predictive hypotheses.
On a superficial level, we could say that the product-oriented methodologies
described in Chapter 3 correspond to Chesterman’s comparative model; the
process-oriented methodologies described in Chapter 4 to the cognitive one;
the context-oriented methodologies described in Chapter 6 to the causal one;
and participant-oriented methodologies might be mapped either onto a cognitive or causal model according to the precise focus of the research. However,
as explained above, an underlying assumption of our approach is that there
cannot be purely descriptive (comparative or procedural) research, because
any (good) research necessarily takes into account possible explanations, and
descriptions are never neutral. Therefore, another way of mapping our model
onto Chesterman’s three types would be to classify it in its entirety as a causal
model that recognizes three different dimensions of causality (linguistic, cognitive and contextual).
A rather complex issue that is necessarily brought to the fore by this mapping
of models and which cannot be addressed in much detail here is the potentially
different ways of understanding causality in the two approaches (Chesterman’s,
as outlined in his 2000 publication, and ours). Our understanding is very broad:
we simply suggest that empirical research needs to address questions of ‘why’
at some point in the research process. Sometimes explanations remain at the
level of speculation but the research should at least point out potential avenues
for further research which could explain the results, and these suggestions need
to be grounded in the evidence and in the state of the art in the field. Koskinen
(2010) suggests that Chesterman (2000) first adopts a Hempellian understanding
of causality, according to which “[c]ausality is … a (probable) relation between
particular premises and observable phenomena” (Koskinen 2010:166), and then
repositions himself in a later article (Chesterman 2008a) where he supports an
‘agency theory of causation’ (Koskinen 2010:179, emphasis in original). Compared
to his earlier approach, Chesterman (2008a) favours a less limited understanding
of causes and follows a teleological model based on the notion of intentionally
making something happen. An agency theory of causation offers a wider range
of avenues for research; apart from probabilistic norms and laws, it considers
Introduction
goals, intentions, motivations, and the ethics of action (Koskinen 2010:179).
Koskinen suggests an alternative way of studying causality which is particularly
useful for case studies and which – instead of attempting to establish correlations or causal laws – focuses on causal mechanisms, i.e. on explaining “how a
particular factor can instigate a change in another factor” (2010:181, emphasis
in original). She describes this approach as a “a more down-to-earth attempt to
identify a plausible account of the sequence of events, conditions or processes
linking the explanans and the explanandum” (ibid.:182). In her own work, Koskinen adopts a nexus model, which “is based on placing the object of study … at
the centre of our attention and then trying to establish the kinds of relations it
enters into and how these relations interact with it and each other” (ibid.:180).
Chesterman (private communication) believes this model should be incorporated
into the typology described in Chesterman (2000). The nexus model is particularly
suited to case studies, a research methodology that allows us to focus on causal
mechanisms rather than causal effects (see Chapter 6). We revisit the difference
between mechanisms and effects in Chapter 2, Section 2.10.3.
Marco (2009) proposes four (non-exhaustive) models of research in TS:
(1) textual-descriptivist, (2) cognitive, (3) culturalist and (4) sociological. Marco’s
classification also overlaps to some extent with the one proposed in this book; our
product-oriented methods are also text-oriented, our process-oriented methods
have a strong focus on cognitive processes, participant-oriented research tends
to be sociological in nature and there is also overlap between cultural and contextual research. The key difference between Marco’s models and ours is that his
establish a closer link between research methods and theoretical approaches or
schools of thought. We prefer to encourage a looser connection between methods and schools of thought so as to offer flexibility in terms of what researchers
take and discard from each methodology and from each school, and encourage
creativity in terms of combining methods and theories.
While the book discusses both methods and methodologies, we decided to
highlight the latter in the title as the more encompassing term and because most
of the chapters discuss methodologies rather than methods (see Chapter 2,
Section 2.3 for a definition of these terms). Every piece of research begins
with theoretical assumptions, for example, about what science is and how
knowledge is constructed. Our choice of methodology depends on those assumptions as well as on our research question and/or hypothesis. The success
of our methodology in addressing the research question(s) depends on how
well the methods suit the research question(s) and the aim of the research.
These are questions of validity and reliability which are at the basis of empirical
research. Understanding such core principles is essential for any researcher
no matter what methodology they adopt, and these are therefore discussed
before we actually delve into methodologies per se, in Chapter 2. This chapter
lays the foundations for high quality research independently of the research
methodology adopted and the aspect of translation we focus on. It also discusses general ethical issues, which are then followed up in other chapters as
appropriate according to the specific characteristics of the methodology.
Gabriela Saldanha and Sharon O’Brien
Chapter 3 discusses how critical discourse analysis and corpus linguistics
can be used to examine translated texts (including transcripts of interpretedmediated events). As explained in that chapter, critical discourse analysis is not
actually a methodology but a school of thought that follows a series of principles
in its understanding of language and its approach to language research. Here, we
focus on the text-oriented methodology developed by Fairclough (2003, 2010)
and how it has been adopted and applied in translation studies. While some
linguists would argue that corpus linguistics is a research paradigm in its own
right (Laviosa 2002), within the context of the present book it is presented as a
methodology that can be used to pursue a wide range of research aims. Corpus
linguistics and critical discourse analysis are presented as both alternative and
complementary methodologies; in other words, they can be used on their own
or combined. Therefore, while we take care to clearly distinguish the principles
underlying each methodology, their strengths and weaknesses, and to discuss
their distinctive tools and procedures, much of the discussion in that chapter concerning general principles of linguistic research applies to both methodologies.
No book on research methodologies in translation studies would be complete
without considering the complex nature of research involving translation quality.
Such research tends to be primarily product-oriented (though alternative approaches are, of course, possible), and thus Chapter 3 also includes a discussion
of research involving translation quality assessment.
Chapter 4 introduces process-oriented research. We outline what the main
objects of inquiry have been to date in translation process research and discuss
general issues of research design, data elicitation and analysis, before focusing
specifically on four methods: verbal reports, keystroke logging, screen recording
and eye tracking. Chapter 5 focuses on the ‘participants’ (also called ‘agents’)
involved in the process of translations, such as translators, trainers, students,
commissioners and agents. This chapter discusses both quantitative and qualitative approaches to participant-oriented research and is divided into two main
parts: the first discusses questionnaires and the analysis of quantitative data
derived from them, and the second discusses interviews and focus groups and
the analysis of qualitative data.
