Upload
rachel-weinstein
View
218
Download
0
Embed Size (px)
Citation preview
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 1/56
The affect structure revisited 1
The affect structure revisited
Véronique Elefant-Yanni and Maria-Pia Victoria-Feser1
University of Geneva
1 Both author acknowledge the support of the Swiss national Science Foudation, grant no PP001—106465.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 2/56
The affect structure revisited 2
Abstract
In affective psychology, there is a persistent controversy about the number, the nature and the
definition of the affect structure dimensions. Responding to the methodological criticisms
addressed to the preceding studies, we conciliated the principal theories regarding the affect
structure with the same experimental setting. In particular, using the semantic items all
around the circumplex we found three bipolar independent dimensions and using only the
PANAS semantic items, we found two unipolar dimensions. Finally, we propose a heuristic
theorization of affect based on a current firmly established in social sciences, coherent from
semantics to sociology, but largely ignored by researchers in affective psychology , that
allows to postulate that affect is all at once cognition, motivation and behaviour.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 3/56
The affect structure revisited 3
Introduction
Over the last years a long debate took place regarding the affect structure. The affect
is the momentary feeling a person has at any time that is induced by the situation as a whole,
including internal and external stimuli. It has to be distinguished from mood and emotion
which are affective phenomena of another nature. The affect structure is commonly described
by means of different dimensions such as valence or evaluation, activation, tension or
potency. These dimensions are either evaluated on a continuum (bipolar dimension) or
separated into a positive and negative directions (unipolar dimension). There is however a
current controversy about the number, the nature and the definition of the affect structure
dimensions. If a consensus was recently established (Feldman-Barrett & Russell 1998,
Watson & Tellegen 1999) to recognize a two-dimensional circumplex affect structure
between the two principal currents of research carried out on the one hand by Russell
(Russell 1980) and on the other hand by Watson (Watson & Tellegen 1985), they still
disagree on the uni- or bipolarity and the nature of the two dimensions. Since several authors
claim that the underlying reason for the existence of this controversy lies in the
methodological aspects of the research plans, the aim will then be to highlight the various
difficulties in implementing research on affect structure, and provide a complete study that
will show that depending on the experimental plan, the analysis results will support with the
same statistical method one or another of the leading theories. In addition, a persisting
alternative current of research postulates an affect structure with three correlated dimensions
(Sjösberg, Svensson & Persson 1979). Lastly, a current firmly established in social sciences,
coherent from semantics to sociology, supports an affect structure with three independent and
bipolar dimensions. The results of the present research are perfectly integrated in this current,
generally ignored by researchers in affective psychology, and allow to postulate that affect is
all at once cognition, motivation and behavior.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 4/56
The affect structure revisited 4
The first studies regarding the affect structure took place by means of introspections
and were made by Wundt, Stumpf and Titchener (see Reisenzein 1992, Reisenzein &
Schonpflug 1992). They focused on the bipolar dimension of valence (pleasure/displeasure)
that they thought to be the base of the conscience. On the other hand, Cannon (1927)
emphasizes the activation dimension (see also Lindsley 1951, Hebb 1955, Duffy 1957,
Berlyne 1960, Schachter & Singer 1962, Zillman 1983, Mandler 1984, Thayer 1989, 1996).
From these two first currents a third approach was born that includes two independent,
bipolar and of same magnitude dimensions of valence and activation (Russell 1978, 1980,
Lang 1978, 1994, Larsen & Diener 1992, Lang, Bradley & Cuthbert 1992, Bradley 1994).
According to these authors, the valence and activation dimensions define a circumplex on
which all affects can be represented (see also Schlosberg 1952). This representation takes
over the circumplex semantic representation for two of the three dimensions of connotative
meaning (i.e. evaluation and activity) found by Osgood to underlie language (Osgood & Suci
1955, Osgood, Suci & Tannenbaum 1957, Osgood 1966, Osgood, May and Miron 1975).
Nevertheless the bipolarity hypothesis of the valence was disputed by the hypothesis of
unipolarity (Nowlis & Nowlis 1956, Bradburn 1969, Bradburn & Caplovitz 1965).
Researches on mood scales seem to confirm this unipolarity hypothesis (Borgatta 1961,
McNair & Lorr 1964, Thayer 1967, Warr, Barter & Brownridge 1983, Watson & Tellegen
1985). However, Cacioppo and Berntson (1994) pointed out that an important part of the
literature regarding the unipolarity had developed because experiments were based on the
PANAS (Positive And Negative Affect Schedule, Watson, Clark & Tellegen 1988). PANAS
is a questionnaire that measures the affective experiences by means of 20 semantic items
chosen to correspond to a high degree of activation and pleasure, and to a high degree of
activation and displeasure (Watson & Tellengen 1985). Actually, the resulting circumplex
model based on the PANAS has the two orthogonal dimensions of Positive Affect (PA) and
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 5/56
The affect structure revisited 5
Negative Affect (NA), and it should be specified that for the authors “these two factors have
been characterized as “descriptively bipolar but affectively unipolar dimensions” (Zevon &
Tellegen 1982, p. 112) to emphasize that only the high end of each dimension represents a
state of emotional arousal (or high affect), whereas the low end of each factor is most clearly
and strongly defined by terms reflecting a relative absence of affective involvement…”
(Watson & Tellegen 1985, p. 221). However, while some authors accumulate evidence in
favour of unipolarity (Tellengen 1985, Mayer & Gaschke 1988, Meyer & Shack 1989, Morris
1989, Watson & Clark 1997), others defend the bipolarity (Russell & Mehrabian 1977, Lorr
& Wunderlich 1980, Lorr & McNair 1982, Russell & Steiger 1982, Diener & Emmons 1984,
Larsen & Diener 1992, Ortony, Clore & Collins 1988, Russell 1989, Lang, Greenwald,
Bradley & Hamm 1993, Reisenzein 1994, Feldman 1995). As when rotated of 45° the PA
and NA factors correspond to the valence and activation dimensions of Russell’s circumplex
model, Watson and Tellegen “suggest a simpler explanation, namely that the extent of their
relations varies with the terms used to construct the Positive and Negative Affect measures”
(1985, p. 233). If their “parsimonious” model corresponded to the true affect structure than
the choice of more indicators representing more affects facets by the use of the entire
semantic circumplex should lead to the same dimensional affect structure. In the present case,
it will be showed however that the too restrictive choice of semantic items consists of a
methodological bias. Watson and Clark (1994) have more recently proposed the PANAS-X
which expanded the first version of PANAS and used 60 semantic terms to assess specific
emotions or enduring affective states.
