IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 15
Copyright © eContent Management Pty Ltd. International Journal of Multiple Research Approaches (2008) 2: 15–30.
Blending qualitative and
quantitative analyses in observing
interaction: Misunderstandings,
applications and proposals
A U G U S T O GN I S C I
Department of Psychology , Second University of Naples, Caserta, Italy
R OGER B A K E M A N
Department of Psychology, Georgia State University, Atlanta GA, USA
V ICENÇ Q U E R A
Department of Behavioral Science Methods, University of Barcelona, Barcelona, Spain
ABSTRACT
In this contribution we discuss how, when observing social interactions, qualitative and quantitative research can each enrich the other. First, we highlight the usefulness of qualitative research for
subsequent quantitative studies. We also mention some possible misunderstandings in the way
qualitative researchers view quantitative research. We discuss criticisms regarding ‘natural’ units of
analysis, the use of ‘pre-defined’ categories, the sequential context, the multifaceted aspect of interaction and the role of transcripts. Second, we present two research examples – one based on time
series graphs, the other on similarity maps – that demonstrate how quantitative analysis can be
used to identify points in interaction that require further qualitative analysis (eg particular phases
of interaction, critical points such as positive/negative shifts, unique cases).
Keywords: quantitative and qualitative methods; social interaction; observation; coding; units; similarity
patterns; critical points; unique cases.
I NTRODUCTION
O
bservational research of social interaction has
been affected profoundly by technology as it
has evolved from film, to analogue tape, to digital
and from expensive and cumbersome to inexpensive and portable recording devices (Bakeman &
Gottman 1997). In many disciplines where the
study of communication, conversation and social
interaction is of concern (eg social psychology, psy-
chiatry, anthropology, linguistics, sociology, ethology), the ability to study social interaction through
repeated viewings of video and audio recordings
has proven immensely useful (Bull 2002). The
consequences for what could be called a revolution
in recording technology have been profound for
the way we think about human communication
(Kendon 1982); this revolution has affected quantitative and qualitative approaches alike (Jacobs,
Volume 2, Issue 1, June 2008 INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES
15
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 16
Augusto Gnisci, Roger Bakemanand Vicenç Quera
Kawanaba & Stigler 1999). Human interaction is
essentially a dynamic process that unfolds in time.
Technology that allows us not just to view interaction, as we do when viewing live, but to view it
repeatedly (literally, to review) has been invaluable
to scholars of interaction.
We will discuss quantitative/qualitative issues
from our point of view as observational researchers
who mainly use quantitative analysis. We will focus
on data collection because, in our view, critical issues
between qualitative and quantitative approaches are
largely concentrated in this research step. For quantitative approaches we will reference mainly our past
studies and research (Bakeman 2000; Bakeman &
Gnisci 2005; Bakeman & Gottman 1997; Bakeman
& Quera 1995a 1995b; Gnisci 2005; Gnisci &
Bakeman 2007; Quera, Bakeman & Gnisci 2007)
and for qualitative approaches we will reference classical contributions that range from ethnography,
ethnomethodology and ethogeny to conversation
and discourse analysis (Edwards & Potter 1993;
Drew & Heritage 1992; Harré 1979; Harré &
Gillett 1994; Harré & Secord 1972; Heritage 1984;
Goodwin & Duranti 1992; O’Barr 1982; Sacks,
Schegloff & Jefferson 1974; Schegloff 1993).
Throughout, we focus on the uses each
approach has for the other. We begin by describing
the usefulness of qualitative research for quantitative researchers and then list and discuss some
common criticisms qualitative researchers have
concerning quantitative research. After having
briefly described the quantitative approach to interaction that we pursue, that is sequential analysis,
we present two examples of it, which show how the
results of quantitative observational research can
inform and guide subsequent qualitative research.
One is based on time series analysis, the other on
similarity maps and both represent concrete possibilities of integration between the two approaches.
The usefulness of qualitative research
f o r q u a n t i t a t i v e re s e a rchers
Qualitative research is often viewed as both a precursor to and a partner for quantitative research (eg
O’Barr & Lind 1981). Its results not only stimu16
late the development of theories about specific
phenomena, but also aid in development of observational schemes to measure those phenomena.
Qualitative observational research applies an
inductive method to the comprehension of interaction, using such methods as participant observation, conversation and discourse analysis,
ethogeny, ethnography and so on. The results of
such research often stimulate subsequent quantitative research (an example is provided in the
next paragraph). Qualitative research aims to
obtain a detailed and deep comprehension of the
phenomenon studied, formulating descriptions,
possible explanations and hypotheses. However,
when rigorous experimental methods are applied
to the study of social interaction, as social psychology often attempts to do, important realworld matters may be overlooked. Each approach
has risks: studies may be based on a too narrowly
constructed a view of reality, on one hand, or
based on too abstract a construction of the world
without sufficient empirical grounding, on the
other (Lewin 1944). From this point of view,
qualitative studies can and often do provide the
quantitative world a needed breath of fresh air.
An excellent example of quantitative research
stimulated by qualitative results is provided by
development of the face model used to analyse
televised political interviews (Bull & Elliott
1998), which postulates that the main concern
for politicians engaged in a public situation is to
avoid losing face (Bull, Elliott, Palmer & Walker
1996). The politicians not only have to appear
competent and prepared on every possible issue,
they also have to try not to disappoint their electorate, annoy different possible voters, contradict
statements previously made by themselves and
their colleagues, or damage the image of either
their party or significant others (Bull 2002; Gnisci 2008). In such communicative conflict situations, shrewd politicians use minimal, elaborated,
implicative, or no-reply answers to protect themselves (Bull & Mayer 1993); whereas interviewers, attempting to force politicians into an
avoidance–avoidance conflict (Lewin 1938; Bave-
INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES Volume 2, Issue 1, June 2008
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 17
Blending qualitative and quantitative analyses in observing interaction
las et al 1990), use face-threatening and coercive
questions. Based on this conceptual analysis,
three different ‘code and count’ category systems
were developed, one for threatening questions,
another for coercive questions and the last for
responding/not responding to the questions (Bull
1994, 2003).
One of the primary influences on this model
was Goffman’s (1955) face theory. Goffman’s theory was developed for explaining the ways in
which people present themselves in everyday life,
support or challenge the claims of others, and, in
particular, deal with challenges to their identity
perpetrated by others (Bull 2002). In his first
article on face-work he exemplified a number of
strategies for both avoiding threats to face and
repairing possible damages to face. Goffman was
primarily a theorist who put forward a conceptual framework for studying social interaction (Bull
2002). Although Goffman considered himself a
‘human ethologist’ (1971), his methods were
uniquely his own, made use of his exceptional
observational capacities and freely ranged from
participant observation of social encounters to
thoughtful reading of etiquette books and advertisements (Bull 2002). In particular, his work was
not based on systematic and detailed analysis of
recorded material.