The focus in Chapter 6 is on external – political, economic, social and
ideological – factors affecting individual translators, the circumstances in which
translations take place and how translations impact the receiving culture. The
object of enquiry is much broader than in previous chapters and a wide range
of methodologies could be used in the investigation of the very different contextual factors that can potentially be accounted for. We have chosen to focus
on the case study for two reasons: first, because of its flexibility in terms of
drawing from a wide range of sources of data, and second, because the label
‘case study’ is often used in translation studies research without consideration
of the particular characteristics and requirements of case study as a methodology (Susam-Sarajeva 2009).
While each of the chapters focuses on different research objects and methodologies, we have attempted – as far as possible – to adopt a similar structure
Introduction
in each of them so as to cover consistently what we see as key aspects of any
methodology: research design, data collection and/or elicitation, and analysis.
In empirical social research a distinction is made between “elicitation and evaluation methods: between ways of collecting data and procedures that have been
developed for the analysis of observed data” (Titscher et al. 2000:6). All chapters
make a distinction between these two stages, but it is important to note that in
some cases these are not two necessarily subsequent stages in a linear progression. This is particularly the case when doing qualitative research that follows an
iterative process as opposed to a linear one.
While many researchers use elicitation and collection as two interchangeable
terms, we distinguish between the two where appropriate. Collection suggests
the recording of data that already exist whereas elicitation evokes a more active
generation of data which are then collected or recorded. Elicitation also suggests
that a stimulus is involved, such as a text that needs to be translated according to
specific instructions, and it is therefore particularly appropriate for the discussion
in Chapter 4 on process-oriented methods.
In our final chapter (Chapter 7) we deal with what is also generally the final
stage of the research process: producing a research report. Research can be
reported in a variety of formats, from conference presentations to PhD theses.
Here we focus on written reports. Since many of the issues around reporting
research span all our chapters and all methodologies, we discuss them at both
a general as well as specific levels in this chapter.
Chapter 2. Principles and ethics in research
2.1
Introduction
The aim of this chapter is to highlight issues that should be of concern to all
researchers and to place these in the context of translation studies research
by offering examples from that domain. We commence with a discussion on
ontology and epistemology, terminology and types of research. This discussion
is necessarily brief: our aim is to introduce the core terms and concepts here,
and we recommend that researchers turn to general works on research methodologies for fuller discussions of the issues that arise. We then turn our attention
to research questions, hypotheses and types of data before considering different methodological approaches (quantitative, qualitative and mixed). The last
section focuses on questions pertaining to research quality and ethics.
2.2
Ontology and epistemology
There are many books on research methodologies in the humanities and social
sciences which cover important philosophical questions such as How do we
know what we know? or What is the truth? Here we will summarize the main
philosophical questions, present the most important concepts and terms, and
explain their importance for research in translation studies.
It is far too easy to delve into a research project without first questioning
one’s own view of the world, and, especially, of knowledge acquisition and ‘truth’.
Having an appreciation for different ways of seeing the world will not only help
with the decision regarding what research approach to take, but will also help us
as researchers to question our own underlying assumptions, thereby hopefully
strengthening our research.
One of the core terms that should be understood prior to engaging in research is ontology. In social research, one way of defining ontology is as “the
way the social world is seen to be and what can be assumed about the nature
and reality of the social phenomena that make up the social world” (Matthews
and Ross 2010:23). A key related term is epistemology, which is “the theory of
knowledge and how we know things” (ibid.). Here, we follow Matthews and Ross
in distinguishing, in very broad terms, three different ways of seeing the social
world – objectivism, constructivism and realism – and three epistemological
positions linked to these ontological categories: positivism, interpretivism and
realism. These categories are somewhat convenient simplifications; in fact, there
are many more than three ontological and epistemological positions, and there
are also several versions of each of the positions we present here. However, analyzing these three approaches should be enough to give us an idea of the range
of perspectives that can be adopted and their implications. Further reading on
Principles and ethics in research
11
these questions is therefore recommended. Guba and Lincoln (2005:193) provide
a helpful table that identifies five different paradigms (positivism, postpositivism,
critical theory, constructivism, and participatory/cooperative) and basic beliefs
associated with them concerning, for example, ontology, epistemology, methodology, ethics, inquirer posture and quality criteria. This may be a good starting
point for considering the position researchers think they might be most comfortable with as this will probably influence the approach taken in research.
According to Matthews and Ross (2010:24-25), objectivism “asserts that the
social phenomena that make up our social world have an existence of their own
[…], apart from and independent of the social actors (humans) who are involved”.
This position derives from the approach adopted by natural scientists when they
investigate phenomena in nature and assume that the researchers’ relationship
to the phenomena they study is one of objective observation. Constructivism,
on the other hand, asserts that social phenomena “are only real in the sense
that they are constructed ideas which are continually being reviewed by those
involved in them [the social actors]” (ibid.:25). In other words, the meanings of
any social phenomenon are not inherent but are ascribed to it by social actors
(ibid.:28). Realism presents an intermediate position between objectivism and
constructivism: it accepts that social phenomena can have a reality that is separate from the social actors involved in it but also recognizes that there is another
dimension that relates to what we know about the social world as social beings.
This dimension includes “structures and mechanisms that trigger or affect the
social reality that can be observed” (ibid.:26).
As mentioned above, each of the ontological positions described is linked
to an epistemological position, that is, it entails some beliefs as to what counts
as knowledge and how knowledge can be obtained. The ontological position of
objectivism assumes a positivist epistemology, which asserts that social phenomena can be objectively researched, data about the social world can be collected
and measured, and the resulting observations must remain independent of the
researchers’ subjective understandings; that is to say, the researcher remains
independent and has no impact on the data. Positivism is often linked with
quantitative approaches to research and to empiricism, i.e. the collection of
observable evidence (see Chapter 1). However, in postpositivist research, empiricism and objectivism are treated as distinct positions; just because research is
‘empirical’ in nature does not mean that it is ‘objective’ (Tymoczko 2007:146).
In postpositivism it is held that observation and measurement are fallible, and
the participation and influence of the researcher are acknowledged. As Crisafulli
(2002:33) puts it,
empirical facts do not exist independently of the scholar’s viewpoint;
indeed, it is the scholar who creates the empirical facts of the analysis by
making observable (raw) data relevant to his/her perspective.