On the other hand, some researchers in the psychology of personality domain proposed
an affect model with three slightly correlated dimensions (Sjösberg, Svensson & Persson
1979, Matthews, Jones & Chamberlain 1990, Steyer, Schwenkmezger, Notz & Eid 1994,
Schimmak & Grob 2000). For example, Matthews, Jones and Chamberlain found the three
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 6/56
The affect structure revisited 6
dimensions as being pleasant-unpleasant (valence), energetic arousal and tense arousal.
Energetic arousal contrasts feelings of vigour and energy with tiredness and fatigue and can
be considered as an activation dimension, whereas tense arousal contrasts tension and
nervousness with relaxation and calmness and can be associated to a tension or potency
dimension. The resulting model is different from Russell’s model not only because one
dimension has been added, but also because the valence is correlated with the activation
dimension. Wundt (1896) already postulated an affect model with three dimensions pleasant-
unpleasant, low-high activated and tense-relaxed and Thayer (1978) showed that participants
were able to subjectively discriminate energetic arousal and tense arousal. It should be noted
that the research on affect structure in the psychology of personality is relatively recent and
needs to be distinguished from the numerous studies on the structure of personality,
personality trait and mood which concerns the enduring dispositions of the participants. For
long the “Gigantic three” dimensions of Eysenck’s model of personality (1991, neuroticism,
extraversion, psychoticism) dominated, but recently a consensus has been achieved about a
five independent dimensions model of which most known is Costa and McCrae’s “Big Five”
model of personality (1992, neuroticism, extraversion, openness, agreeableness,
conscientiousness). In both theoretical frameworks, the associations between neuroticism,
extraversion, and bad mood, good mood are highlighted by various authors (Watson & Clark
1992, Tellegen 1985, Emmons & Diener 1986, Watson, Clark & Tellegen 1988, Thayer
1989, Williams 1989, Matthews, Jones & Chamberlain 1990, Costa & McCrae 1992) and
tend to accredit that for mood the negative and positive dimensions are indeed unipolar and
independent.
In another domain, the three dimensions of valence, activation and tension have been
considered for a long time now. In linguistics, Osgood and colleagues (Osgood & Suci 1955,
Osgood, Suci & Tannenbaum 1957) have developed “semantic differential” bipolar scales
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 7/56
The affect structure revisited 7
based on semantic opposites such as good/bad, strong/week, active/passive and so on, in
order to measure the attitude of participants regarding symbolic concepts represented by
words. These authors were interested in the connotative meanings of words as apposed to
their denotative meanings. The denotative usage gives to things represented by words their
essential and objective meanings as in dictionaries. On the other hand, the connotative usage
presents things enriched by their affective associations which, though intangible, are
nonetheless real. These authors highlight a semantic structure based on the three dimensions
of Evaluation, Potency and Activity, known as EPA. Osgood has presented convincing
evidence that this simple structure is a universal characteristic of human thinking and is
exhibited by all of the cultural groups and languages (Osgood 1960, Osgood, May & Miron
1975). It should be noted that semantic differential scales actually measure the connotative
meaning of symbolic concepts by examining how people "feel" about, that is by measuring
their induced affects.
Largely inspired by Osgood's cross-cultural research, Heise (1979) created his Affect
Control Theory (ACT). The ACT postulates that all individuals evolving in any situation
evaluate the situation according to the affective meanings of all its elements, which,
combined, produce a global impression that gives a meaning to the situation. This last
meaning is transient since the affective meanings of the situation elements change while the
situation progresses. The meaning of the situation is compared by the individual to the
expected sentiment that his social group attributes to that situation (Heise 1966). Usual
situations produce transient impressions that match expected sentiments, whereas situations
that generate impressions deviating widely from those sentiments seem abnormal (Heise &
MacKinnon 1987). According to the ACT, people manage situations so as to match transient
impressions with their expected sentiments and to maintain normality in their experiences.
For Heise (1979), since affects reflect the impressions that the situation in progress generates,
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 8/56
The affect structure revisited 8
they are a way of directly sensing the kinds of social identities people are taking, the
operative social structure in the situation and also a way of sensing in which direction the
situation is evolving according to its expected sentiment. Given the central importance of
affect in his theory, it was quite naturally that Heise got interested in the controversy about
the affect structure (Morgan & Heise 1988), but none of his results conformed to a
circumplex model of affect structure. On the contrary, his results highlighted that the affects
"array themselves naturally in a three-dimensional space involving evaluation, activation and
potency" (p. 27). He concluded that "the affective dimensions correspond to basic mental
processes […] and that much of people's cognitive information concerning emotions is
generated within the dimensional framework" (p. 29).
Therefore, since the different models of affect structure disagree on the number of
dimensions, their bipolar or unipolar nature, their definitions, and since the methodology to
unveil them differ sometimes substantially, we investigate the different experimental plans
used until now and analyze in this paper the important aspects on which they differ, their
potential influence on the empirical construction of the affect structure which has led to the
present controversy.
Important aspects of the experimental plan
Type of affective phenomena
Many authors view the distinction pleasant-unpleasant (i.e. valence dimension) as the
primary criteria that discriminates affective states in self-reported affective experiences
(Wundt 1896, 1906, Ortony & al. 1988, Schimmack & Reizenzein 1997, Shaver & al. 1987).
When we ask people to answer about their mood or more globally about how they feel on a
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 9/56
The affect structure revisited 9
relatively long period of time, they aggregate momentary feelings in their minds (Thomas &
Diener 1990, Parkinson, Briner, Reynolds & Totterdell 1995, Schimmack & Reisenzein
1997, Reisenzein & Schimmack 1999). As valence is the salient criteria of discrimination,
they differentiate the global amount of “pleasant state” from the global amount of
“unpleasant state”, and evaluate simultaneously how good they felt as well as how bad they
felt at the precise moment when they are asked. If time is given to recall information from the
episodic memory, participants are surely able to remember a bad or good feeling at a “special
moment” in which they were globally in a good or bad mood. The contrast with our mood in
background is precisely why those moments are specially remembered. This pattern of
outcome has led to the view that affective states are supported by two orthogonal unipolar
dimensions, namely positive affect and negative affect (Watson, Clark & Tellegen 1988).
However a participant will deny being able to be at the same time cheerful and sad, because
in that case, no confusion is made about the moment at which the feeling occurs, i.e. at
exactly the same point in time (Diener & Emmons 1984, Zautra, Potter & Reich 1997,
Russell 2003). Since by definition affective experiences are temporary states that change
from moment to moment, the main problem in determining these states resides in the
definition of “moment”. There is an agreement to differentiate trait affect and state affect, in
that a distinction is made between mood and emotion as in common language (Diener & Iran-
Nejad 1986, Reisenzein & Schönpflug 1992, Fridja 1993, Clore & al. 1994). However for the
valence dimension in affect structure, we must first address the questions of the duration and
definition of the different affective states. No consensual taxonomy has been reached until
now, and a good share of the debate is due to the difficulty in distinguishing clearly between
various types of affective phenomena (Scherer 2000). We propose here to consider three of
them, namely emotion, mood, and affect.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 10/56
The affect structure revisited 10
The emotion consists of a relatively short episode of synchronized response by several
or all subsystems of the organism at a time of evaluation of an external or internal event
having a major importance for the participant. Emotions are, for example, anger, sadness, joy,
fear, shame, pride, exaltation, despair. Regarding emotion, the principal discussion between
researchers is about the changes in different modalities that are necessary and sufficient
components of an emotional episode. There is a consensus on the "reactional triad" of the
emotion that is supposed to be composed at least of physiological activation, motor
expression (vocal, facial and body) and participative feeling. Some authors also include
motivation and cognitive evaluation (Buck 1985, 1993, Frijda 1986, 1987, Scherer 1984 a &
b, 1993). Although we still often speak about "emotional state" suggesting a relative stability
in time, the authors seem to consider an emotion as a process implying fast changes in the
duration of an episode.
The affect consists of an elementary process accessible to conscience in terms of
valence, activation and tension (Wundt 1896, Thayer 1978). It corresponds, for a part, to the
component of the subjective experience of emotion (Scherer 1993, Russell 2003), but it is not
necessarily directed by a precise event and in course of time it is always present, fluctuating
for reasons we are not always aware of, like for instance weather, circadian cycle (Thayer
1989, Watson, Wiese, Vaidya & Tellegen 1999). The affect varies in intensity throughout the
day, it is particularly salient when its degree of activation is rather low or high into positive or
negative, but with an average degree of activation it has tendency to disappear in the back of
our consciousness. Perhaps also it remains generally unconscious because it exists first and
mainly at the present time to which we do not have access. And it is only by a work of self-
listening that we get aware of what we just felt. For example, it is this affective phenomenon
that is in question when with the alarm clock one feels depressed, tired, or merry. The affect
is also called activation by Thayer (1986), felt, feeling, sentiments or emotions by other
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 11/56
The affect structure revisited 11
authors. It is difficult to be consistent in the denomination because the various authors use the
same terms to indicate affective phenomena of different nature.
Although sometimes also mistaken as affect, the mood is distinguished from affect by
its duration and the fact that it can be without apparent cause (Russell & Feldman Barrett
1999). It consists of an affective diffuse state of relatively low intensity that acts as a
background on which are played the fluctuations of affect. Thus a participant can be basically
happy but for example feels temporarily sad with the listening of a melancholic melody. It is
this affective phenomenon that is in question in mood disorder, for example dysthymia,
characterised by a lack of enjoyment/pleasure in life that continues for at least six months
(APA, DSM-IV 1994).
Hence, in experimental plans, when studying affect structure, it is crucial that the
participants receive clear instructions about what type of affective phenomena they should
describe. Indeed, we observed that the participants use the semantic scales in ways suggesting
they have frames of reference based upon a lifetime of making such judgments about what
they feel. However, mood has undoubtedly an impact on the coloration and intensity of
affects, and therefore they are difficult to discriminate. If the period during which the
participant has to evaluate what he/she feels is long, the participant could be tempted to retain
only what showed invariant over this period and to answer according to his mood rather than
according to his affect. On the other hand, the participant could answer according to his
different affects that followed one another by melting them together. Diener and Emmons
(1984) thus concluded that the valence dimension was bipolar when it was examined in a
short period of time, but that positive and negative affects seemed independent when they
were examined over an extended time period (Warr & al. 1983, Gotlib & Meyer 1986).
Therefore, a suitable way to distinguish between mood, emotion and affect, is to create a
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 12/56
The affect structure revisited 12
situation in which the participant is placed and his affective experience is immediately
recorded. This is what we will do in our experimental plan.
Choice of semantic items
The hypothesis of bipolarity comes mainly from the semantic representation of items
relating to affect. The concepts of joy and sadness appear to us like bipolar opposites
primarily because merry and sad are regarded as antonyms. The semantic differentiation
studies of Osgood (1969) as well as the interlinguistic comparisons (Russell, Lewicka & Niit
1989, Russell 1991) support the hypothesis of bipolarity. To test this assumption empirically,
it is thus initially necessary to specify a model that is bipolar, i.e. selected semantic items
must be antonyms or at least be semantically interpreted as bipolar opposites. However, all
terms related to positive affects are not bipolar opposites of each term related to negative
affects. Thus we regard sad as opposite to merry rather than tense, and tense as opposed to
relax rather than sad. The terms can have between them more complex degrees of relation
than only antonym, synonym or independent. The operationalization of affect bipolarity has
to go through the selection of terms that are bipolar opposites on valence but also on
activation as we now have a consensus that it constitutes a second independent dimension
(Bush 1973, Averill 1975, Neufeld 1975, 1976, Russell 1978, Whissell 1981, Thayer 1989).
It is also necessary to ensure that the selected terms adequately come from the entire range of
pleasant or unpleasant affects descriptors. Otherwise, the retained semantic items would not
be representative of the entire affect structure, and consequently bipolarity could artificially
not be represented. Currently, a majority of authors recognize a semantic representation of
items related to affect under the form of a circular structure (circumplex structure) in which
the horizontal diameter corresponds to the bipolar dimension of valence and the vertical
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 13/56
The affect structure revisited 13
diameter corresponds to the bipolar dimension of activation. The semantic items are placed
on the circle according to theirs scores in valence and activation obtained on the basis of
empirical research (Russell & Feldman Barrett 1999). The antonyms are then diametrically
opposite. More generally, bipolar opposites are theoretically separated by angles of 180°,
whereas synonyms are angularly very close. The number of clusters of semantic items that
we can define is arbitrary, but it is important to note that choosing to work with items from
some sections rather than the whole circumference can lead to an affect structure that is
specific to the selected items.
By selecting a cluster of items without selecting items that are diametrically opposed
on the circumplex structure, we exclude the semantic items that are the bipolar opposites. By
doing so, the bipolarity we seek to test is excluded. Therefore several authors questioned the
use of the PANAS as able to adequately measure affects (Morris 1989, Larsen & Diener
1992, Green & Al. 1993, Nemanick & Munz 1994, Carver 1996). Indeed, Watson and his
colleagues chose their positive semantic items according to a theoretical dimension defined
by high degree of activation and pleasure, and their negative semantic items according to a
theoretical dimension defined by high degree of activation and displeasure. Since the
semantic items are not opposed on activation, the resulting model excludes in fact bipolarity.
Watson’s clusters correspond to two adjacent quadrants in the circumplex where positive
items are mirrored into their negative counterparts compared to valence dimension. Therefore
it is not surprising that Watson found correlations close to zero between what he had named
"positive affect" and "negative affect". In our research we will supplement the PANAS
semantic items to include all the bipolar opposites so as to cover the entire range of activation
of positive and negative affects.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 14/56
The affect structure revisited 14
The affect space
The choice of semantic items is not the unique issue in setting an experiment plan that
is able to uncover the affect structure. But, it may give an idea of the hierarchical bond
between affect space and semantic space in which participants draw to formulate their
answers based on introspection of what they felt. Indeed, the semantic items are used by
participants to describe what they just felt but are not pure inner feelings states in that they
only reflect some particular aspects of an affective experience (Feldman 1995). Therefore, in
order to unveil the affect structure, one must have a substantial number of different situations
inducing different affects (observations) from the entire affect space. In the literature,
experiments on self-reported affect experiences often ignore to take into account the
variability of the situations inducing the different affects, by for example asking the
participants to remember a personal situation. Although in these situations one could expect
to have a large enough spectrum of different affects, it is well know that the experimental
condition itself will influence the participant in his choice of affect (Tulving & Thomson
1973, Golden & Baddeley 1975, Tulving 1979, Baddeley 1982, Tulving 1983), and since the
experimental condition is the same for all participants, there is a doubt that all the reported
affects reflect the entire affect space. The consequence is that the unveiled affect structure
can be biased in the direction of fewer dimensions. To ensure that the observations reflect the
entire affect space, the participants must be put into a given situation and the set of all
situations must potentially produce all kinds of affects. This can be done using for example
the International Affective Picture System (IAPS, Lang, Bradley & Cuthbert 1995) as we will
develop it later.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 15/56
The affect structure revisited 15
The response format
To measure each semantic item one can use different scales, bipolar and unipolar
ones. In the first case, the graduated scale has at its ends one semantic item of the bipolar
opposites pair and in its medium the neutral point (zero). In the second case, the two semantic
items of the bipolar opposites pair are measured separately on two graduated scales having
each one a neutral point (zero) for the situations where the participants consider the
corresponding item not relevant to qualify their affects. Between these two formats, there is a
whole range of rather bipolar or rather unipolar ambiguous scales. Since Russell and Carroll
(1999) showed that in these ambiguous cases participants subjectively rebuild unipolar scales
in bipolar scales, it is essential to measure semantic items in strictly unipolar scales in order
not to force bipolarity artificially.
Measurement errors
Another potential source of bias in the evaluation of affect structure is the
measurement error due to the participant’s response styles on likert scales, a longstanding
methodological concern (Bentler 1969). Using multitrait-multimethod analyses on repeated
measurements, several authors concluded that there is little if not no difference in the
estimated affect structure compared to the one unveiled when using single measurement per
participant (see e.g. Green & al. 1999). Watson and Clark (1997) and Russell and Carroll
(1999) reached the same conclusion in a different manner. Moreover, it can be argued that
even if there exists a source of variability due to the participant’s response style, there is also
a confounded variability due to the type of affect induced by the situation (for example a
picture). The participants will especially respond differently according to the unicity of their
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 16/56
The affect structure revisited 16
own background, i.e. describe different affects when presented with the same situation.
Judgment of a particular stimulus on a series of semantic scales is really a comparative
judgment against a multitude of previous stumuli scales allocations. Although it is not easily
measurable, it is reasonable to suppose that this second source of variability is much larger
than the one induced by response style, and since it is confounded with the variability of the
affect itself, i.e. the observation, it shouldn’t bias the results.
In summary, the experimental plan has to pay a detailed attention so that the
investigated affective phenomenon is clearly definite for the participants, their affective
experiences are immediately recorded, the indicators represent the whole circumference of
the semantic circumplex and include the semantic bipolar opposites to allow to test the
bipolarity of affect dimensions, the variability of the induced affects is sufficient to highlight
the affect structure, and finally the graduated scales are strictly unipolar not to force
bipolarity artificially. Those important methodological aspects taken care of, we can now
operationalize the question about the number, nature and definition of the affect structure
dimensions.
Method
Population
A total population of 162 women (age 18-56, mean 24.3) participated in this research.
Most of them (128) were second year students in the section of psychology enrolled for
course credit. The remaining women responded to an advertisement and participated on a
voluntary basis. We did not retain male participants to avoid an increase of variability due to
the difference in response between gender when the affect is induced by pictures (Lang,
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 17/56
The affect structure revisited 17
Bradley & Cuthbert 1995). The front page of the questionnaire collected various information
on the participants that was also used to form a code and identify participants while
preserving their anonymity (more details on the sample upon request to the authors).
Material
We use a series of 604 pictures elaborated and digitized by the Center for the Study of
Emotion and Attention of the University of Florida (1995). This series of pictures is known
under the name of International Affective Picture System (IAPS) and was already largely
used in studies on affect (see e.g. Lang, Greenwald, Bradley & Hamm 1993, Spence, Shapiro
& Zaidel 1996, Vanman, Boehmelt, Dawson & Schell 1996). The various pictures were
subjectively evaluated in term of valence, activation and of control by the participants of
Lang (Lang, Bradley & Cuthbert 1995). We chose 162 pairs of pictures that are close on their
rating on valence as well as on activation. The chosen pairs are distributed on the whole
space so that affects induced by the pictures are varied., i.e. spread over the entire latent
space. We also chose one pair and a picture that are used respectively as trial pictures and as
final picture after the experimental pictures.
Experimental plan
Each participant was submitted to the experiment twice, at a week of interval. One of
the time chosen randomly, the participant had to respond on strictly unipolar likert scales, the
other time on strictly unipolar continuous scales. The aim was to investigate if the type of
scale (likert or continuous) had an influence in the estimation of the affect structure. Each
time, the participant had to describe her affects induced by 4 pictures (scenes): the trial
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 18/56
The affect structure revisited 18
picture, two experimental pictures and the final picture. The two experimental pictures (2nd
and 3rd
scenes) were different between the two times, but belong to the same pairs, i.e. they
are matched on valence and activation as given in the IAPS. The choice of the pairs and the
order of the pictures within the pair and between the times were chosen randomly. We chose
to use two experimental pairs of pictures in order to control for the possible variability due to
the style of response and/or the interaction between the participant and the induced affect. In
order to have enough pairs, we “roled them over” the participants, in that the second
experimental pair of the ith participant was used as the first experimental pair for the (i+1)th
participant. Soon after the participant has seen a picture for 10 seconds (time calibrated on the
basis of a preliminary study), she was asked to evaluate her affect using a questionnaire
containing various semantic items presented in the following paragraph.
Semantic indicators
As there is not a consensual list of semantic items, we constructed one by taking care
that the entire circumplex structure is represented and that each item has its bipolar opposite.
We started with semantic items that have been used in similar studies. We also included the
semantic items of the PANAS and for each one we added, if necessary, its bipolar opposite
on valence and on activation. Once the list of the semantic items in English was built, it was
translated by four researchers using this time the antonyms and synonyms in French
according to the method of back-translation (see Appendix). The French translation of the
PANAS items used in our research were mainly taken from Lapierre, Gaudreau & Cableway
(1999). The resulting list includes twenty semantic pairs of bipolar opposites terms that we
presented on a circumplex structure on Figure 1. In the questionnaire, the two semantic items
of the same pair of bipolar opposites were presented on the same line side-by-side in order to
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 19/56
The affect structure revisited 19
prevent the participants to rebuild subjectively unipolar scales in bipolar scales (Russell &
Carroll, 1999). The presentation of the items was controlled to avoid a positive or a negative
coloration of the questionnaire.
Results
Data analysis method
With the data collected on the likert scales, we estimated polychoric correlations
matrices (Olsson 1979) as input for an exploratory factor analysis. We used two different
software and methods that are in some sense equivalent (see Jöreskog 2003) to estimate the
factor loadings. In LISREL 8.54 (Jöreskog & Sörbom 2003), more specifically in PRELIS
2.54, we carried out exploratory factor analysis using the estimation method MINRES
(MINinmum RESiduals) that was first introduced by Harman (1960) and is based on the
direct minimization of least squares to fit a matrix of factor loadings to the correlation matrix.
The MINRES method can be used with small samples even when the number of variables is
large and when the correlation matrix is not positive definite as it might be the case with
polychoric correlation matrix. In M plus 3.0 (Muthén & Muthén 1998-2004), we used the
ULS method (Unweighted Least Squares) of Jöreskog and Sörbom (1977) which is based on
eigenvalues and eigenvectors of the reduced correlation matrix. Again, the ULS method is
known to work well with relatively small samples even when the number of variables is
large. We note that we used exploratory factor analysis since not only our aim is to determine
the number of dimensions, but also because structure theory postulates that all the manifest
variables saturate on all the latent factors and therefore constrained loadings to a value of
zero has no theoretical support.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 20/56
The affect structure revisited 20
To determine the number of dimensions, we used the eigenvalues of the estimated
correlations matrices. Again the difference in the affect structure models could be due to
methodological preferences, as most researchers use the eigenvalue-greater-than-one criterion
to determine the number of factors while at the same time others found that this criterion can
underestimate or overestimate the number of factors (Horn 1965, Cattell & Jaspers 1967,
Browne 1968, Linn 1968, Lee & Comrey 1979, Revell & Rocklin 1979, Yeomans & Golder
1982). In our analysis, we considered both the percentage of explained variance (at least
60%) and also the criteria provided by parallel analysis (Horn 1965). The later method
compares the eigenvalues obtained from uncorrelated variables by simulation to the ones
obtained with the sample, so that one can find eigenvalues “significantly” higher than one
found randomly with uncorrelated variables. Zwick and Velicer (1986) showed in a
comparison study that parallel analysis provided always the best solution.
It should be noted that we didn’t use the data collected on the continuous scales to
estimated the affect structure since preliminary descriptive analyses showed that for all
scenes, the distributions of all the variables (semantic items) were far from the necessary
normality assumption and could not be corrected neither by a robust method, nor by a
transformation. The possible explanation concerns the structural problem of a use of
continuous scale. Indeed, since participants are used to likert scales, they seem to use
continuous scales as likert ones, with the important consequence that most responses were set
to the minimal value (zero). The responses can be seen as dichotomized between a value of
zero and another positive value. Since the hypothesis of normality was not respected, we
could neither carry out a factor analysis, nor calculate correlation coefficients.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 21/56
The affect structure revisited 21
Affect structure
Our analyses are consistent across software, methods and datasets (different experimental
scenes), so that from now on we report the analyses made using M plus on the 2nd
scene. First
we retained three factors, i.e. three dimensions, on the basis of the parallel analysis and
contrasted eigenvalues. It should be noted that this criterion is known to be more accurate
than the criterion “eigenvalues greater than one” used by most authors (Russel 1980, Watson
& Tellegen 1985). These three factors explain 67.28% of the total variance (2 factors explain
only 57.74%).
The exploratory factor analysis (with varimax rotation) lead to three (independent)
factors that can be interpreted as activation-tension (Factor 1), valence (Factor 2) and
activation-energy (Factor 3). In Table 1 are presented the factor loadings given by pairs of
opposites for the three factors, as well as the uniqueness that can be used to evaluate how
well the factorial structure explains the different semantic items. Some socially stigmatized
negative semantic items show a relatively bigger uniqueness (guilty, bored, resigned,
inattentive, ashamed, hostile), and generally negative semantic items have a slightly bigger
uniqueness than their positive counterparts. This could be explained by the “Pollyanna
effect”, which consists of a human tendency, showed stronger among women, to use positive
semantic items in an evaluative context, as for personal judgments, more frequently, easily
and diversely than negative semantic items (Boucher & Osgood 1969, Warr 1971, Matlin &
Gawron, 1979, Osgood & Hoosain 1983). Consequently, the tendency to avoid brain
processing for the negative semantic items (Matlin & Stang, 1978) results in a part of
variance that cannot be explained by the three factors and could artificially make believe to
the necessity of more explanatory variables, particularly for those that have strongly marked
taboo value in French as in English (Armstrong & Hogg 2002). In addition, some semantic
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 22/56
The affect structure revisited 22
items resulting correctly from the back-translation seemed to have a different usage in French
than in English. Thus the French terms corresponding to strong, determined, inspired and
resigned were objects of question from the participants, who, while seeming to know the
adjectives, appeared perplexed when asked to use them to qualify what they just felt, with
again for consequence relatively important uniqueness and loadings deviating from what
could be expected. However, without those particular cases, the uniqueness of the semantic
items are quite satisfactory (from pleased .104 to dull .449 with an average of .292)
highlighting that the variables are globally well explained by the three-dimensional structure.
For the loadings also there are some exceptions that primarily correspond to cases in which
the loadings are very small and could be considered as not significantly different from zero,
but most of the semantic items have loadings of opposite signs that their bipolar opposites on
the three dimensions.
Although the results of the factor analysis already show that the hypothesis of
bipolarity can not be rejected, we nevertheless examine the polychoric correlations of the
opposite semantic items in Table 2. Except for the pair determined/scared and very low
correlations for inspired/worried and resigned/hopeful, which indicate again that the usage of
those semantic items is culturally marked, all correlations are negative from medium to high
(from -.315 to -.674, with an average of -.534) accrediting the bipolarity.
To summarize this first stage, responding to the criticisms addressed to the precedent
studies, the innovation of this research lies in the particular care taken to the experimental
plan and the statistical analysis method which allows us to unveil three independent and
bipolar dimensions, namely activation-tension, valence and activation-energy. In what
follows, we will strengthen this finding with the choice of the semantic items.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 23/56
The affect structure revisited 23
Bipolarity
As argued in the introduction, Watson’s analyses showed that the affect structure is
two-dimensional with a positive affect (PA) and a negative affect (NA) dimensions because
the items were chosen without their opposites. We reproduce here the experiment by
performing the same factor analysis as before, but with only the items of the PANAS. The
parallel analysis lead us to two dimensions (64.80% of explained variance) and the
exploratory factor analysis (with varimax rotation) indicates that these factors can indeed be
interpreted as NA (Factor 1) and PA (Factor 2). As expected, the items with a rather negative
connotation (like afraid, scared, ashamed, guilty, distressed, etc.) have high loadings on the
NA factor and the others have high loading on the PA factor. There is a cultural exception,
which could not be avoided by a correct back-translation, with the semantic item excited (in
French “agité”) because it has a negative connotation in French whereas excited has a
positive connotation in English. The factor loadings along with the uniqueness are presented
in Table 5. Our results thus confirm that the unipolarity found by Watson can be attributed to
an artefact of the experimental plan.
Since one could argue that the unipolarity found by Watson can be only due to the
smaller number of items (20 instead of 40), we performed the following analysis. We retain
as semantic indicators only every second pair along the circumference of the circumplex and
thus obtain 2 groups A and B of 10 pairs of bipolar opposites with which we replicate the
same statistical analysis. For both the parallel analysis retained three factors, i.e. three
dimensions. These three factors explain 64.98% for group A and 75.52% for group B of the
total variance (2 factors explain only 55.31% and 64.99% respectively). The exploratory
factor analysis (with varimax rotation) lead to three (independent) factors that can be
interpreted as activation-tension (Factor 2 or 1), valence (Factor 1 or 2) and activation-energy
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 24/56
The affect structure revisited 24
(Factor 3). In Tables 3 and 4 are presented the factor loadings given by pairs of opposites for
the three factors, as well as the uniqueness to evaluate how well the factorial structure
explains the different semantic items of the groups A and B. As previously, we observe
bigger uniqueness for negative semantic items due to the “Pollyanna effect” and for the same
socially stigmatized terms, but again most variables are well explained by the three
dimensional structure (from pleased .072 to dull .507 with an average of .294). With the same
exceptions that we saw before, the majority of the semantic items have again loadings of
opposite signs that their bipolar opposites on the three dimensions. Thus, with twenty or ten
pairs of bipolar opposites chosen from the whole circumference of the semantic circumplex,
the results highlight the same affect structure based on three independent and bipolar
dimensions. This was not the case when the items were chosen without their bipolar
opposites.
Interpretation and theoretical extensions
It is interesting to note that the three independent and bipolar dimensional affect
structure model was already highlighted by Russell and Mehrabian (1974, 1977) using an
exploratory approach. They postulated “that the three dimensions of pleasure-displeasure,
arousal-nonarousal, and dominance-submissiveness are both necessary and sufficient to
describe a large variety of emotional states” (p. 291). Although they qualified the affective
phenomena as emotional states, we believe that they correspond to affects as we have defined
them, since they described the affective phenomena through “a person is viewed as being in
some emotional state at all times, a state that can be described as a region within a three
dimensional space” and ““emotion” does not merely include occasional passionate states” (p.
274). They recognized (see also Mehrabian 1972a & b, Mehrabian & Ksionzky 1974) that the
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 25/56
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 26/56
The affect structure revisited 26
view is that the independent bipolar three dimensional model of affect structure is more
consistent because it results from an exploratory step, is conceived without theoretical a
priori, and while being parsimonious it takes into account the three necessary and sufficient
dimensions on which we are able to subjectively evaluate any temporary affective state.
The three-dimensional structure has also been defended by Sjösberg, Swensson &
Persson (1979), Matthews, Jones & Chamberlain (1990), Steyer, Schwenkmezger, Notz &
Eid (1994) and Schimmack & Grob (2000), but with correlated dimensions (oblique rotation).
However, in our analyses the correlations between valence and activation-tension is relatively
small (-.228) as well as the one between valence and activation-energy (-.110). The
possibility of correlation between dimensions was mainly introduced for goodness-of-fit
reasons rather than for theoretical reasons (Schimmack & Grob 2000). From a psychometric
point of view, a model with orthogonal dimensions seems preferable. For example diurnal
rhythms studies demonstrate that activation-energy follows a circadian rhythm but not
activation-tension or valence (Thayer 1989, Watson & al. 1999).
In fact, the affect model we estimate is in agreement with Heise's affect model (1988)
based on the three dimensions of Evaluation, Potency and Activation. This last model was
developed upon Osgood semantic model and one may legitimately wonder if the three
dimensional model is not simply the semantic structure model inducted by the semantic
items. If this were the case, then all studies using most of the semantic items (not only those
of the PANAS), should always find a three dimensional structure. This is not the case, since
we have seen that when the affective phenomenon is the mood, the structure is two-
dimensional (Bradburn 1969, Warr 1978, Lorr & Shea 1979, Zevon & Tellegen 1982, Warr,
Barter & Brownbridge 1983, Diener & Emmons 1984, Diener & Iran-Nejad 1986, Diener
1994).
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 27/56
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 28/56
The affect structure revisited 28
The affective dimensions and the psychological processes associated with those
dimensions provide an extremely efficient way of people to assimilate knowledge
about social life. After learning basic social categories, distributional constraints on
categories, and sentiments associated with each category, people can generate
automatically a wide range of social expectations. Only the core of social knowledge is
learned piecemeal; the rest is assembled as needed in various circumstances. Thus the
affective dimensions are central in the social psychological sense. (Morgan & Heise,
1988, p. 30)
Affective meaning can be seen as a metalanguage that translates the social status, power and
expressivity in more fundamental psychological dimensions of evaluation, potency and
activity (Parsons & Shils 1951, Kemper 1978, 1987). "Classifications of places, peoples,
objects and behaviours get transformed into a domain of feelings, where they lose their
qualitative uniqueness, become comparable to one another, and begin obeying quantitative
principles" (Heise 1987, p. 6). The relation between social structure and psychological
processes isolates something basic to the structuring of human judgments.
Whether it refers to affective, subjective or connotative meaning, essentially the same
phenomenon is being considered, which from the psychological point of view can be
formulated within the general framework of perception and learning. Affects are the primary
mode of apprehending the environment, of appraisal of the present situation in regard of the
coping resources at hand (Frijda, 1993, p. 384), that could correspond to the Evaluation
dimension. As affective dimensions correspond to basic mental processes that account for our
knowledge of the world, people's cognitive information is generated within the three-
dimensional framework which can then be found in all the domains (Mehrabian 1980).
Affects account for availability or unavailability of coping resources in a given situation
(Morris 1992) or the individual's trust in them (Moms, 1992), that could correspond to the
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 29/56
The affect structure revisited 29
Potency dimension. But in the process, they are also at the origin of increased readiness or
lowered thresholds for consonant responses vis-à-vis the situation in progress or in the
eventuality of an expected event in this situation, that could correspond to the Activation
dimension. Viewed in this way, the three affective dimensions allow to postulate that affect is
all at once cognition, motivation and behavior.
Conclusion
In this paper, through the analysis of real data issued from an appropriate
experimental plan and data analysis method, we show that the affect structure has three
independent dimensions, namely activation-tension, valence and activation-energy. This
three-dimensional structure was already uncovered by several researchers (Bush 1973,
Averill 1975, Russell & Mehrabian 1977, Morgan & Heise 1988). In order to give validity to
our results, we paid special attention to the experimental plan which across studies on this
subject, can vary quite substantially. First, it was important to specify what is really intended
by affect, especially its difference from mood which, as several studies have shown, has not
the same structure as affect. Second, the choice of the semantic items is crucial, since we
have shown that choosing the ones in the PANAS only, lead to a two-dimensional structure
with a PA and a NA dimensions. Third, the statistical criteria chosen to determine the number
of dimensions can also play an important role, in that when it is based on the number of
eigenvalues greater than one, two dimensions are found (Russell 1980, Watson & Tellegen
1985) while when using parallel analysis, three dimensions are suitable.
We postulate that these three independent and bipolar dimensions of the affect
structure model highlighted by our results correspond to the human primary modalities to
apprehend our material and social environment. We believe that these dimensions correspond
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 30/56
The affect structure revisited 30
to the Evaluation, Potency and Activation dimensions of Osgood’s mediation theory of
meaning where they play the role of a common logical structure of all human beings, prior
and independent of the language, in the cognitive processes. Those same dimensions are also
at the basis of Heise’s Affect Control Theory in which as a metalanguage they translate
everything in our environment: objects, persons, behaviors in comparable affective quanta
used by a on-going monitoring judgment process which makes us assimilate knowledge about
our social and material environment, directs our response and makes us behave accordingly.
That expression of the three affective dimensions allows to postulate that affect is all at once
cognition, motivation and behavior.
Admittedly more than a consensual taxonomy, it still remains much to conceptualize,
the particular nature of affects and their place relatively to other concepts, the type of
connection between affect and mood, affect and emotion, emotion and mood, mood and
personality, but the theorization that we propose in this article based on affect takes into
account a large panel of domains and offers a promising structural framework to affect
psychology.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 31/56
The affect structure revisited 31
References
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental
disorders (4th ed.). Washington, DC: Author.
Armstrong N. & Hogg C. (2002). The Pollyanna Principle in English and French Lexis:
Some Results and Issues of Methodology. Leeds Working Papers in Linguistics and
Phonetics, 9, 1-33.
Averill J. R. (1975). A semantic atlas of emotional concepts. Catalog of selected documents
in psychology, Vol.53 (30), MS 421.
Baddeley A. D. (1982). Domains of recollection. Psychological Review, 89 , 708-729.
Bentler P. M. (1969). Semantic space is (approximately) bipolar. Journal of Psychology, 71,
33-40.
Berlyne D. E. (1960). Conflict, arousal and curiosity. New York, McGraw-Hill.
Borgotta E. I. (1961). Mood, personality, and interaction. Journal of General Psychology, 64,
105-137.
Boucher J., & Osgood C. E. (1969). The Pollyanna hypothesis. Journal of Verbal and
Learning Behavior , 8(1), 1-8.
Bradburn N. M. & Caplovitz D. (1965). Reports on happiness: A pilot study of behaviour
related to mental health. Chicago, Aldine.
Bradburn N. M. (1969). The structure of psychological well-being. Chicago, Aldine.
Bradley M. M. (1994). Emotional memory: A dimensional analysis. In: Ed° Van Goouen S.
H. M., Van de Poll N. E. & Sergent J. A., Emotions: Essays on emotion theory, 97-
134, Hillsdale, New Jersey, Erlbaum.
Browne M. W. (1968). A comparison of factor analytic techniques. Psychometrika, 33, 267-
334.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 32/56
The affect structure revisited 32
Buck R. (1985). Prime theory: An integrated view of motivation and emotion. Psychological
Review, 92(3), 389-413.
Buck R. (1993). What is this thing called participantive experience? Reflections on the
neuropsychology of qualia. Special Section: Neuropsychological perspectives on
components of emotional processing. Neurropsychology, 7(4), 490-499.
Bush L. E. II (1973). Individual differences multidimensional scaling of adjectives denoting
feelings. Journal of Personality and Social Psychology, 25, 50-57.
Cacioppo J. T. & Berntson G. G. (1994). Relationship between attitude and evaluative space:
A critical review, with emphasis on the separability of positive and negative
substrates. Psychological Bulletin, 115, 401-423.
Cannon W. B. (1927). The James-Lange theory of emotion: A critical examination and an
alternative theory. American Journal of Psychology, 39, 106-124.
Carver C. S. (1996). Some ways in which goals differ and some implications of those
differences. In: Gollwither P. M. & Bargh J. A. (Ed°), The psychology of action, 645-
672, New York, Guilford.
Cattell R. B. & Jasper J. (1967). A general plasmode for factor analytic exercieses and
research. Multivariate Behavioral Research Monographs, 3, 1-212.
Center for the Study of Emotion and Attention (CSEA-NIMH). (1995). The international
affective picture system: Digitized photographs. Gainsville, University of Florida:
Center for Research in Psychology.
Clore G. L., Schwarz N. & Conway M. (1994). Affective causes and consequences of social
information processing. In: Ed° Wyer R. W. & Srull T. K., Handbook of social
cognition, 323–417, Hillsdale, NJ: Erlbaum.
Costa P. T. & McCrae R. R. (1992). Four ways five factors are basic. Personality and
Individual Differences, 13, 653-665.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 33/56
The affect structure revisited 33
De Saussure F. (1916). Cours de linguistique générale. Ed° Bally C. & Sechehaye A.,
Lausanne & Paris, Payot.
Diener E. & Emmons R. A. (1984).The independence of positive and negative affect. Journal
of Personality and Social Psychology, 47, 1105-1117.
Diener E. & Iran-Nejad A. (1986). The relationship in experience between different types of
affect. Journal of Personality and Social Psychology, 50, 1131-1138.
Diener E. (1994). Assessing Participantive Well-Being: Progress and Opportunities. Social
Indicators Research, 31, 103-157.
Duffy E. (1957). The psychological significance of the concept of arousal or activation.
Psychological Review, 64,265-275.
Emmons R. A. & Diener E. (1986). Influence of impulsivity and sociability on participantive
well-being. Journal of Personality and Social Psychology, 50, 1211-1215.
Eysenck H. J. (1991). Dimensions of personality: 16, 5 or 3?-Criteria for a taxonomic
paradigm. Personality and Individual Differences, 12, 773-790.
Feldman L. A. (1995). Variations in the circumplex structure of mood. Personality and
Social Psychology Bulletin, 21, 806-817.
Frijda N. H. (1986). The emotions. Cambridge & New York, Cambridge University Press.
Frijda N. H. (1987). Emotion, cognitive structure and action tendency. Cognition and
Emotion, 1, 115-143.
Frijda N. H. (1993). Moods, emotion episodes, and emotions. In: Ed° Lewis M. & Haiviland
J. M., Handbook of emotion, 381-404, New York, Guilford Press.
Golden D. R. & Baddeley A. D. (1975). Context-dependent memory in two natural
environments: on land and underwater. British Journal of Psychology, 66, 325-331.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 34/56
The affect structure revisited 34
Gotlib I. H. & Meyer J.-P. (1986). Factor analysis of the Multiple Affect Adjective Checklist:
A separation of positive and negative affect. Journal of Personality and Social
Psychology, 50:1161-1165.
Green D. P., Goldman S. L. & Salovey P. (1993). Measurement error masks bipolarity in
affect ratings. Journal of Personality and Social Psychology, 64, 1029-1041.
Green D. P. & Salovey P. (1999). In what sense are positive and negative affect independent?
Psychological Science, 10, 304-306.
Harman H. (1960). Modern Factor Analysis. (3rd
revised ed.1976), Chicago, IL: University of
Chicago Press.
Hebb D. O. (1955). Drives and the CNS (conceptual nervous system). Psychological Review,
62, 243-254.
Heise D. R. (1966). Social status, attitudes and word connotations. Sociological Inquiry, 36,
227-239.
Heise D. R. (1979). Understanding Events: Affect and the Construction of Social Action.
New York: Cambridge University Press.
Heise D. R. (1987). Affect Control Theory: Concepts and Model. Journal of Mathematical
Sociology, 13, 1-33.
Heise D. R. & McKinnon N. (1987). Affective bases of likelihood judgments. Journal of
Mathematical Sociology, 13, 133-151.
Heise D. R. (1988). Delusions and the construction of reality. In: Ed° Maher B. A. &
Oltmanns T. F., Delusional beliefs, 259-272, Oxford, England: John Wiley & Sons.
Heise D. R. & Thomas L. (1989). Predicting impressions created by combinations of emotion
and social identity. Social Psychology Quaterly, 52, 141-148.
Heise D. R. (1998). Conditions for Empathic Solidarity. In: Ed°. Doreian P. & Fararo T. The
Problem of Solidarity: Theories and Models, 197-211, Amsterdam: Gordon & Breach.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 35/56
The affect structure revisited 35
Horn J. L. (1965). A rationale and test for the number of factors in factor analysis.
Psychometrika, 30, 179-185.
Jöreskog K. & Sörbom D. (1977). Factor analysis by least-squares and maximum-likelihood
methods. In: Ed° Enslein K., Ralston A. & Wilf H. S., Statistical methods for digital
computers. New York: Wiley.
Jöreskog K. (2003). Factor analysis by MINRES. Online:
http://www.ssicentral.com/lisrel/minres.pdf
Kemper T. D. (1978). A Social Interactional Theory of Emotions. New York: Wiley.
Kemper T. D. (1987). How Many Emotions are There? American Journal of Sociology, 93,
263-289.
Kumata H. & Schramm W. (1956). A pilot study of cross-cultural meaning. Public Opinion
Quarterly, 20, 574–84.
Lang P. J. (1978). Anxiety: Toward a psychophysiological definition. In: Akiskal H. S. Webb
W. L. (Ed°), Psychiatric diagnosis: Exploration of biological predictors, 365-389,
New York, Spectrum.
Lang P. J., Bradley M. M. & Cuthbert B. N. (1992). A motivational analysis of emotion:
Reflex-cortex connections. Psychological Science, 3, 44-49.
Lang P. J., Greenwald M. K., Bradley M. M. & Hamm A. O. (1993). Looking at pictures:
Affective, facial, visceral and behavioural reactions. Psychophysiology, 30, 261-273.
Lang P. J. (1994). The motivational organization of emotion: Affect reflex connections. In:
Ed° Van Goouen S. H. M., Van de Poll N. E. & Sergent J. A., Emotions: Essays on
emotion theory, 61-93, Hillsdale, New Jersey, Erlbaum.
Lang P. J., Bradley M. M. & Cuthbert B. N. (1995). International affective picture system
(IAPS): Technical manual and affective ratings. Gainsville, University of Florida,
Center for Research in Psychology.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 36/56
The affect structure revisited 36
Lapierre A. M., Gaudreau P. & Blondin J. P. (1999). Evaluation de l'affectivité positive et
négative en contexte sportif: Traduction francophone du PANAS. Communication
présentée au Congrès annuel de la Société Québecoise pour la Recherche en
Psychologie, Québec.
Larsen R. J. & Diener E. (1992). Promises and problems with the circumplex model of
emotion. In: Ed° Clark M. S., Review of personality and social psychology:
Emotion, Vol. 13, 25-29, Newbury Park, CA, Sage.
Lee H. B. & Comrey A. L. (1979).Distortions in a commonly used factor analytic procedure.
Multivariate Behavioral Research, 14, 301-321.
Lindsley D. B. (1951). Emotion. In: Ed° Stevens S. S. Handbook of experimental psychology,
473-516, New York, Wiley.
Linn R. L.(1968). A Monte Carlo approach to the number of factors problem. Psychometrika,
33, 37-71.
Lorr M. & Shea T. M. (1979). Are mood states bipolar? Journal of Personality Assessment ,
43, 468-472.
Lorr M. & McNair D. M. (1982). Manual: Profile of mood states (bipolar form). San Diego,
CA, Educational and industrial testing services.
Lorr M. & