Methods for Researching the Second Language Classroom

Formal research Research that is conducted by an external researcher

drawing on one or more of the established research traditions. In many cases,
the issues investigated are of both theoretical and pedagogical significance.
They intend to contribute to theory.
Practitioner research Research conducted by teachers in their own
classrooms drawing on the principles of action research or exploratory practice.
It is motivated by pedagogic concerns and is directed at enabling teachers to
solve problems they are experiencing with their teaching or develop a deeper
understanding of some aspect of the quality of life in the L2 classroom.
Formal L2 Classroom Research
Interaction Hypothesis L2 acquisition is facilitated when a communication
problem arises that causes learners to try to resolve it through the negotiation
of meaning. Negotiation helps to make input comprehensible, prompts learners
to notice the difference between their own erroneous output and the correct
target language form through feedback, and pushes learners to self-correct.
Not all formal research is driven by theory. Much of the L2 classroom research
has been descriptive in nature, aimed at understanding a specific aspect of
teaching or learning and also at providing information that can ultimately be
used to shape a theory.
Formal L2 classroom research, then, whether experimental or descriptive, aims
to contribute to research-based language pedagogy.
That is, it seeks to provide teachers with information that they can use to
decide what and how to teach. Its characteristics are as follows:
l. The phenomenon investigated is determined by the researcher.
2. The researcher 'borrows' a classroom in order to carry out the study.
3. The researcher may also solicit the help of a classroom teacher to conduct
the research.
4. The research is either theoretically driven (as in experimental research) or
carried out with a view to developing theory (as in descriptive research).
5. The results of the research are written up in accordance with the
requirements of academic articles and with a view to publishing them in
academic journals.
6. In many cases the research is intended to contribute to research-based
language pedagogy.
Such research is of undoubted value -both for testing and developing theory
and for language pedagogy. A limitation, however, is that it may not address
the kinds of issues that preoccupy teachers and, in fact, may never reach
teachers as they are unlikely to read the journals in which it is published.

Practitioner Research
Practitioner research is research conducted by practitioners (usually teachers)
in their own classrooms either acting independently or in collaboration with
others. Practitioner research, then, is directed at enabling teachers to become
'expert knowers about their own students and classrooms'. In this way it aims
to make a direct connection between research and practice. To this end, the
research topics are not derived from theory but from teachers' desire to
experiment with some innovation in their classroom, to seek a solution to some
problem they have identified with their teaching or their students, or simply to
develop a fuller understanding of some aspect of life in their classrooms. lt
follows that it should be evaluated not in terms of whether it contributes to our
general understanding of some issue of theoretical significance but in terms of
the contribution it can make to teachers' practice of teaching and, through the
reflection that it can promote, to teacher development.
Arguably, then, the value of practitioner research lies more in the process of
conducting it than in the product of the research. Practitioner research is not
likely to be published in the same academic journals as formal research.
A key point that emerges in discussions of practitioner research conducted by
practitioners is the need for a model (in a way, a theory) for how to conduct it.
Action Research
Action research is a form of self-reflective enquiry undertaken by practitioners
in their own contexts of action. It is intended to lead to improvement in the
practice of some professional activity -in our case, in language teaching. It can
be undertaken by individual teachers or collaboratively involving a team of
teachers.
The model for conducting action research for teachers emphasizes a number of
features: (1) it is context-specific, (2) it is practical, (3) it is systematic, (4) it is
reflective, and (5) it is cyclical. The model proposes a number of iterative
phases:
1. Identifying an issue or problem relevant to a specific instructional context
(the initial idea).
2. Obtaining information relevant to the problem/issue (fact finding).
3. Working out a possible solution to this problem and devising ways of trying
this out (the action plan).
4. Trying out the solution in the specific instructional context (implementation).
5. Collecting data to investigate whether the solution is effective (monitoring).
6. If necessary, revising the action plan and proceeding through steps (4) and
(5) again or alternatively identifying a new issue thrown up by the initial study.
Such a procedure is context -specific in that the problems are identified by
teachers and are located in their own teaching. It is practical in that it is

directed at improving teaching. It is systematic in that it offers a set of clearly

delineated steps for conducting an action research study. It is reflective in that
it requires teachers to examine problems in their own teaching, identify
possible solutions and evaluate their effectiveness. It is cyclical in that it
recognizes the importance of continuous research to find solutions and the
possibility of new problems arising out of teachers' attempts to find solutions to
initial problems.
Clearly, it is not sufficient to simply identify a problem; it is also important to
consider the reasons for why the problem exists, as only then is it possible to
formulate a research question that can guide the study.
Nunan (1990b) reported that teachers' proposals tended to be rather grand and
unmanageable because they had failed to identify specific research questions.
The starting point for such research is not a problem but rather some 'task' that
the teacher is interested in using with the students. The idea behind my
proposal is that teachers would benefit exploring empirically whether the task
'worked' and that this would stimulate reflection on the choice of task and the
manner in which it was implemented.
I outlined a five step procedure for conducting a micro-evaluation: (1)
description of the task, (2) planning the evaluation, (3) collecting information,
(4) analyzing the information, and (5) reaching conclusions and making
recommendations.
I distinguished three types of micro-evaluation: (1) student-based (involving
eliciting the views of the students about the task, (2) response-based
(collecting data to see if the performance of the task accorded with its
pedagogic aims), and (3) learning-based (investigating whether any language
learning resulted from the performance).
Exploratory practice
Allwright (2003) formulated a number of general principles to guide the
practice of exploratory research. At the center of these is the notion of
classroom life', which Allwright suggests should be considered in terms of the
'quality of the learning', the 'quality of education' and, finally, the 'quality of
life' with the latter involving factors that lie outside the classroom itself. His
principles were:
1. Put the 'quality of life' first (i.e. 'practical problems' are best considered in
context and will involve a holistic understanding of the 'lives' of the participants
involved).
2. Work primarily to understand the language classroom life. Allwright
emphasizes that 'only a serious effort to understand life in a particular setting
will enable you to decide if practical change is necessary, desirable and/or
possible' (p.128). This requires converting a 'practical problem' into a 'puzzle'.
3. Involve everybody. Allwright sees exploratory research as a collaborative
endeavor, involving learners as 'co-researchers'.

4. Work to bring people together. The need for social harmony requires that
exploratory research be directed at achieving collegiality.
5. Work also for mutual development. Participants in the research should work
for each other's development.
6. Integrate the work for understanding into classroom practice. Allwright
argues that 'practitioner research must not become parasitic upon the life it is
trying to understand' (p. 129). This implies that any investigative tool used
should be part of the instructional activities that are a natural part of a lesson.
7. Make the work a continuous enterprise. Exploratory research needs to be
seen as 'a continuous, indefinitely sustainable enterprise' (p. 130). A corollary
of this is that teachers should avoid time-limited funding as this will
compromise the whole enterprise.
Exploratory practice, then, aims to make the time that teachers and learners
spend together 'pleasant and productive' and, in so doing, create the
conditions for pedagogic change.
For Allwright, teacher research needs to be feasible: he argues that if teachers
see classroom research as too demanding, they will not engage in it. It is for
this reason that exploratory practice needs to integrate enquiry into classroom
practice. He sees the role of academic researchers as 'consultants' rather than
'directors'; they should advise on the conduct of investigations but not attempt
to control them. Above, all exploratory research needs to be a collegial
enterprise; the investigations should involve learners as well as teachers and
should focus on 'puzzles' rather than on 'problems'.
There are a number of differences between action research and exploratory
research. One is the starting point -a 'problem' or, perhaps, a 'task' in the case
of action research and a 'puzzle' in the case of exploratory research. Another
difference lies in the methodology for the two approaches. Action research
employs similar methods of data collection to those found in formal research
and involves going beyond the materials used for teaching; exploratory
research embeds data collection into the actual practice of teaching. What they
have in common is an emphasis on the continuous nature of the enquiry. Action
research is 'cyclical' (although to what extent this is actual\y achieved by many
teachers is doubtful); exploratory research is a long-term enterprise and,
because it is part of teaching is potentially more sustainable.
Practitioner research: concluding comments
Practitioner research is research carried out by teachers working in their own
classrooms with a view to improving practice (in the case of action research) or
understanding the 'life of the classroom' (in the case of exploratory research).
Where such research is published, it typically takes the form of narrative
accounts. Both Penner's action research study and Bloom's exploratory
research study were presented as 'stories' about these two teachers'

experiences of how they investigated their classrooms and what they

discovered through these investigations.
MAIN RESEARCH TRADITIONS
Confirmatory Research
-

Theoretical underpinnings

1. The phenomenon under study (the language classroom) exists as an

objective reality that can be studied scientifically (i.e. by excluding the
subjective viewpoint of the researcher as far as this is possible).
2. Confirmatory research is theory driven. That is, the research questions are
derived from an explicit theory that predicts that certain types of relationships
in the phenomenon being investigated will be found. The purpose of
confirmatory research is to 'test' these predictions (called 'hypotheses') by
providing evidence that either confirms or disconfirms them.
3. In order to form hypotheses it is necessary to divide the phenomenon being
investigated into c1early-defined elements or constructs. The validity of the
research depends on whether these constructs can be shown to be
theoretically sound.
4. Once identified, the constructs under study can then be separately and
quantitatively measured. This is commonly referred to as an etic perspective.
The measurement of variables must be demonstrably reliable. That is, the
researcher needs to show that no bias or inconsistency has crept into their
measurement.
5. By ensuring the research is valid and reliable it is possible to arrive at
conclusions about the phenomenon under study that can be applied to the
wider population from which the sample being investigated was drawn. That is,
the findings of confirmatory research are not limited to the specific
classroom/teachers/learners under investigation.
Confirmatory studies of L2 teaching draw on theories of language teaching and
learning to identify the 'variables' that can be studied. A 'variable' is a general
theoretical construct. Examples of variables derived from a theory of teaching
are 'production-based instruction' and 'comprehension-based instruction'.
Examples derived from a theory of L2 learning are 'implicit learning' and
'explicit learning'. Some variables straddle the boundaries of teaching/learning
theory. 'Focus on form' for example is a variable that was derived from a theory
of L2 acquisition that claimed that attention to form in the context of meaningfocused communication constitutes the ideal condition for learning to take
place, but it is also an instructional variable in that it is realized through
specific instructional techniques such as corrective feedback.
For research purposes, theoretical variables need to be made operational. The
key point, though, is that in confirmatory research the variables chosen for
study are determined a priori by some general theory.

The relationship between variables can be of two kinds, leading to two different
types of confirmatory research -experimental and correlational. Experimental
research assumes a causal relationship between variables -that is variable A
(called the independent variable) causes some change in variable B (called the
dependent variable). In the case of language teaching research, this typically
involves investigating whether some aspect of instruction (e.g. corrective
feedback) results in some change in the students (e.g. a gain in grammatical
accuracy). Correlational research assumes a link between two variables but
without claiming that the link is causal.
In the context of the L2 classroom, the relationship between two variables is
very Iikely to be influenced by other variables, known as moderating variables.
-

Design

Experimental research - A true experimental study requires the following:

1. The independent and dependent variables are identified and clearly defined.
2. Some form of intervention in the domain under study is devised with the
purpose of investigating what effect the independent variable(s) has on the
dependent variable(s). This is referred to as the 'treatment'.
3. A number of groups are formed, some constituting the experimental group(s)
and one the control group (i.e. a group that does not receive the treatment).
4. Participants are assigned to the different groups randomly.
Classroom experimental studies are best described as quasi-experimental.
Experimental studies vary greatly in what constitutes the independent and
dependent variables, leading to design differences. In such studies it is often
necessary to include a pre-test to demonstrate what learners know prior to the
instruction. Performance on the pre-test can then be compared to performance
on a post-test. Delayed post-tests are needed to investigate whether any
learning evident in the immediate post-test is durable. A testing regime only
becomes essential when the research addresses whether the treatment
resulted in some change in the participants.
Finally, it should be noted that not all classroom experimental studies are
entirely product-oriented. A product-oriented study is one that focuses on the
effects of the treatment on the dependent variable. But L2 classroom
researchers have been increasingly aware of the need to also investigate the
'processes' that arise during the treatment. Investigating the 'processes' is very
important for explaining how the treatment had the effect it did (or didn't).
Correlational research - A correlational study aims to investigate the
relationship between two or more variables. In such studies there is no
'treatment' and no a priori division of the participants into groups. Instead, the
researcher obtains numerical measures of the variables under study and then
investigates statistically the extent to which these variables are related.

Many correlational studies are not classroom-based. They examine learner or

teacher variables by collecting data from learners outside the classroom.
-

Data collection

The data collected in confirmatory research will vary depending on the

research question. In all three cases, as in all confirmatory research, the data
were numerical.
I have identified three of the most common variables in language teaching
research: L2 learning, task performance, and learner factors.
L2 learning There is a growing recognition of the need to collect 'free
constructed responses' (e.g. by using communicative oral tasks) as these
provide the best data for investigating whether instruction has had any effect
on learners' interlanguage development.
Studies that have investigated task performance have collected a whole range
of different types of data. These studies collect samples of learners'
performance of different tasks, which are then analyzed to provide different
kinds of measures of L2 use, including purely linguistic measures (such as
fluency, complexity and accuracy) and discourse measures (such as
negotiation sequences and language-related episodes). Learner factors can
figure as moderating variables, as dependent variables, and in correlational
research. Numerical data can be obtained from classroom observation
schemes, from learners' responses to questionnaires or from tests of specific
learner factors. PAGE 24
-

Data analysis

Given that the data in confirmatory research are numerical, the methods used
to analyze them are statistical.
There are two basic kinds of statistics; descriptive statistics and inferential
statistics. Both are important as the results obtained from inferential statistics
cannot be properly understood without reference to descriptive statistics.
Descriptive statistics provide information about the measures obtained for
whole groups, although in some studies the scores obtained by individual
participants are also included. Group statistics include means (i.e. the scores
obtained by the individual participants divided by the number of participants)
and various measures of dispersion (e.g. standard deviation and range), which
provide information about the extent to which the scores of the individual
participants cluster around the mean.
Inferential statistics are of two major kinds. One kind provides a measure of the
extent to which the group scores are different. This kind is used in experimental
studies.
The second kind of inferential statistics provides a measure of the degree of
relatedness between two variables and thus is used in correlational studies. A
popular statistic of this kind is the Pearson Product Moment Correlation. This
provides a score (r). A perfect correlation (never achieved in the kinds of
research we will be considering) occurs when r = 1.0. There are also other

correlational statistics, including Multiple Regression Analysis, which seeks to

show not just the strength of the relationship between two variables but also
the extent to which the scores on one variable are predictive of scores on
another.
Descriptive Research
-

Theoretical underpinnings

Classroom descriptive research aims to produce qualitative and quantitative

accounts of classroom processes, the factors that shape these and their
implications for language learning.
l. Descriptive research adopts an emic perspective by providing a rich account
of specific instructional contexts.
2. For this reason, it typically involves only a few 'cases' (e.g. teachers, learners
or classrooms) and does not seek to generalize beyond these cases.
3. Researchers investigate these cases as they find them (i.e. there is no
attempt to intervene through some form of instruction).
4. Descriptive research emphasizes the need to understand phenomena in their
cultural and social contexts.
5. It often involves a research-then-theory approach. That is, the variables are
not usually decided a priori on the basis of some theory but rather emerge as
the research progresses. The findings of the research, however, may be used to
build a theory.
6. It assumes that knowledge and understanding of phenomena is subjective
and seeks to counter the subjectivity of the researcher by demonstrating that
the research findings are consistent with the data (i.e. are trustworthy) and
reflect the viewpoints of different participants.
- Types of descriptive research
Descriptive research relies on observation of the phenomena under study and,
in some types, on self-reports obtained from the participants involved. It is
possible to group the different approaches under two broad headings: (1)
interactional research and (2) ethnography.
Interactional research - Interactional research documents the interactional
features of classroom discourse. A variety of tools are available for achieving
this -discourse analysis, conversation analysis, and microgenetic analysis.
Discourse analysis assumes that classroom interaction is hierarchical in
structure and aims to provide a systematic description of this. Conversation
analysis seeks to 'characterize the organization of interaction by abstracting
from exemplars of specimens of interaction and to uncover the emic logic
underlying the organization'. The assumption here is that there are norms that
enable speakers to produce and interpret actions. Microgenetic analysis aims to

document change by showing how interaction assists participants to perform

specific tasks collaboratively when they are unable to perform these
independently.
A key characteristic of classroom interactional research is that it relies on the
evidence available from the interactions themselves (i.e. it rarely takes into
account the participants' explicit reflections on their practices).
Ethnographic research - Ethnographic research has a broader goal. It is 'the
study of people's behavior in naturally occurring, ongoing settings, with a focus
on the cultural interpretation of behavior'. It emphasizes the importance of
obtaining multiple perspectives by collecting a variety of data (e.g. through
participant observation, interviews, assembling relevant documents, and
member-checking) and using these to describe and understand common
patterns of behaviour. Nunan and Bailey (2009) identify three main
characteristics of ethnography: it involves longitudinal enquiry, it is
comprehensive, and it views people's behaviour in cultural terms.
Ethnography of communication - This focuses on the 'ways of speaking' that
occur in different classroom speech communities. It differs from interactional
research in that it 'looks for strategies and conventions governing larger units
of communication and involves more holistic interpretation'.
-

Design

The design of confirmatory research is usually cross-sectional (i.e. it involves

collecting data within a single or short period of time). ln contrast, descriptive
research can be either cross-sectional or longitudinal (i.e. data are collected
over a period of time).
Most (but certainly not all) interactional research is cross-sectional. 1t involves
identifying the specific phenomenon to be studied (e.g. corrective feedback;
teachers' questions; turn-taking) and formulating research questions, choosing
the specific classroom(s) to investigate, observing the interactions that take
place there (and, these days, audio-or video-recording them), extracting
episodes relevant from the data and describing them in accordance with the
procedures of the preferred research tool (discourse analysis, conversation
analysis, microgenetic analysis).
Ethnographic research is emergent and therefore longitudinal in design. The
scope of the research gradually narrows as it progresses. Watson-Gegeo (1988)
distinguished three stages. The 'comprehensive stage' involves obtaining a
general picture of the situation being investigated. Its aim is to study 'all
theoretically salient aspects of a setting' (p. 584). This is essential in order to
enter the next stage the 'topic-oriented stage' -where the specific topic to be
investigated is identified. It is only at this stage that research questions are
formulated. The final stage is the 'hypothesis-oriented stage'. This requires
further data collection in order to test specific hypotheses. However, many
reports of ethnographic research do not document these stages but instead
describe the findings of the final stage only.

Data collection

Interactional research calls for data relating to the interactions that take place
in the classroom. Such data can be collected in two principal ways: using an
interactional analysis scheme to record the different interactional behaviors
that occur or recording the interactions that take place and then preparing
transcripts of these for analysis. Long (1980) distinguished three different types
of interaction analysis systems: in a category system each event is coded each
time it occurs, in a sign system each event is recorded only once within a fixed
time span, while in a rating scale an estimate of how frequently a specific type
of event occurred is made after the period of observation.
One of the best-known interactional analysis systems is the Communicative
Orientation of Language Teaching (COLT).
Interaction analysis systems have gradually lost popularity as audio-and video
recording of classrooms has become easier. Interactional analysis systems
provide quantitative data (i.e. the frequency of different behaviors). Audio-and
video-recording afford both qualitative data (in the form of transcribed
extracts) and quantitative data (when the data have been coded in accordance
with a coding system).
Ethnographic research also employs classroom observation using interactional
analysis systems or recordings but is likely to supplement these with field notes
which can provide detailed contextual information not otherwise available.
Ethnographic research also collects other kinds of data; participant self-reports
(e.g. by means of informal and formal interviews or diaries) and documentary
information. One type of self-report that is becoming increasingly popular is
'stimulated recall'. The researcher plays back an extract from the recording and
then asks the participant in the extract (the teacher or a learner) to recall the
thoughts they had at the time.
-

Analysis

Quantitative data in descriptive research is analyzed using descriptive and

inferential statistics. As the data typically take the form of frequency counts, a
very useful statistical tool is chi-square. This allows the researcher to establish
whether the distribution of frequencies across a number of different categories
is statistically significant.
Qualitative data can be analyzed in a number of different ways. In a deductive
analysis, a set of predefined or expected themes or categories are used to
'code' the data. In an inductive analysis the data itself is allowed to determine
the themes/categories by using what is called a 'grounded theory' approach.
This involves 'examining the data first to see into what kind of chunks they fall
into naturally and then choosing a set of concepts that helps to explain why the
data fell that way'.
-

Hybrid research: process-product studies

Research paradigms be considered in terms of three dimensions: (1) design,

(2) type of data collected and (3) method of data analysis. Using these he
describes two 'pure' research paradigms (corresponding to the distinction
between 'confirmatory' and 'descriptive' research) and six 'mixed' forms. Of
these mixed types, one that is very common in language teaching research is
'experimental-qualitative-statistical'. That is, the basic design is experimental,
but qualitative data are collected, which are then quantified by counting the
frequency of occurrence of specific categories established qualitatively. These
frequencies can then be analyzed using statistical methods. This type of
research is often referred to as 'process-product research'.
The importance of conducting process-product research became clear as a
result of the indeterminate results produced by the method-comparison
studies.
Process-product studies constitute one of the most powerful ways of
investigating language teaching. They enable the researcher to investigate
what happens when a particular experimental treatment is implemented. Thus
they provide essential evidence for explaining the results of product-based
analyses. It does not follow that externally defined treatments will result in
actual differences at the process level. This can explain why no differences in
learning outcomes occur. Where product differences are evident, it is possible
for the researcher to explore which of the specific process features account for
these differences.

ACTIVITY: Student Seminar on Ellis (2012) Ch.2, Methods for researching the
second language classroom, pp. 21-49.
INSTRUCTIONS: Based on your reading evidence submitted, work in pairs or trios to prepare the
answer to the topic question you will be randomly assigned. You will have 15 minutes to get prepared for
participation in the Seminar. Also a moderator will be chosen to lead the discussion during the Seminar.
A) What are the main characteristics of formal L2 classroom research? Please, provide examples.
B) What are the main characteristics of practitioner research? Please, illustrate with examples.
C) What are the main research traditions in second language classrooms? Please, describe each of
them.
D) Concluding remarks on Ellis Chapter.
NB. Each team will have 10-15 minutes to present A, B, C, or D