Interpretivism is linked to the ontological position of constructivism; it prioritizes
people’s subjective understandings and interpretations of social phenomena and
12
Gabriela Saldanha and Sharon O’Brien
is often linked with qualitative approaches to research, where the researchers
attempt to explore the social world from the point of view of the actors and reflect on their own subjective interpretations. Realism is both an ontological and
epistemological position. As an epistemological approach it claims that certain
social phenomena exist outside the human mind and can be objectively investigated using approaches similar to those in the natural sciences. In this respect,
realism agrees with positivism. However, it also recognizes the existence of invisible but powerful structures and mechanisms that cannot be directly observable
but whose effects are apparent, and these effects can provide evidence of the
underlying structures and mechanisms (Matthews and Ross 2010:29); Realist
approaches to research might typically adopt both quantitative and qualitative
tools and methods.
We will not prescribe a specific ontological or epistemological framework
here. In fact, the approaches outlined are not necessarily mutually exclusive,
and we consider the way in which one standpoint or the other has divided
researchers in the past to be unhelpful. As Guba and Lincoln state, “there is
no single ‘truth’ … all truths are but partial truths” (2005:212). However, as
researchers bring a number of beliefs, prior knowledge and experience to research, it is helpful to reflect on these prior to commencing research.
2.3
Research terminology
Along with the terms used above, many other terms are used in research in
a way that assumes general agreement about the meaning assigned to those
terms. However, even seasoned researchers can use research terminology inconsistently, and this can lead to much confusion and frustration on the part of the
reader and, especially, the novice researcher. Terms such as model, framework,
theory, typology, concept, method and methodology often go unexplained or
are used synonymously, resulting in a lack of comprehension. Here we provide
some definitions for common terms, drawing mainly on Silverman (2006:13),
with the exception of the definitions for ‘framework’ and ‘typology’, which are
taken from Matthews and Ross (2010:34 & 112 respectively). Not everyone will
agree with these definitions, which represent but one way of defining the concepts. What is important for each researcher is to carefully consider their use of
research terminology, to justify the definitions used for their given purpose and
to use terms consistently, while being aware and drawing attention to the fact
that others might use the terms in a different way.
A model is a representation of the ‘reality’ of your research topic or domain.
In Chapter 1 we compared the model of translation studies research suggested
by this book with the models proposed by Chesterman (2000) and Marco (2009).
Note, however, that it is frequently the case that models are not made explicit
in research projects and that sometimes there can be a disconnect between the
assumed model and the object of investigation (Tymoczko 2007). A framework
is the set of ideas and approaches that can be used to view and gather knowledge about a particular domain. As described in Chapter 3, Halliday’s systemic
Principles and ethics in research
13
functional grammar is often used as an analytical framework in corpus-based
translation and critical discourse analysis research. A concept is an idea deriving
from a model or a framework. A theory organizes sets of concepts to define and
explain some phenomenon or, in Chesterman’s words, a theory is “an instrument
of understanding” (2007:1). A typology is a typical model of the way items tend
to be found in relation to each other. For example, one might try to construct a
typology of translation strategies used in specific circumstances. A methodology
is a general approach to studying a phenomenon whereas a method is a specific
research technique. In Sealy’s words, “methodology is the science of method”
(2010:61). Saukko differentiates between the two concepts in the following way
(2003:8; our emphasis):
whereas methods refer to practical ‘tools’ to make sense of empirical
reality, methodology refers to the wider package of both tools and a
philosophical and political commitment that come with a particular
research approach.
The relation between a theory and a method is expressed by Chesterman in the
following way: “methods
methods are the ways in which one actually uses, develops, applies and tests a theory in order to reach the understanding it offers” (2007:1).
Methods and tools are also frequently confused. Perhaps the best way to demonstrate how these terms might be applied to a domain within translation studies
would be to take the example of translation process research. In this example, our
model might be a particular model of cognitive processing, i.e. a representation
of the phenomenon whereby the brain perceives signals, processes them and
converts them into meaning and instructions. Our framework might be cognitive
load, i.e. a set of ideas about brain processing capacity during a specific task,
where we expect that there is a limit to the amount of information (signals) that
can be processed by the human brain at any moment in time. Concepts within
that framework might include the translation process, i.e. an activity the brain
engages in when a human is translating from one language to another, short-term
memory, long-term memory, limitations on the capacity of the brain, to name
just a few. Our theory might be the MiniMax theory (otherwise known as the
‘principle of least effort’ (Krings 1986a, Séguinot 1989, Lörscher 1991), which
posits that humans (translators in this case) opt for a strategy whereby they
employ the highest level of cognitive processing possible and do not proceed
to a deeper level of processing, which entails a greater cognitive load, unless
the first level proves to be unsuccessful or unsatisfactory. Our methodology for
studying this might be empirical (we will gather evidence from translators at
work) and might combine both qualitative and quantitative methods such as
think-aloud protocol and keystroke logging. We may wish to propose a typology
of the translation strategies used to achieve the principle of least effort. Finally,
the tools we might use are a voice recording device, screen recording software
and a keystroke logging tool. We summarize this example in Figure 2.1.
14
Gabriela Saldanha and Sharon O’Brien
Model
Frarne\Nork
Theory
Methodology
•
X Model of cognitive processing
• Cognitive load analysis
• Concepts: translation process,
STM, LTM , restrained capacity ...
• MiniMax
• Empirical/Experimental
• Qua ntitative
Methods
- Eye Tracking
- Keyboard Logging
• Qualitative
- TAP
- Que s tionnaire
Tools
• Eye Tr acker (Hardvvare &
Softvvare )
• Keyboard L ogging Softvvare
• Dictaphone
• On-line Questionnaire
Figure 2.1 Example of Research Terminology Applied to TS
Laying out one’s research domain in these terms is not always an easy task, and
we expect that there will be some who do not agree with our categorization in
Figure 2.1, but the exercise is worthwhile because it forces the researcher not
only to think about the research terminology but also about the concepts one
is subscribing to, how to communicate these to the wider research community,
and, ultimately, what the researcher’s view of the world is.
2.4
Types of research
There are many questions to be answered before conducting research, such as
what is the research question, which method or methods are most appropriate,
what kind of data will be collected, how will the data be analysed and so on. We
have argued that it is worthwhile thinking about one’s epistemological framework before diving into such details. Likewise, we argue that it is important to
consider what type of research we are engaging in. An initial question pertaining
to type of research is what logical system it subscribes to, i.e. whether it is being
conducted from an inductive or a deductive positioning. Induction involves the
development of theories and hypotheses from the data collected (it moves
Principles and ethics in research
15
from particular instances to general statements), whereas deduction involves
the testing of existing theories or hypotheses through data (it moves from
general statements to specific instances). A third position, abduction, is also
possible. This position was first mentioned by C.S. Pierce in 1878; it proposes
to isolate the most convincing reasons (hypotheses) from a research result and
to research these hypotheses further. Johnson and Onwuegbuzie (2004:17)
helpfully characterize the three as discovery of patterns (induction), testing of
hypotheses (deduction) and seeking understanding by uncovering and relying
on “the best of a set of explanations for understanding one’s results”.
In addition to the question of logical positioning, there is the question of
the nature of the research. As explained in Chapter 1, this book focuses on
empirical research. Williams and Chesterman (2002:58) explain that empirical
research “seeks new data, new information derived from the observation of
data and from experimental work; it seeks evidence which supports or disconfirms hypotheses, or generates new ones”. This type of research is generally
seen in opposition to conceptual research, which “aims to define and clarify
concepts, to interpret or reinterpret new ideas, to relate concepts into larger
systems, to introduce new concepts or metaphors or frameworks that allow a
better understanding of the object of research” (ibid.). However, as discussed
in Chapter 1, the distinction is not always clear cut and these two types of
research are not mutually exclusive (see the discussion of argument-centred
research designs in Chapter 3). Empirical researchers can engage in either basic
or applied research. Although the distinction between these two types is not
clearcut either, basic research is generally understood to mean fundamental research, the primary aim of which is to acquire new knowledge. Applied research
is generally understood to mean research on practical problems, research that
has an application in life. Research may also be characterized as experimental,
in which case the researcher seeks to establish cause and effect relations (if X
happens, then what is the effect on Y?). Such research might be carried out in
a controlled environment, although this is not always practical in humanities
and social science research, and is often comparative; it compares two groups
and their properties of behaviour when certain variables are manipulated.
It may be designed in such a way that there is an ‘experimental group’ (also
known as a ‘treatment group’) and a ‘control group’. Members of the former
group are exposed to some sort of ‘treatment’, or manipulation, while the latter
are not. Note that the groups are not necessarily populated by humans, but
can also be composed of texts, for example. The creation of control groups in
translation studies research is not without challenges, however. Comparable
groups of translators or translations may simply not exist. To compensate for
this, Tymoczko (2002:21) suggests the use of other translations of the translated
text that is under investigation or even other passages from the translated text
that ‘are neutral with respect to the issues being investigated’, or the use of
a corpus of parallel texts (see the discussion on obtaining comparable textual
data in Chapter 3, Section 3.3.2). Experimental research in translation studies
is discussed in more detail in Chapter 4.
16
Gabriela Saldanha and Sharon O’Brien
Basic or applied research does not necessarily have to be experimental,
though, and might also be explorative. An example of explorative research is
phenomenology, where rather than asking what is the effect on Y if X happens,
or what X is, the lived experience or appearance of a particular phenomenon is
explored. Phenomenology is an interpretive, subjective approach to research,
which is interested in gaining insights from personal experiences. For further
discussion on phenomenology see, for example, O’Leary (2010) and Lewis and
Staehler (2010).
Research can also be evaluative, attempting to establish the value of a particular initiative once it has been implemented (summative evaluation) and the
intended or unintended effects of the initiative, or it might evaluate the delivery
of an initiative (formative or process evaluation).
The goal of research can extend beyond that of evaluation or looking for
relationships between X and Y; it can also lead to change, and this is where the
term action research is applied. Action research tackles “real-world problems in
participatory and collaborative ways in order to produce action and knowledge
in an integrated fashion through a cyclical process” (O’Leary 2010:146). Action
research is collaborative: it seeks to empower the stakeholders and moves away
from the concepts of the ‘researcher’ and the ‘researched’. See Chapter 5, Section 5.7.1, for an example of action research in interpreting.
Research might also be ethnographic, when it explores cultural groups “in a
bid to understand, describe, and interpret a way of life from the point of view of
its participants” (ibid.:116). One example is Koskinen’s (2008) study of the Finnish translation unit at the European Commission. The ethnographic approach is
discussed in detail in Chapter 6. For a more detailed discussion of ethnographic
research methods, see, for example, Madden (2010).
2.5
Research questions and hypotheses
Before we as researchers select methodologies, we must first identify at least a
tentative research question, and possibly several sub-questions, which are often
refined as the research develops. The sub-questions allow the researcher to unpack the main research question into more specific, highly focused questions.
Williams and Chesterman (2002) have identified many research domains in
translation studies. The research question may very well ‘belong’ to one of these
research domains but it may also straddle more than one domain, or explore
new domains. As translation studies expands its horizons, we can expect research
questions to touch on many more and varied topics than it has done to date.
There are different types of research questions (Matthews and Ross 2010:57).
A question might be explorative, in which case it seeks to find out what knowledge exists about a particular phenomenon. If we return to our previous example
of translation processes and the theory of a MiniMax strategy, an explorative
research question might be What evidence is there to show that the MiniMax
strategy is employed by translators? A descriptive research question seeks to
elicit data through which a phenomenon can be described in detail, e.g. What
Principles and ethics in research
17
micro-strategies do translators employ when they apply the MiniMax macrostrategy? An explanatory research question is a ‘why’ question. In our example,
this might be formulated as Why do translators employ the MiniMax strategy
while translating? The fourth type of question is an evaluative question which
seeks to understand the value of a phenomenon, e.g. What is the impact on
translation quality when translators employ the MiniMax strategy?
We stated in Chapter 1 that we do not believe in a clear-cut distinction
between descriptive and explanatory research, and it is important to stress
here again that research questions do not always fit neatly into one of the four
categories above. A researcher may have an explorative question, which is then
followed by a descriptive or an evaluative sub-question, for example. Questions might also have a hierarchy of sorts, with one being a primary research
question, followed by one or several secondary research questions. Indeed,
some primary research questions cannot be ‘operationalized’ until they are
broken down into more specific sub-questions (Sunderland 2009); see below
for a discussion of operationalization.
Researchers will select questions based on their interest in the topic, but
the question should also be one that is of interest to the community at large.
Unfortunately, questions worthy of future research are not always made
explicit in research publications, but it is still possible to extract questions by
identifying what has not been said by authors. This requires a critical reading
of research publications, where the reader considers what questions might
arise from the argument being put forward and whether or not they are addressed by the author(s).
It is generally accepted that research questions evolve over time. This is a
normal development in the research cycle: as we become more familiar with the
domain we are better able to critique our own research question and to refine
it; to do so is to be recommended. This refinement frequently involves reducing
the scope of the research question, or making it more specific, or introducing
some sub-questions which will allow us to investigate the general question in
more detail. Some research methods almost demand that questions evolve over a
period of time, by the very nature of the method itself. For example, ethnographic
research in general or case studies in particular might demand multiple cycles
where research questions evolve as the research takes shape (see Chapter 6).