In sum, there is a profound influence of Goffman’s face theory (1955), based on qualitative
methodology, on Bull’s and associates’ face model
of political interviews, a model based on the use
of category systems and inferential analysis. We
can easily track the lines that connect Goffman’s
qualitative work to the quantitative research
realised by Bull’s research group. Goffman’s face
theory guided Brown and Levinson (1978,
1987), who in their politeness theory focused on
positive and negative face. Jucker (1986), discussing the politeness theory on politicians’ news
interviews, identified 13 different ways in which
the face of a politician can be threatened in the
1
course of the interview. Bull and associates (Bull
et al 1996) found these categories too general for
studying specific question-answer exchanges and
expanded the list to 19 that, in turn, could be
organised in three superordinate and not mutually exclusive categories of questions: ones that
threaten directly the face of the interviewed
politician, his or her political party, or significant
others. A hypotheses-testing study then applied
this category system to a set of 18 interviews
from the 1992 British General Elections and
eventually to many others.
A second example of the influence of qualitative on quantitative research is provided by the
Interruption Coding System (ICS), a comprehensive classification of turn-taking (Roger, Bull &
Smith 1988) that is based on a well known, seminal paper by Sacks, Schegloff and Jefferson
(1974). In this paper the mechanism of alternation of turns was extensively studied using a
detailed empirical analysis of phone calls to a suicide prevention center. The calls were transcribed
in considerable detail, which let the investigators
identify the basic features of turn alternation in
conversation and ultimately propose a system of
rules that described how turn-taking is organised
(for the development of ICS see also Bull 2003:
82–87). Again, qualitative work provided basic
insights for subsequent quantitative hypothesestesting research. 1
Qualitative research is important, not just
when formulating hypotheses, but during other
phases of quantitative observational research as
well. Many handbooks describe these research
phases (ie Reis & Judd 2000; Eid & Diener
2006). Here we wish only to state that qualitative
analysis can contribute during various steps of the
research process: from the planning of research to
data collection; establishing reliability and (particularly, content) validity; the data analysis stage
and interpretation and communication of results.
In particular, it plays a vital role in the formation
An interesting example of the blended use of quantitative and qualitative sequential research that the reader may wish to
consult is Muntigl and Turnbull’s (1998) study of facework in arguing. Another instructive example is O’Barr and Lind’s
(1981; O’Barr 1982) ethnographic and experimental studies on linguistic styles in legal examinations.
Volume 2, Issue 1, June 2008 INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES
17
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 18
Augusto Gnisci, Roger Bakemanand Vicenç Quera
of hypotheses and the development of coding systems and it is essential in new areas of investigation (Bakeman & Gottman 1997).
Some common qualitative criticisms
o f q u a n t i t a t i v e re s e a rch and some
answers
Fruitful cooperation between qualitative and
quantitative approaches can come from two
sources: from research that integrates the
approaches and from discussion of what we think
are misunderstandings, false beliefs even, that each
approach has toward the other. Here we focus primarily on qualitative critiques of quantitative
approaches because, in our experience, they often
serve as obstacles to reciprocal comprehension. We
emphasise matters that often concern qualitative
researchers: quantitative researchers’ unit of analysis and their behavioral coding schemes, the issue
of ‘pre-defined’ categories, the concept of sequential context, the problem of how to grasp the complexity of social interaction and finally the role of
transcripts in observational research.
The ‘natural’ units of analysis
According to many qualitative researchers a basic
concern is that quantitative researchers’ units of
analysis do not correspond to how people organise themselves during interaction and the coding
categories do not reflect how participants in fact
parse their interaction (Edwards & Potter 1992;
Goodwin & Duranti 1992). They argue that, in
our understanding of interaction, we should privilege the interpretations, the meanings and the
orientations of the participants (Schegloff 1997).
In qualitative terms, the unit that organises the
interaction and the behavioral categories that
describe interaction in quantitative studies are
not ‘natural’. Consider first the issue of unit.
Some misunderstandings can arise from confusing the coding unit with the descriptive statistics derived from coding, such as, eg rates, mean
durations, etc. The coding unit in an interactional study is the ‘entity’ to which coders, following
rules, assign a particular category (Stevens 1946).
18
Descriptive statistics are elaborations of the data
collected. Therefore, they concern data analysis,
not data collection. Typically the entity coded is a
behavior, but often coders are asked not just to
assign a code to a behavior, but also to note its
onset and offset time as well. Thus the coding
unit would be a behavior but the measurement
unit one of time (eg a second). Measurement and
coding units are often confused, probably because
in some cases they overlap (eg when time intervals are coded), but they are different concepts.
Qualitative researchers often call the units they
use ‘natural’ because they reflect the organisation
people give to interaction. A striking example is
provided by conversational analysis where transcripts are often segmented into turns-of-talk
(Sacks et al 1974). However, units as turn or
speech are used also by quantitative analysts
(Gottman 1979; a similar term is ‘thought unit’).
Gottman (1983), for example, used such units in
a quantitative way in his well known study on
how children become friends. As a further refinement, each natural unit can be coded on more
than one dimension (Bakeman & Gottman
1997, term this cross-classifying events), which
we regard as important because it allows the use
of modeling strategies for analysing data (eg
Gnisci 2005, 2008 in press).
T h e i s s u e o f ‘ p re - d e f i n e d ’ c a t e g o r i e s
Beyond the coding unit, behavioral codes or categories are sometimes criticised as being ‘predefined’ (Peräkylä 2004) by qualitative analysts.
Psathas (1995), for example, considers coding
systems as arbitrary and reductive, fixed and
abstract; they distort behavior by forcing it to fit
preconceived categories and do not give sufficient
consideration to the context and the meaning of
interaction. From a qualitative point of view,
observers should not impose their own preconceived categories onto reality, but, on the contrary, should make use of the way people
categorise themselves as manifested in their own
talk (Bull 2002). Here we will not question qualitative assumptions, asking eg by what method
INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES Volume 2, Issue 1, June 2008
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 19
Blending qualitative and quantitative analyses in observing interaction
qualitative analysts claim that they are not imposing categories but merely labeling what can be
observed to exist, a criticism that Billig (1999)
addresses to conversation analysis (see Mey
2001). Aided by Bull’s (2002) discussion of these
assumptions, we will try to address what we view
as misunderstandings.
We would argue that participants’ categories can
be used in a quantitative way. Indeed, researchers
can and often do, create ad hoc category systems,
that is, systems that are inductively based on the
recorded material used in the research that attempt
to reflect what participants experience and how
they categorise their experience. If the point of
view of participants is paramount and a qualitative
analyst can grasp it, why cannot a quantitative one
do likewise when developing an ad hoc category
system? It is worthwhile here to contrast the criticism on pre-defined or preconceived categories (see
Psathas 1995) with the vivid description of the
beginning of a quantitative observation provided
by Bakeman and Gottman (1997: 12):
We do not think that an investigator need have
a coding system before collecting data of interest. For example, Gottman began his research
on acquaintanceship in children (1983) by collecting tape recordings of children becoming
acquainted. He had no idea at first which situations and experimental arrangements were
best for collecting data. Although the literature
suggested some social processes to study, … he
had little idea how to operationalize these constructs. A great deal of work was necessary
before a useful category system was devised.
He also found that several different coding systems were necessary to capture different
aspects of the children’s interaction.
Bakeman and Gottman then comment: ‘The
wonderful thing about observational research is
that it maximises the possibility of being surprised’
(1997: 13). Similar words are quoted by Rosenblum (1978) when describing the initial stages of
observational research. The deep, rich and detailed
qualitative work, often carried out with a kind of
ethnographic method by quantitative researchers,
Bakeman and Gottman characterise as important
hypotheses-generating observation.
True, after development and once the coding
phase of a research project begins, quantitative
researchers are extremely reluctant to make additional changes to their coding systems because to
assure comparability this would require them to
recode material previously coded (see Section 2
above). However, this does not mean that their
coding systems are ‘pre-defined’ in an absolute
sense. The researcher can adapt and modify them
in the preliminary phase of qualitative observation, as we have shown; and he/she can modify
and improve the categories, their contents and
many other features in future research: ‘Coding
systems can also change over time’ (Bull 2002:
20), they can be improved to provide a better
representation of what they initially were thought
to observe. They develop in a Darwinian sense in
contrast to the static picture that qualitative analysts sometimes provide of them. The observational capabilities of an analyst during qualitative
observation can benefit from previously carried
out qualitative observation; likewise with quantitative analysis the use of categories, the coding
process and the reflection on these activities and
practices in previous research may bring the
quantitative researcher to modify, improve and
redefine all or part of the category system. Even if
initially adopted, a category system can in any
case be adapted for future research. Bakeman and
Gottman (1997) provide an example of such
development, noting how Parten’s (1932) seminal
coding scheme for preschool children’s play was
first modified by Smith (1978) and then further
elaborated by Bakeman and Brownlee (1980).
Two final points: quantitative research is usually well-served when investigators take into
account what people feel, how they organise conversation and how they perceive and categorise
situations or events in their speech, but this does
not necessarily invalidate category systems that
are less or not at all participant-oriented. On the
contrary, they can be highly informative, precisely
Volume 2, Issue 1, June 2008 INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES
19
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 20
Augusto Gnisci, Roger Bakemanand Vicenç Quera
because they are devised by outside observers
(Bull 2002). We are not convinced that the only
valid view is from the inside.
Moreover, according to Bull (2002), a positive
aspect of category systems – one that is often
under-appreciated by qualitative researchers – is
that, instead of being a hindrance, they can actually aid perception. Indeed, they focus attention on
phenomena that scholars in the field have found
important both empirically and theoretically.
True, when adopted mechanically and uncritically,
coding systems can work as blinders, but in general they act as corrective lenses that allow a better
view (Bakeman 2000). For this reason category
systems are very useful when training observers or
professional people in communication skills (Bull
2002); they also provide focus for observers,
which can improve their reliability.
T h e s e q u e n t i a l c o n t e x t re f l e c t e d i n
the categories
Qualitative researchers sometimes maintain that
categorical observation does not take context
into consideration, particularly sequential context, that is, what happened just before and
immediately after the act that is being considered (Heritage 1984; Peräkylä 2004). This is a
common criticism, for example, of Bales’ (1950)
Interaction Process Analysis (IPA) that in fact
argues for the use of context only when classification is uncertain or dilemmatic (Peräkylä
2004). However, the IPA is a very early category
system and over time systems have become
increasingly more sophisticated regarding context. For example, one of the basic tenets of the
turn-based category system called InitiativeResponse Analysis (I-R) is the concept of linguistic context (Linell, Gustavsson & Juvonen
1988). According to the authors, each turn has a
retroactive and a proactive strength that links
each turn to the previous and following one or
ones. From this Bachtinian view (Bakhtin
1986), each turn is both context-determined
and context-renewing, a determination which
some qualitative authors consider a prerogative
20
of qualitative analysis (Drew & Heritage 1992).
In I-R each categorisation requires a shrewd
evaluation of the context by the observer.
One well known dispute concerns the category
system used by the Linguistic Category Model
(Semin & Fiedler 1988, 1991) that, according to
qualitative-oriented authors, abstracts words from
context and ignores the sequential, contextualised, interaction-oriented nature of talk in interaction (Edwards & Potter 1993, 1999). Although
the dispute is far from being resolved, it is worth
noting that the authors of the category system
reject this claim, pointing out that contextual
information is intrinsic to the four basic dimensions on which the category system is built (see
Bull 2002). We do not know who is right. However, the fact that context can be intrinsic to the
categories is a point worth making. In sum, many
category systems now attempt to incorporate the
notion of context and the notion that contextualism is a prerogative purely of qualitative methodologies no longer seems tenable (see Dawson,
Fischer & Stein 2006); measurement and context
can go together.
The complexity of the interaction
Qualitative scholars sometimes question whether
quantitative studies can take into consideration
the many different levels and aspects of interaction
sufficiently. Sometimes, but not always, this is a
reasonable concern; but it is not an inherent limitation of quantitative research. For example, what
Peräkylä (2004) attributes to the Sacksian research
tradition – its focus on discrete aspects of the
organisation of interaction such as turn-taking,
sequence organisation, repair, choice of words, etc.
and their intersection – when contrasting it with
the quantitative tradition represented by the Balesian one, could likewise be attributed to many
quantitative observational studies.
A good example of a quantitative study that
organises its coding systems both hierarchically
and sequentially is Bearison and Dorval’s (2001)
study of child negotiation. They were primarily
interested in those times when children, who had
INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES Volume 2, Issue 1, June 2008
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 21
Blending qualitative and quantitative analyses in observing interaction
been instructed to co-invent a board game,
entered into what they termed a negotiation
episode. They asked observers, working from
video records, first to identify episodes of a certain sort before moving on to another level of
coding, which in fact is a relatively common and
useful coding strategy. The primary coding unit
for Bearison and Dorval (2001) consisted of a
negotiation episode, but, reflecting their interest
in the process of negotiation itself, they asked
observers to consider a second unit: the conversational turns individuals take, which are related
hierarchically to (ie are nested or embedded within) negotiation episodes. Thus coding proceeded
on two levels: first, it detected negotiation
episodes, then it identified the conversational
turns within them.
Several dimensions of negotiation episodes
were of concern, including their topic and their
outcome. Topic codes included rules or methods
of play and goals or purpose of playing; negotiation outcome codes included unresolved, passive
acquiescence, active acceptance, joint compromise and joint elaboration. Additionally, two
dimensions of conversational turns were coded:
their function and the justification provided.
Function codes included initial proposal, counter
proposal, disagreement and agreement; justification codes included: none, factual and perspective-taking (justifications that consider the other
child’s point of view in some way). The actual
study included more dimensions and categories
than given here, but these few examples should
give some sense of what a realistic and relatively
complex set of coding schemes might attend to.
Some of these codes may not seem immediately
obvious or familiar to all readers, but this only
illustrates how codes necessarily reflect the theoretical context of the study they serve, in this case
a study of relations between social context and
cognitive development based on the works of
Piaget and Vygotsky (Mooney 2000). In sum, if a
task of conversational analysis is considering larger units than a sentence or utterance ‘conceived as
sequences of activity and their component unit
turns as turns-within-sequences’ (Drew & Heritage 1992: 18), then considering negotiation
episodes and their embedded turns seems a task
similar to one usually considered a prerogative of
qualitative analysis.
The ro l e o f t r a n s c r i p t s
Qualitative analysts often complain that quantitative results do not allow the reader to gain a sense
of the original data (eg video recordings or transcripts) whereas qualitative data does. Conversation analysts, for example, argue that their
transcripts reproduce the sound of conversation
as much as possible and that they still remain
accessible to ‘linguistically unsophisticated readers’ (Sachs et al 1974: 734) who are familiar with
the same socio-cultural context (Edwards & Potter 1992; Goodwin & Duranti 1992). Actually,
often transcripts omit basic features (tempo,
pitch, loudness) and can be difficult for others to
understand in other ways as well, as the following
example from the original article by Sacks, Schegloff and Jefferson (1974) shows: ‘I’d a’ cracked
up ‘f duh friggin (gla-i(h)f y’kno(h)w it)
sm(h)a(h) heh heh’ (reported by Bull 2002).
In any case, the assertion that the products of
qualitative analysis (ie transcripts) are understandable to participants or persons like them
and the ones of quantitative analysis are not (eg a
graph or a table with numerical data) remains at
present an assertion of principle; it has yet to be
tested empirically with whatever methods are
deemed most appropriate.
In any event, even if their transcripts could do
the work the qualitative analysts assume they do,
such a ‘return to data’ may not be so essential.
Once interaction data has been coded and their
sequences have been analysed quantitatively and
consistent temporal associations have been discovered, we do not necessarily need to return to
the original data (eg transcriptions) in order to
understand the interaction process or dynamics.
Sometimes it may help, other times, not. Partly it
depends on how we view transcripts. For qualitative analysts transcripts are paramount and are
Volume 2, Issue 1, June 2008 INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES
21
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 22
Augusto Gnisci, Roger Bakemanand Vicenç Quera
considered ‘data’; in our view, transcripts are primarily an important aid in observing, a ‘form of
assistance’ (Bull 2002: 9).
TWO EXAMPLES FROM THE
SEQUENTIAL ANALYSIS APPROACH TO
INTERACTION
This section discusses two examples of possible
qualitative applications of quantitative analysis of
interaction, taken from the broad approach to
interaction named sequential analysis. Sequential
analysis is an analytic approach that views interaction as a dynamic, sequential process that
unfolds in time between two or more actors who
may exert reciprocal influence (Argyle 1967;
Bakeman & Gottman 1997; Heyns & Lippitt
1954) and tries to grasp this dynamic character in
a quantitative way.
Qualitative uses of quantitative
research I: Gottman’s use of time series
graphs derived from categorical data
One problem in studying interaction is taking
into account different levels of analysis. Often the
different levels are organised hierarchically, as for
example when turns of talk are nested in different
phases of interaction. Sometimes researchers have
an a priori taxonomy of the phases of interaction
based on the literature. However, a taxonomy of
phases of interaction could also be based on
empirical evidence. For example, Gottman (1979,
1983) created time-series analysis data from a
stream of categorical data from marital interaction. From the categorical data he derived a new
variable that was the total positive minus the total
negative turns up to that point in time, separately
for husbands and wives. The resulting graphs for
two different couples are shown in Figure 1.
When both husband’s and wife’s graphs were
negative (first graph), interaction was in a phase
of negative affect reciprocation, when they were
both positive (second graph), there was a phase of
positive affect reciprocation and if one partner’s
graph was positive and the other’s was negative
the interaction was in a phase of compensation
22
with a partner who gave in most of the time in
response to the partner’s complaints. Therefore,
taxonomies of couples and of phases of interaction can be classified in an inductive way. Once
the phases of interaction are visualised and segmented, it is possible to come back to transcriptions and video recordings and provide a
qualitative analysis of the different phases. In this
example, quantitative analysis can be the basis for
a subsequent qualitative analysis. For example, if
we went back to transcripts, we might see that
cross-complaining sequences characterise negative
reciprocation in the agenda-building phase of
marital interaction, as in the following extract:
W: I’m tired of spending all my time on the
housework. You’re not doing your share.
H: If you used your time efficiently you
wouldn’t be tired.
Once quantitative analysis is able to detect
particular phases of interaction, such as positive
or negative reciprocation, cross-complaining, validation or contracting sequences, then qualitative
analysis can deepen our comprehension of the
different phases.
Graphs from time-series, as the ones we have
described, can prove useful when a researcher is
looking for critical points in interaction. Critical
points are when the interaction begins in a very
positive/negative way and dramatically changes
becoming negative/positive, respectively. In this
case the slope of the graphs suddenly inverts.
Therefore, quantitative analysis can prove useful
to identify critical shifting points that qualitative
analysis can in turn analyse.
Qualitative uses of quantitative
r e s e a rc h I I : E x p l o r i n g p a t t e r n s o f
similarity in interaction sequences
w i t h p ro g r a m R A P
Researchers who study human interaction are
usually interested in detecting temporal patterns
within observed sequences of codes. A pattern is
‘the repeated or regular way in which something
INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES Volume 2, Issue 1, June 2008
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 23
Blending qualitative and quantitative analyses in observing interaction
FIGURE 1: G RAPHS
OF I N T E R A C T I O N S F R O M TWO C O U P L E S ( F R O M
happens or is done’ (Collins COBUILD English
dictionary) and it is then synonymous with order
and predictability. Therefore, searching for patterns in sequences of codes that represent interaction amounts to detecting whether two or more
B A K E M A N & G OTTMAN 1 9 9 7 : 5 5 )
codes often occur in succession, whether they
tend to occur at the same time, or whether they
tend to occur within a specific time intervals of
each other. Different types of patterns can be
potentially detected depending on how interac-
Volume 2, Issue 1, June 2008 INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES
23
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 24
Augusto Gnisci, Roger Bakemanand Vicenç Quera
tion codes are represented (as event sequences,
timed event sequences, etc. For a classification of
types of sequential data see Bakeman & Quera
1995b). For example, the following coding
scheme could be used for representing verbal
interaction in a couple (w and h denote wife and
husband): wa and ha (approves), wc and hc (complains), we and he (emotes), wp and hp
(empathises), wn and hn (negates), wo and ho
(other utterances). Each verbal turn of wife and
husband is assigned one code and only the order
in which the codes occur is of interest. Suppose
that a couple was observed for a certain period of
time and the following sequence was obtained
(where every wife turn is followed by one husband turn and vice versa):
we he we he we he we hc wa hp we hc wc hc
wc hc wc he we he we he we he we he we hc
wc hc wc hc wc hc wc hc wc hc wc hc wc hc
wc ho wc hc we he wn hn wo ha wo ho wo
hn wo ho we he we he we hp wp ha wa ha
wa hp wa ha wa hp wa ha wa ha wa hn wn
hn wn hn wn he we he we he we he we he
wp hp wp he we he
A classical approach to study such sequences of
acts consists in tallying transitions among adjacent codes (eg wife-to-husband for all the codes),
organising them in a sequential cross-tables and
applying to these tables the adequate descriptive
and inferential statistics that show if participants
respond to their partner randomly or in systematic ways (eg husband complains after that wife
complains) (Argyle 1967).
However, besides that kind of quantitative and
molecular result, an exploration of the sequence
as a whole can provide new insights as to whether
certain patterns exist, where in the sequence they
occur and even whether they tend to repeat in
different but comparable sequences. Such global
exploration can be carried out by first computing
indices of similarity among parts of the sequence
(each part defined as a window containing several
consecutive codes or turns), then representing
them as a two-dimensional map. Code repeti24
tions, chains composed of certain codes occurring
in succession and possible patterns consisting of
codes separated by a more or less constant interval may be detected by visual inspection of that
map. Moreover, the map as a whole can indicate
whether the sequence is probably random or
whether it contains patterns, and, if so, where
they tend to be located in the sequence.
Program RAP (for RAndom Projection, Quera
2008) transforms event or timed event interaction sequences into such a map; the program represents successive windows in the sequence by a
quantitative vector by means of an analytical
technique called ‘random projection’ (eg Mannila
& Seppänen 2001), then computes the similarity
between every window and every other window.
Similarity maps provided by RAP are analogous
to recurrence plots, which are used by physicists,
engineers and chemists, among other scientists,
for depicting patterns in time series of quantitative, continuous variables (like temperature, pressure, speed and so on) and are particularly useful
for discriminating among random, chaotic (ie
complex) and predictable series (eg Eckmann,
Kanphorst & Ruelle 1987).
Similarities, as provided by program RAP, are
values ranging from 0 (no similarity between the
two windows) to 1 (perfect similarity) and are
displayed as grey pixels in an image, the greater
the similarity the darker the pixel. Researchers
can then visually inspect the image and navigate
it with the mouse cursor on a computer screen in
order to explore its regions; this exploration
makes it possible to gain qualitative knowledge
on the basis of a graphical image obtained by
means of a quantitative technique.
Figure 2a shows a self-similarity map created
by RAP for the previous sequence of verbal interaction; the sequence is represented twice, top to
bottom and left to right, as indicated in the figure. The image was obtained by comparing every
code in the sequence with every other code; when
two codes are identical their similarity equals 1,
otherwise it equals 0 and thus a black pixel indicates that the code in the row is identical to the
INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES Volume 2, Issue 1, June 2008
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 25
Blending qualitative and quantitative analyses in observing interaction
one in the column of the image. That is the simplest representation, in which every successive
window contains just one code. This particular
map is in fact a dot plot, a representation of
sequence similarity that is mainly used in bioinformatics and text analysis for representing similarities within sequences of biomolecules and of
sections of documents, respectively (eg Church &
Helfman 1993). In a dot plot the main diagonal
represents the similarity of every code with itself,
which is obviously maximum and the image is
symmetrical around the diagonal. In this example, chequered regions indicate that a particular
chain of two codes repeats a number of times;
specifically, the 6 x 6 region in the upper left corner of Figure 2a corresponds to the six first codes
in the sequence, we he we he we he: dots in the
first row of that region indicate that code we
repeats at positions 1, 3 and 5, dots in the second
row indicate that code he repeats at positions 2, 4
and 6 and so on.
Figure 2b shows a self-similarity map of that
same sequence when successive windows containing not just one but three codes are applied. That
is, the first three overlapped windows contain:
[we he we] (sequence positions 1, 2 and 3), [he
we he] (2, 3 and 4) and [we he we] (3, 4 and 5).
Pixels in that image are not just black or white;
grey pixels indicate that certain windows have a
non perfect similarity (eg [ha wa hp] and [ha wa
ha]). Diagonal segments in the image, parallel to
the main diagonal, indicate that certain chains of
similar windows repeat in different parts of the
sequence. For example, the lower right quarter of
Figure 2b contains five diagonal black pixels close
to the main diagonal; they correspond to five successive windows of the section wa ha wa hp wa ha
wa hp wa ha wa in the last quarter of the
sequence, which starts at position 67 and ends at
position 77: window [wa ha wa] starting at 67, is
repeated at 71 and 75; window [ha wa hp] starting at 68, is repeated at 72 and 74 and so on.
Program RAP displays the contents of the windows when the mouse cursor is placed on the
pixel that represents their similarity.
Other hints about possible patterns are vertical
and horizontal lines, either continuous or fragmented, which indicate that a certain code (Figure 2a), or chain of codes (Figure 2b), occurring
in a position in the sequence repeats in several
other positions. For example, in Figure 2a a fragmented horizontal line composed of eight dots
can be seen in the upper left quarter of the image,
indicating that a certain code (hc) repeats eight
(B)
(A)
FIGURES 2 A
AND B:
T WO
S E L F -S I M I L A R I T Y M A P S O F T H E S E Q U E N C E S H O W N I N T H E T E X T W H E N S U C C E S S I V E
WINDOWS ARE ONE ( A ) AND THREE CODES ( B ).