Also, the use of abduction, as mentioned above, can contribute to the evolution
of research questions and hypotheses.
It might seem overly simplistic to say that a research question should be formulated as a question, but not doing so is a common mistake, especially among
novice researchers. When novice researchers are asked about their research
questions they often describe their topic instead, confusing the two and failing
to formulate a question (Sunderland 2009). Williams and Chesterman (2002:57)
reinforce the importance of the question type by recommending the selection
of the research model based on the type of question asked. Olohan and Baker
(2009:152) make an important point regarding the wording of research questions for doctoral studies: “Almost every word used in a research question sets
18
Gabriela Saldanha and Sharon O’Brien
up specific expectations, some of which a student may not be able or willing to
fulfill”, but this issue might also be relevant beyond Master’s or doctoral research.
Sunderland (2009) echoes this point and adds that the researcher needs to understand, and explain, exactly what each word in the question means.
According to Matthews and Ross, hypotheses are specific types of research
questions that are not phrased as questions but as statements about relationships; they define a hypothesis as “[a] testable assertion about a relationship or
relationships between two or more concepts” (2010:58, emphasis in original).
A research question, then, can sometimes be rephrased as a hypothesis. If we
take the descriptive research question mentioned above (What micro-strategies
do translators employ when they apply the MiniMax macro-strategy?), we might
express the following hypothesis in relation to this question: When translators
employ the MiniMax strategy, they make use of micro-strategies that are different from those they use when they are not employing the MiniMax strategy (but
see comments about the null hypothesis below). In other words, the researcher
is asserting that there is a relationship between the use of the MiniMax strategy
and the type of micro-strategies employed. Note that the hypothesis is not just
an expression of the research question in the form of a statement. We have
had to refine it somewhat in order to express it as a hypothesis, and it probably
still needs further refinement. It can be illuminating to ask oneself what one’s
hypotheses are, once the research question(s) has/have been formulated. In
doing so, we are asking what we expect to find and the research project should
aim to find evidence which either supports our hypotheses or contradicts them.
Note that even if our hypothesis is not supported (or fully supported), this is still
a valuable research outcome.
Not all research questions can be expressed in terms of a hypothesis. In fact,
those who disagree with the positivist approach to research would claim that
hypotheses are reductionist devices which constrain social research (O’Leary
2010:55). Olohan and Baker (2009), in their discussion of research training for
doctoral students in translation studies, comment that they favour open research
questions over hypotheses for several reasons, including, for example, that an
open research question provides broader scope for interrogating data from
several perspectives; they encourage students to keep an open mind about the
data and potential findings. Hypotheses are commonly (though not exclusively)
used in situations where data can be gathered to measure each concept and
where statistical tests can be executed to establish if there is a relationship
between concepts.
The relationships expressed in a hypothesis can be causal or associative
(Matthews and Ross 2010:59). Causal relationships assert that a change in
one concept (X) causes a change in the other (Y). Associative relationships
recognize the influence of one concept on another. In the latter case, there
might be a third factor, Z, an intervening variable (cf. Silverman 2006:289),
which influences Y.
Chesterman (2007) recognizes four types of hypotheses: descriptive, explanatory, predictive and interpretive. The first three can be grouped together
Principles and ethics in research
19
as empirical hypotheses, whereas the interpretive kind has a different status.
According to Chesterman, a descriptive hypothesis is formulated along the
lines of ‘All Xs have features F, or belong to class Y’; an explanatory hypothesis
states that ‘X is caused or made possible by Y’; and a predictive hypothesis
is formulated as ‘In conditions ABC, X will (tend to) occur’. The interpretive
hypothesis asks whether something (X) can be usefully interpreted as Y, or
better understood if we ‘see it’ as something else. Chesterman (2001a) notes
that classifications and categories (e.g. types of equivalence) are interpretive
hypotheses in themselves. Interpretive hypotheses pertain to conceptual (as
opposed to empirical) research and are “conjectures about what something
means” (Chesterman 2008b:49); they are “what we use when we try to understand meaningful yet obscure phenomena” (Chesterman 2008b:56). An
example of an interpretive hypothesis would be ‘translation is best conceptualized as a type of artistic performance rather than as a reproduction’. As
Chesterman notes (2000:23), interpretive hypotheses are rarely presented
explicitly as such, to be tested like any other hypotheses.
In research that adopts a mainly quantitative approach, it is traditionally
assumed that no relationship exists between two variables, and statistical tests
are based on this assumption (Rasinger 2008:176). The hypothesis mentioned
above would therefore be phrased as: When translators employ the MiniMax
strategy, they make use of the same micro-strategies they use when they are not
employing the MiniMax strategy. In other words, there is no relationship between
the MiniMax strategy and the type of micro-strategies used in translation. This
is called the null hypothesis and is given the notation Ho. We are usually interested in disproving the null hypothesis, in demonstrating that its opposite, or
the alternative hypothesis (H1) is true. We discuss the falsification of hypotheses
in more detail in Section 2.10.4.
2.6
The literature review
It was mentioned above that one way of identifying interesting research
questions is by performing a thorough literature review. The literature review
gives researchers an opportunity to explain their motivation and potential
contribution. According to Fink (2005:3), the literature review is “a systematic,
explicit, and reproducible method for identifying, evaluating, and synthesizing
the existing body of completed and recorded work produced by researchers,
scholars, and practitioners”.
Let us examine each of these qualifiers in turn: Systematic means that it is not
random, i.e. that all key sources of published research on the topic have been
identified, read and evaluated. Explicit implies that there is clarity regarding what
works, authors, time period, domain, languages, regions, etc. have been included
and, of equal importance, what has been excluded and why. Reproducible demands that everything is documented clearly, with appropriate referencing, so
that any other researcher could track down the sources used and confirm the
summary of the included works.
20
Gabriela Saldanha and Sharon O’Brien
The literature review identifies all relevant work and synthesizes core concepts and findings. Care is needed in the synthesizing task as this does not
simply involve repeating verbatim what other researchers have said, but rather
summarizing the main themes, ideas, questions/hypotheses and conclusions.
There are two significant challenges when synthesizing: avoiding plagiarism
and deciding how to structure one’s work. Plagiarism will be discussed below
in Section 2.11.8.
A number of questions could be asked when considering how to structure
the literature review: do you work author by author, era by era, language by
language, etc., or do you amalgamate into common themes and topics? Opinions
will vary on this, but generally speaking a literature review that is structured
along thematic lines might be more effective and accessible than one structured
chronologically and/or according to author.
Arguably one of the most important features of a literature review is that it
evaluates critically. For a novice researcher, who is perhaps new to doing research
and to the topic itself, this is one of the most challenging aspects of the literature
review. Assuming a position of modesty, the novice researcher might think that
they are not in a position to criticize an author who has published one or several
papers or books. However, it is the job of the researcher to critically explore
previous research. We should aim to identify both strengths and weaknesses in
earlier work, concepts that have not been fully investigated or researched at all,
concepts that have been researched particularly well or that have been overresearched, weaknesses in assumptions, methods, research questions, and so
on. There is an important comparative aspect to this commentary too: we should
aim to highlight contradictory findings as well as findings which support those
of previous research, and we should aim to identify differences in assumptions,
theories and definitions and how these can lead to different conclusions.
The literature review is an important vehicle through which researchers can
identify and describe the most relevant theoretical framework(s) for their own
research. Tymoczko (2007) directs attention to the interrelationship between
data and theory, emphasizing that in postpositivist approaches to research the
recognition of the interdependence between data and theory is essential. In the
analysis of data, researchers have an opportunity to explore this interrelationship
and to make explicit links to the theoretical framework(s) they have identified
as being important.
2.7
Data
To find answers to research questions, we need to collect appropriate data for
analysis. Data can be spoken or written, non-verbal, structured in different ways,
produced by individuals or groups, be factual or representing opinions, and it can
include the researcher’s own reflections (Matthews and Ross 2010:181).
Methods for data collection and analysis will be discussed in more detail in
relation to the different methodologies presented in Chapters 3 to 6. For the mo-
Principles and ethics in research
21
ment, we need to differentiate between primary and secondary data. Primary
data are collected by the researcher him or herself while the term secondary
data refers to collections of data, e.g. interview transcriptions, questionnaire
responses, translations etc., that have been collected by other researchers
and made available to the research community for analysis. Corpora could be
considered in this category, so an example of secondary data for translation
research would be the Translational English Corpus, a computerised collection
of contemporary translational English text held at the Centre for Translation and
Intercultural Studies at the University of Manchester.1 A researcher interested in
analyzing some aspect of translated English could use this resource as secondary
data while also creating their own corpus. Primary and secondary data might be
structured in different ways. When comparing primary and secondary data, it is
important to take into account that the circumstances under which data were
collected, and the number and nature of the people who generated the data
and the time of data collection or elicitation might vary and this may affect the
comparability of the two data sets.
The type of data collected is also important because it will determine whether
we use qualitative and/or quantitative approaches in our research. Quantitative
approaches will generate structured data which can be represented numerically
and analyzed statistically, whereas the qualitative approach will generate semior unstructured data. In questionnaire surveys, for example, structured data are
generated by asking the same questions to all research participants and limiting
the ways in which they can provide answers (through tick boxes in questionnaires, for example). Qualitative interviews, on the other hand, generally result
in semi- or unstructured data because the questions asked vary to some degree,
the respondents are given some or a lot of freedom when answering and not all
questions are necessarily answered. See also the discussion of quantitative and
qualitative approaches to textual analysis in Chapter 3.
For data that can be quantified, it is also important to take into account what
kind of measurements we can apply. Rasinger (2008:25-26) distinguishes four
different levels of measurement. The first level pertains to categorical scale data
(also termed nominal data), where data can fall into only one category, such as
‘pregnant’/’not pregnant’. The second level is ordinal scale data, where a concept
can be ranked, but where it is not possible to measure differences between each
label. The example given by Rasinger here is for ranking of university lectures on a
dullness scale, where it is impossible to say that ‘very dull’ is twice as dull as ‘dull’.
The next level is interval scale data, where again categories can be labelled, but
the difference between them is fixed. A typical example mentioned by Rasinger
is the grading system used to evaluate student work. The final level is ratio scale
data where, like interval data, there is a fixed value between points, but unlike
interval data, ratio data have an absolute zero point.
http://www.llc.manchester.ac.uk/ctis/research/english-corpus/ [Last accessed 2 December
2012].
1
22
Gabriela Saldanha and Sharon O’Brien
At what point do we have sufficient data? This is a frequently asked question which is difficult to answer because it depends on so many variables (the
methodology, the research questions, hypotheses, time allocation, among others). Chapters 3 to 6 discuss data collection and address some of the relevant
issues in more detail, including the length of text and corpora for analysis and the
concept of saturation in participant-oriented and case-study research. In social
science and humanities research in general, and translation research in particular,
the trend is for researchers to work on their own. Examples of large teams of
people analyzing data sets are few. Consequently, it is not always possible to
analyze very large data sets. Also, the nature of the data collected in translation
research – for example, written or spoken linguistic data, behavioural data, narratives – compared with the natural sciences, means that automatic analysis is
challenging and not always desirable. This, in turn, tends to restrict the amount
of data analyzed. While some automation is possible in translation research (see,
for example, later chapters on corpus analysis and translation process research),
most of the analysis is manual.
Before researchers decide whether they have collected ‘enough’ data to address their research questions, they will first have to consider issues of validity
and credibility, which are addressed below. It is sometimes helpful to carry out
a small-scale pilot study prior to the main data collection phase. This will allow
the researcher to test selected methods of analysis and will give a feeling for
how much data might need to be collected to establish some level of credibility.
Pilot studies are discussed in more detail, where relevant, in relation to each of
the methodological approaches presented in Chapters 3 to 6. Another approach
to establish whether data are sufficient is to add layers of data over time, until
one sees a stabilization in the variability of results.
2.8
Qualitative, quantitative and mixed-methods approaches
The approach to take to one’s research should be determined by the research
question(s) and how best it/they might be addressed. The quantitative approach
is associated with the positivist epistemological position we mentioned earlier
while a qualitative approach is generally associated with the interpretivist position. According to O’Leary (2010:113), the qualitative tradition
calls on inductive as well as deductive logic, appreciates subjectivities,
accepts multiple perspectives and realities, recognizes the power of
research on both participants and researchers, and does not necessarily
shy away from political agendas.
Each approach has specific methodologies associated with it. A qualitative apOne current example is the PACTE translation competence research group at the Universitat
Autònoma de Barcelona, in Spain: http://grupsderecerca.uab.cat/pacte/es [Last accessed 2
December 2012].
Principles and ethics in research
23
proach in translation research can include critical discourse analysis, interviews,
focus groups, questionnaires (see Chapters 3, 5 and 6) while the quantitative
approach might be associated with corpus analysis, eye tracking, keystroke logging (see Chapters 3 and 4). It is important to point out that some methods can
produce data that can be analyzed both qualitatively and quantitatively (e.g.
survey data, think-aloud protocols and corpus analysis). A mixed-methods approach is the term used when several methods are used to collect or analyze
data. This is often understood to mean using both qualitative and quantitative
approaches. The two types of data can be collected simultaneously. Alternatively,
the researcher might opt for an initial qualitative phase followed by a quantitative phase, or vice versa. The first sequence has the advantage of allowing the
researcher to explore data qualitatively and to follow this exploration up with a
more focused quantitative analysis of the topic or sub-topic, while the alternative
of commencing with a quantitative phase has the potential advantage of exposing some trends that can then be further probed via qualitative data. Chapter 5
discusses mixed methods in participant-oriented research in more detail and
illustrates different ways of ‘quantitizing’ and ‘qualitizing’ data, that is, deriving
quantitative data from qualitative data and vice-versa.
Guba and Lincoln (2005:201) raise an important concern regarding the commensurability of competing paradigms (e.g. positivism and interpretivism), stating
that commensurability can be an issue “when researchers want to ‘pick and mix’
among the axioms of positivist and interpretivist models, because the axioms
are contradictory and mutually exclusive”. As we mentioned above, in the end,
the research question will dictate what the most appropriate approach is, but
it is worth taking potential contradictions into account when adopting a mixedmethods approach. As Creswell and Plano Clark (2007) point out, mixed methods
research is not just a way of combining qualitative and quantitative approaches,
but also “a research design with philosophical assumptions” (ibid.:5). The central
premise is that “the use of quantitative and qualitative approaches in combination provides a better understanding of research problems than either approach
alone” (ibid.). An argument along similar lines is made in Chapter 3 in relation to
the combination of corpus linguistics and critical discourse analysis.
When two methods are used to collect and analyze data on the same research
question, this is called triangulation, which means cross-checking the results
one set of data provides with results from another set of data. This is a practice
we would generally endorse, and we point to opportunities for triangulation of
results where appropriate in the chapters that follow.
2.9
Research operationalization
An important question to ask about the data to be collected and analyzed pertains to the unit of data. Data can pertain to either the macro or micro level
(Matthews and Ross 2010:114). Macro-level data are collected, for example,
from organizations, countries, systems and social entities, while micro-level
data are at the level of the individual, word, or text. In the case of translation
24
Gabriela Saldanha and Sharon O’Brien
research, macro-level data might pertain to professional translator associations,
country-specific laws regarding language and translation, to translation practices
within organizations, or to literary polysystems, to mention just a few examples.
Micro-level data might pertain to the use of specific strategies in a translated
text, individual translation strategies, or the length of time taken to translate a
text. Tymoczko (2002) aligns macro-level research with the cultural approach to
research in translation and micro-level research with the linguistic approach. In
her attempt to connect the two approaches, rather than allowing them to be
seen as competing and exclusive of one another, she encourages a convergence
which makes use of both macro-level and micro-level analysis, with data from
one type of analysis complementing and, hopefully, confirming the other.
Yet another important concept is the unit of analysis. This is not the same
as the unit of data. For example, the unit of data might be at the micro-level of
‘text’ and, while a researcher might analyze text in general, it is quite likely that
the unit of analysis (or measurement) will be further broken down into measurable concepts such as lexical items, sentences, clauses, phrases, collocations
and so on. On the macro level, the unit of data might be ‘legislation pertaining
to language and translation in country X’, but the unit of analysis in this context
might be specific laws or legal clauses.
The unit of analysis is linked with the important concept of operational
definitions or operationalization. Strictly speaking, operationalization refers to
the operations involved in measuring the dependent variable. Operationalization does not pertain only to quantitative approaches to research but is equally
important for qualitative approaches, where the operational definition can be
thought of as an explicit and precise definition that isolates the core components
of the variable under investigation. Let us go back to the example of the theory of
a MiniMax strategy we used earlier which, as a reminder, posits that translators
opt for the minimum amount of cognitive effort possible before proceeding to
deeper levels of processing. An important question from the outset is how can
I operationalize the MiniMax theory, i.e. how can I turn this somewhat abstract
theory into a measurable entity? An example of how this might be achieved
comes from Lörscher (1991), who equates evidence for a MiniMax strategy with
a dominance of sign-oriented translation, which he defines as a transfer of the
source language form into a target language form, without recourse to the sense
of the text segments involved. Sign-oriented translation could be operationalized
both via the translated product (where there is evidence of form substitution
without recourse to sense) and via the utterances of translators produced in
think-aloud protocols either during or after the translation process (for example,
where a translator might say during the course of a translation: ‘Entwicklung, yes
that’s development in English…, that’ll do’). Operationalization becomes even
more challenging with more abstract concepts, such as culture or the status
For a discussion of the problem of identifying a unit of analysis in an emerging field like
translation studies, see Tymoczko (2007:153). We discuss the unit of analysis with specific
reference to translation process research in Chapter 4.
Principles and ethics in research
25
of the translator. A good example is provided by Koskinen’s operationalization
of ‘culture’ in the context of an ethnographic study of institutional translation
(2008:40-43). Koskinen chooses a definition of organizational culture proposed
by Martin (2002, in Koskinen ibid.:41) that is useful for her purposes because it
describes aspects of culture, both material and ideational, that are manifested
and observable, such as stories people tell, relations among people, official
policies and formal arrangement (to name just a few). This definition is then
related to Scott’s (2001, ibid.) systemic view of the three pillars of institutions:
regulative systems, normative systems and cultural cognitive systems. Finally,
Koskinen identifies methods that will allow her to explore the manifestations of
culture described in the definition and belonging to those three pillars from an
ethnographic perspective.
Two crucial questions ought to be asked when operationalizing a concept: (1) what influence does the researcher’s beliefs and ideology have on the
proposed operationalization of the concept, and (2) whether or not the tools
selected can actually measure the concept the researcher wishes to measure?
Take, as one example, the concept of ‘source text difficulty’, i.e. how complex
the source text is and how this might impact on the translation process and/or
product. There are numerous ways in which the concept ‘source text difficulty’
might be and has been measured in translation research, such as subjectively
using native speakers as evaluators, using traditional readability indices, or
using rhetorical structure theory (Taboada and Mann 2006). The ability of any
of these methods to measure the construct or the degree to which they can do
this is open to question.
Operational definitions will dictate the approach for gathering data and the
type of analysis that can be performed on the data. Previously, we mentioned
using secondary data. There is a possibility that the operational definition used
when collecting secondary data differs from the researcher’s own operational
definition and, therefore, the implications of using secondary data that were collated under a different operational definition should be considered. At the very
least, differences in operationalization ought to be acknowledged.
2.9.1
Measurable variables
A variable “is simply something that varies … in some way that we seek to measure” (Langdridge and Hagger-Johnson 2009:40, original emphasis). This concept
is used primarily in quantitative approaches to research. The dependent variable
is the core concept we are trying to assess in our research question. We expect it
to change when it is exposed to varying treatment. The independent variables,
on the other hand, are things that we manipulate in order to see what the effect
is on our dependent variable.
Let us consider, for example, the research question: What is the effect on translation quality when time pressure is increased? The null hypothesis might be expressed
as: There is no change in translation quality when time pressure is increased. (In
fact, as we stated earlier, we expect that the opposite – alternative – hypothesis
26
Gabriela Saldanha and Sharon O’Brien
is true.) Let us assume for a moment that we have found acceptable ways of
operationalizing translation quality and time pressure. Then, we might expect
the results shown in Figure 2.2.
,.
•
•
•
•
10... " ...... ·,
......
Figure 2.2 A theoretical model expressing the relationship between time pressure (X axis) and translation quality (nominal scale on the Y axis)
In other words, as time pressure increases we expect translation quality to decrease. Of course, we would be naïve to expect such a straight linear correlation
between time pressure and quality, but this figure illustrates our alternative hypothesis and our expected relationship between the dependent variable (quality)
and the independent variable (time pressure).
One of the challenges is how to successfully isolate dependent and independent variables so that they are not influenced by other (‘confounding’)
variables. In the example above, it would not be unreasonable to expect that
the complexity of the text for translation might also have an impact on quality,
especially under conditions where there is significant time pressure applied.
So, text complexity might be another independent variable, as might degree of
experience of the translator (can more experienced translators cope better with
time pressure than less experienced ones?), or degree of specialization (hypothesis: translators specialized in a specific domain will produce higher quality
texts even under time pressure than those not specialized in the domain), to
name just two examples. We can design a research project to investigate the
effect of multiple independent variables on one dependent variable or, indeed,
even on multiple dependent variables.
Dependent and independent variables are typically associated with quantitative approaches to research, but qualitative approaches do not exclude them. A
qualitative approach to a research question may also be able to identify depen-
Principles and ethics in research
27
dent and independent variables.
When items are complex and harder to measure than other items, they are
typically called ‘research constructs’ rather than variables: “Constructs are unobservable variables, and therefore variables that have to be measured indirectly”
(Langdridge and Hagger-Johnson 2009:40). Complex ideas such as ‘status’, ‘acceptability’, or ‘equivalence’ may be best explored as constructs in translation research.
However, see Ji (2012) for examples of how the relationships between (1) texts and
source texts, (2) target texts and the target language linguistic and cultural system,
and (3) translation style and the historical development of the target language
system, can be mapped in terms of dependent and independent variables. In relation to the second type of relationship, Ji suggests that the independent variables
can be the ideological stances of the translator, and the dependent variables the
particular translation strategies developed for their work (ibid.:56).
On the topic of variables in translation research, another useful distinction is
proposed by Chesterman (2001a), who identifies two main variable types: profile
variables, which refer to aspects of the form of a translation such as stylistic or
syntactic features, and context variables, which refer to aspects of the translation
context and its consequences, such as text type or skopos, to name just two.
2.10 Research quality
Short of reliable methods and valid conclusions, research descends into
a bedlam where the only battles that are won are by those who shout
the loudest.
(Silverman 2006:310)
The issues discussed so far in this chapter are important to ensure that research
is planned in such a way that it meets high quality standards: considering our
epistemological viewpoint prior to commencing the project; identifying the model,
concepts, frameworks etc. with which we are working; giving due consideration
to the formulation and scope of our research question and hypotheses; carrying
out a literature review according to recommended practices; thinking about the
approaches we wish to take and the nature of the data we wish to collect; and
identifying independent and dependent variables in advance where appropriate.
However, undertaking a research project also includes undertaking to make some
contribution to the knowledge that already exists about a topic, and to ensure that
a contribution is made, the research should meet certain quality criteria: validity,
reliability and generalizability, each of which is discussed below in some detail
(Section 2.10). Since some of these criteria have been criticized as not being applicable to qualitative research and an interpretivist stance, we also devote another
section to alternative ways of measuring research quality (Section 2.10.4).
2.10.1 Validity
Validity is a multi-faceted, and somewhat contested, topic and hence significant
space is given to its discussion here. The central question around validity, according
28
Gabriela Saldanha and Sharon O’Brien
to Guba and Lincoln (2005:205) is whether one’s findings are “sufficiently authentic (isomorphic to some reality, trustworthy, related to the way others construct
their social worlds) that I may trust myself in acting on their implications”. The
very possibility of validation in translation studies research is contested. Tymoczko (2007:155) points out that the expansion of the concept of ‘translation’ and
the subsequent inclusion of a broader range of scholars has led to disagreement
over the possibility of validation. However, she insists on the importance of addressing the question of validation to strengthen research methods in this field
and suggests that it is important to acknowledge the limitations around claims
of validity and replicability and to face these methodological problems head on
(ibid.:159).4 Here, rather than attempting to argue the possibility, or impossibility,
of validation in translation research, we feel it will be more fruitful to highlight
the challenges and potential solutions for achieving validity in qualitative and
quantitative approaches to research so that researchers can make up their own
minds as to what understanding of validity they subscribe to.
At the most basic level, the validity of our results will depend on the extent to
which the data we collate and analyze can contribute to answering our research
question. For example, let us imagine we are interested in researching translation students’ attitudes towards the teaching methods they have been exposed
to in a particular university programme. We may decide to use a questionnaire
to collect data. The inclusion of a question about student attitudes towards the
particular university’s infrastructure (facilities, services, buildings, etc.) will not
help us answer our research question, and the validity of the data accumulated
through responses to that question is therefore questionable (notwithstanding
the fact that the responses might give us some interesting secondary data).
The definition of validity by Guba and Lincoln quoted above, according to
which validity is the degree to which results match as closely as possible the
real state of the world, assumes a positivistic research perspective. A less rigid
(and postpositivistic) understanding of validity “concerns the extent to which
justifiable inferences can be made on the basis of the evidence gathered”
(Le Grange and Beets 2005:115, drawing on Messick 1989; our emphasis).
The reference to inferences implies that, especially when using qualitative
approaches to research, one cannot claim absolute validity. In the past, this
resulted in some tension...
Purchase answer to see full
attachment