Volume 2, Issue 1, June 2008 INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES
25
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 26
Augusto Gnisci, Roger Bakemanand Vicenç Quera
(A)
(B)
FIGURES 3 A
AND B:
S IMILARITY
M A P S FOR TWO H Y P O T H E T I C A L S E Q U E N C E S ( A A N D B ) , EACH ONE W I T H
W I N D O W S CONTA I N I N G O N E , T W O , T H R E E C O D E S .
times forward in the sequence, specifically in section hc wc hc wc hc wc hc wc hc wc hc wc hc wc
hc wc. On the other hand, horizontal or vertical
white lines (ie regions with no dots at all) would
indicate that a certain code, or chain of codes,
does not repeat at other positions in the sequence;
particularly, a horizontal white line as wide as the
image (except for the black diagonal pixel) would
show that the code, or chain of codes, corresponding to the row, is unique and never repeats.
As a whole, a similarity map reveals general
features of the sequences and can be useful for
classifying the interaction as patterned or random. Figure 3a shows three similarity maps (for
windows containing 1, 2 and 3 codes, respectively) for a hypothetical sequence of couple interaction in which husband responds to wife at
random and vice versa; Figure 3b shows three
similarity maps for another hypothetical sequence
containing long runs of reciprocal interactions,
specifically cross-approvals (wa ha wa ha…),
26
cross-empathising (we he we he…) and crosscomplaining, (wc hc wc hc…), which correspond
to three big chequered squares along the main
diagonal and indicate a development from positive to negative reciprocation. For a random
sequence, black pixels in the similarity map tend
to vanish rapidly as the window width increases,
while for a highly patterned sequence the proportion of black pixels remains more stable.
All these kinds of graphical representations
make it possible to describe bodies of data qualitatively and globally on the basis of quantitative
indices of similarity or distance among portions
of the data. Qualitative analysts may intervene
then at two levels: when they inspect and interpret the maps, providing interactional meanings
to the patterns identified (similarities, positive
and negative reciprocation and compensation,
and atypical sequences); and when they return to
the points of the interaction identified by means
of the maps and enrich their interpretations.
INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES Volume 2, Issue 1, June 2008
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 27
Blending qualitative and quantitative analyses in observing interaction
CONCLUSIONS
In this contribution we began by noting that traditionally qualitative analysis has served as a basic
resource for quantitative studies of social interaction and then discussed some possible sources of
misunderstandings between qualitative and quantitative approaches. Specifically, we discussed
some concerns qualitative analysts have about
quantitative research, particularly as concerns
‘natural’ units of analysis, use of ‘pre-defined’ categories, the notion of sequential context, the
multifaceted aspect of interaction and, finally, the
role of transcripts in observing and analysing
interaction. In the next few paragraphs, we briefly
provide some tentative conclusions and then
comment on the proposed research applications.
In general, there are desirable features that
qualitative analysts attribute to qualitative analysis and undesirable features that they claim characterise quantitative analysis. However, many of
the desirable features are not necessarily a prerogative of qualitative analysis alone. Attention to
participants’ point of view, natural units, grasping
the sequential context and the multifaceted
aspect of interaction, etc., can effectively be
realised and actually are, by quantitative approaches. Likewise, some alleged undesirable features
qualitative analysts attribute to quantitative
research, although they may apply in individual
cases, do not necessarily apply to the practice of
quantitative research itself (eg the ‘pre-defined’
nature of categories, the fixity of the category systems, the disregard of context, etc.). In sum,
desirable features are actually shared and undesirable features do not always apply.
Moreover, some criticised characteristics are not
necessarily negative. For example, the systematic
use of categories in itself or of non-participant-oriented categories is not necessarily a betrayal of
reality, as some qualitative analysts seem to imply.
First, non-participant-oriented category systems
may be useful precisely because they represent the
reality of interaction from an external point of
view, focusing on aspects not necessarily perceived
or expressed, verbalised or nonverbally, by partici-
pants. Second, we tend to believe that the task of
social psychology, or other disciplines where social
interaction is of concern, is to identify many
points of view for the same reality, identifying
their intersections, rather than focusing on one
view while disregarding others. Third, category
systems do not necessarily misrepresent reality;
they may even focus our attention on aspects of
the phenomenon considered crucial by careful
scholars before us. This seems a crucial task not
only when training observers or conceptualizing
research, but also when training professional people in communication skills (Argyle 1967), such
as teachers, politicians, social workers, on so on.
Based on these considerations, we tried to
show with concrete examples and applications
how a quantitative analysis of interaction can
focus subsequent qualitative analysis. Often qualitative analysis is regarded as having a preliminary
role with respect to quantitative one, but it can
also be the reverse.
We have shown, for example, that the analysis
of time series graphs allows us to inductively classify broad sequences of interaction or the interactions themselves between partners and how we are
thereby able to grasp and identify critical points of
interaction. Once quantitative analyses have identified them, qualitative analysis can further explicate them, thereby deepening our comprehension
of the dynamics at stake which in turn can lead to
new hypotheses and new insights.
The main application we presented regarded
identifying similarities between the same (or different) sequences of interaction by means of a
new program, RAP (Quera 2008). This application showed a possible way to identify orderly
and predictable patterns of interactional behaviors between interactants by means of two tools:
one is the summary index of similarity within the
same sequence or between sequences and the
other is the map. Graphical inspection of the
maps resulting from quantitative analysis demonstrated a powerful tool to understand whether
interactants responded to each other on a random
basis or whether there are chains of similar
Volume 2, Issue 1, June 2008 INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES
27
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 28
Augusto Gnisci, Roger Bakemanand Vicenç Quera
sequences that repeat in the same interaction.
The maps can also be used in many other ways,
depending on the interests of the researcher. For
example, a white horizontal line would identify a
unique sequence, one that does not repeat in any
other position within the sequence. Therefore,
inspection of similarity maps can also provide
what usually is considered a privilege of qualitative analysis, the identification of unique cases.
Again, we think that a qualitative analysis can
benefit from having identified similar chains in the
course of the interaction, thanks to the visual
inspection of the maps. From maps the analyst can
easily return to the original sequences of codes and,
more, to the original material (transcripts and
recordings) associated with these chains and exert
his/her critical capabilities on the crucial phases so
identified, the better to understand the interaction
at stake and/or to improve the tools with which
he/she observes reality, the category systems.
R e f e rences
Argyle M (1967) The Psychology of Interpersonal
Behavior. Penguin: Armondsworth.
Atkinson JM and Drew P (1979) Order in Court.
The organisation of verbal interaction in judicial
settings. Humanities Press: New Jersey.
Bakeman R (2000) Behavioral observation and
coding. In Reis HT and Judd CM(Eds)
Handbook of Research Methods in Social and
Personality Psychology (pp. 138–159) Cambridge
University Press: Cambridge.
Bakeman R and Brownlee JR (1980) The strategic
use of parallel play: A sequential analysis. Child
Development, 51: 873–878.
Bakeman R and Gnisci A (2005) Sequential
observational methods. In Eid M and Dieneer E
(Eds), Handbook of Multimethod Measurement in
Psychology (pp. 127–140) APA: Washington, DC.
Bakeman R and Gottman JM (1997) Observing
Interaction. An introduction to sequential analysis
(2nd edn) Cambridge University Press: New York.
Bakeman R and Quera V (1995a) Log-linear
approaches to lag-sequential analysis when
consecutive codes may and cannot repeat.
Psychological Bulletin, 118: 272–284.
Bakeman R and Quera V (1995b) Analyzing
Interaction. Sequential analysis with SDIS and
GSEQ. Cambridge University Press: New York.
28
Bakhtin M (1986) The problem of speech genres.
In Emerson C and Holquist M (Eds), Speech
Genres and Other Late Essays. University of Texas
Press: Austin, TX.
Bales RF (1950) Interaction Process Analysis. A
method for the study of small groups. AddisonWesley: Reading, MA.
Bavelas JB, Black A, Chovil N and Mullett J (1990)
Equivocal Communication Sage: Newbury Park, CA.
Bearison DJ and Dorval B (with LeBlanc G, Sadow
A and Plesa D and Commentary by Stetsenko A)
(2001) Collaborative Cognition: Children
negotiating ways of knowing. Ablex: Westport, CN.
Billig M (1999) Whose terms? Whose ordinariness?
Rhetoric and ideology in Conversation Analysis.
Discourse and Society, 10: 543–558.
Brown P and Levinson SC (1978) Universal in
language usage: Politeness phenomena. In Goody
E (Ed.) Questions and Politeness (pp. 56–310)
Cambridge University Press: Cambridge.
Brown P and Levinson SC (1987) Politeness: Some
universal in language use. Cambridge University
Press: Cambridge.
Bull P (1994) On identifying questions, replies and
non-replies in political interview. Journal of
Language and Social Psychology, 13: 115–131.
Bull P (2002) Communication Under the Microscope.
The theory and the practice of microanalysis.
Routledge: London.
Bull P (2003) The Microanalysis of Political Communication: Claptrap and ambiguity. Routledge: London.
Bull P and Elliott J (1998) Level of threat: Means of
assessing interviewer toughness and neutrality.
Journal of Language and Social Psychology, 17:
220–244.
Bull P and Mayer K (1993) How not to answer
questions in political interview. Political
Psychology, 4: 651–666.
Bull P, Elliott J, Palmer D and Walker L (1996)
Why politicians are three-faced: The face model
of political interviews. British Journal of Social
Psychology, 35: 267–284.
Church KW and Helfman JI (1993) Dotplot: A
program for exploring self-similarity in millions
of lines of text and code. Journal of Computational
and Graphical Statistics, 2: 153–174.
Dawson TL, Fisher KW and Stein Z (2006)
Reconsidering qualitative and quantitative
research approaches: A cognitive developmental
perspective. New Ideas in Psychology, 24: 229–239.
Drew P and Heritage J (1992) Analyzing talk at
work: An introduction. In P. Drew and J.
Heritage (Eds) Talk at Work. Interaction in
INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES Volume 2, Issue 1, June 2008
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 29
Blending qualitative and quantitative analyses in observing interaction
institutional settings (pp. 3–65) Cambridge
University Press: Cambridge.
Eckmann JP, Kamphorst SO and Ruelle D (1987)
Recurrence plots of dynamical systems.
Europhysics Letters, 5: 973–977.
Edwards D and Potter J (1992) Discursive
Psychology. SAGE: London.
Edwards D and Potter J (1993) Language and
causation: A discursive action model of description
and attribution. Psychological Review, 100: 23–41.
Edwards D and Potter J (1999) Language and
causal attribution: A rejoinder to Schmid and
Fiedler. Theory and Psychology 9: 823–836.
Eid M and Diener E (2006) Handbook of
Multimethod Measurement in Psychology. American
Psychological Association: Washington, DC.
Gnisci A (2005) Sequential strategies of accommodation: A new method in courtroom. British
Journal of Social Psychology, 44: 621–643.
Gnisci A (2008) Coercive and face-threatening
questions to left-wing and right-wing politicians
during two Italian broadcasts: Conversational
indexes of par conditio for democracy systems.
Journal of Applied Social Psychology, 38(5), in press.
Gnisci A and Bakeman R (2007) Sequential
accommodation of turn-taking and turn length:
A study of courtroom interaction. Journal of
Language and Social Psychology, 9: 234–259.
Goffman E (1955) On face-work: An analysis of
ritual elements in social interaction. Psychiatry,
18: 213–231. Reprinted in (1967) Interaction
Ritual: Essays on face to face behavior (pp. 5–45)
Anchor: Garden City, NY.
Goffman E (1971) Relations in Public. Microstudies
of the public order. Penguin Press: London.
Goodwin C and Duranti A (Eds) (1992) Rethinking
Context. Language as an interactive phenomenon.
Cambridge University Press: Cambridge.
Gottman JM (1979) Marital Interaction. Experimental
investigations. Academic Press: New York, NY.
Gottman JM (1983) How children becomes
friends. Monographs of the Society for Research in
Child Development, 48(3).
Harré R (1979) Social Being. Blackwell: Oxford.
Harré R and Gillett G (1994) The Discursive Mind.
Sage: Thousand Oaks CA.
Harré R and Secord P (1972) The Explanation of
Social Behavior. Basil Blackwell: Oxford.
Heritage J (1984) Garfinkel and Ethnomethodology.
Polity: Cambridge.
Heyns RW and Lippitt R (1954) Systematic
observational techniques. In Handbook of Social
Psychology. Theory and method (Chap. 10), Vol. 1.
Addison-Wesley: London.
Jacobs JK, Kawanaba T and Stigler JW (1999)
Integrating qualitative and quantitative
approaches to the analysis of video data on
classroom teaching. International Journal of
Educational Research, 31: 717–724.
Jefferson G (1984) On the organisation of laughter
in talk about troubles. In Atkinson JM and
Heritage JC (Eds) Structures of Social Action:
Studies in conversation analysis (pp. 347–369)
Cambridge University Press: Cambridge.
Jefferson G (1985) On the interactional unpackaging
of a ‘gloss’. Language and Society, 14: 435– 466.
Jucker J (1986) News Interviews: A Pragmalinguistic
Analysis. Gieben: Amsterdam.
Kendon A (1982) Organization of behaviour in face
to face interaction. In Scherer KR and Ekman P
(Eds) Handbook of methods in nonverbal
behaviour research, 26: 22–63.
Lewin K (1938) The conceptual representation and
measurement of psychological forces.
Contributions to Psychological Theory, 1.
Lewin K (1944) Constructs in psychology and
psychological ecology. University of Iowa Studies
in Child Welfare 20: 17–21.
Linell P, Gustavsson L and Juvonen P (1988)
Interactional dominance in dyadic
communication: a presentation of iniziativeresponse analysis. Linguistic, 26: 415.
Mey I (2001) The CA/CDA controversy. Journal of
Pragmatics, 33: 609–615.
Mooney CG (2000) Theories of Childhood: An
Introduction to Dewey, Montessori, Erikson,
Piaget and Vygotsky. Redealf: New York.
Muntigl P and Turnbull W (1998) Conversational
structure and facework in arguing. Journal of
Pragmatics, 29: 225–256.
O’Barr WM and Lind EA (1981) Ethnography and
experimentation-Partners in legal research. In
Sales BD (Ed.) Perspectives in Law and
Psychology: The trial process (pp. 181–208)
Plenum Press: New York, NY.
O’Barr WM (1982) Linguistic Evidence: Language
power and strategy in the courtroom. Academic
Press: New York, NY.
Parten MB (1932) Social participation among
preschool children. Journal of Abnormal and
Social Psychology, 27: 243–269.
Peräkylä A (2004) Two traditions of interaction
research. British Journal of Social Psychology, 43:
1–20.
Psathas G (1995) Conversation Analysis: The study of
talk-in-interaction. Sage: Thousand Oaks CA.
Volume 2, Issue 1, June 2008 INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES
29
IJMRA_2_1_Internal
19/6/08
3:38 PM
Page 30
Augusto Gnisci, Roger Bakemanand Vicenç Quera
Psathas G and Anderson T (1990) The ‘practices’ of
transcription in conversation analysis. Semiotica,
78: 75–99.
Quera V (2008) RAP: A computer program for
exploring similarities in behavior sequences using
random projections. Behavior Research Methods,
40: 21–32.
Quera V, Bakeman R and Gnisci A (2007)
Observer agreement for event sequences:
Methods and software for sequence alignment
and reliability estimates. Behavior Research
Methods, 39: 39–49.
Reis HT and Judd CM (Eds) (2000) Handbook of
Research Methods in Social and Personality Psychology.
Cambridge University Press: Cambridge.
Roger DB, Bull PE and Smith S (1988) The
development of a comprehensive system for
classifying interruptions. Journal of Language and
Social Psychology, 7: 27–34.
Rosenblum L (1978) The creation of a behavioral
taxonomy. In Sackets GP (Ed.) Observing Behavior
(vol. 2): Data collection and analysis methods (pp.
15–24) University Park Press: Baltimore.
Sacks H, Schegloff E and Jefferson G (1974) A
simplest systematics for the organisation of turntaking for conversation. Language, 4: 696–735.
Schegloff EA (1993) Reflections on quantification
in the study of conversation. Research on
Language and Social Interaction, 26: 99–128.
Schegloff EA (1997) Whose text? Whose context?
Discourse and Society, 8: 165–187.
Semin GR and Fiedler K (1988) The cognitive
functions of linguistic categories in describing
persons: Social cognition and language. Journal
of Personality and Social Psychology, 54: 558–568.
Semin GR and Fiedler K (1991) The linguistic
category model, its bases, applications and range.
In Stroebe W and Hewstone M (Eds) European
Review of Social Psychology, 2: 1–50.
Smith ML (1994) Qualitative plus/versus quantitative:
The last word. In Reichardt CS and Rallis SF (Eds)
The Qualitative-Quantitative Debate: New
perspectives (pp. 37–44) Jossey-Bass: San Francisco.
Smith PK (1978) A longitudinal study of social
participation in preschool children: Solitary and
parallel play reexamined. Developmental
Psychology, 14: 517–523.
Stevens SS (1946) On the theory of scales of
measurement. Science, 103: 677–680.
Received 14 November 2007
Accepted 31 March 2008
RECENT TITLES FROM POSTPRESSED
Following Vygotsky to A Lear ner-Centred School
by Robert G Grandin ISBN 1-876682-96-5
This book illustrates the author's passion for children
and the belief that they are all curious individuals who
want to learn. It outlines the author's experiences in
addressing the problem of individuality in learning,
his search for knowledge, reorganisation of schooling
structures and discovery of real ways to support
teachers as they attempt this most difficult of
challenges. With a greater diversity of learners
remaining within the schooling system, engaging
them in learning and recognising their achievements
will be aided by these recommendations.
Enhancing Lear ning and Teaching: Pedagogy,
Technology and Language
Jeong-Bae Son and Shirley O'Neill (eds)
ISBN 1-876682-85-X
The knowledge economy has raised questions about
the effectiveness of traditional approaches to teaching
and learning, in a collaborative and interactive community. This book provides an entrée into developments in
postmodern pedagogies in terms of approaches,
emerging concepts about learners and learning, the
use of information and communication technologies
(ICTs) and the social construction of knowledge
through networked and e-learning.
Pedagogies: Stor ylines and Stor yspaces
by Jennifer M. Nayler ISBN 1-876682-74-4
This book both focuses on the power of narrative in professional development and also by the way author links
the narratives that “call us into being” to readers own
concrete personal/professional experiences and inviting
them to act. Framed by a commitment to teaching in
socially-just ways, this resource encourages teachers to
theorise their practices and experiences, and to articulate them in relation to the author's learning journey.
D o c t r i n a P e r p e t u a : B ro k e r i n g c h a n g e , p ro m o t i n g
i n n o v a t i o n a n d t r a n s f o r m i n g m a rg i n a l i s a t i o n i n
u n i v e r s i t y l e a r ning and teaching
Jeanne McConachie, Bobby Harreveld, Jo Luck, Fons
Nouwens and Patrick A Danaher (eds)
ISBN 1-876682-93-0
A case study of a regional university, this book provides
insights into university learning and teaching that are
grounded in experience and conceptual rigour.
eContent Management Pty Ltd, PO Box 1027, Maleny QLD 4552, Australia
Tel.: +61-7-5435-2900; Fax. +61-7-5435-2911; [email protected]
www.e-contentmanagement.com
30
INTERNATIONAL JOURNAL OF MULTIPLE RESEARCH APPROACHES Volume 2, Issue 1, June 2008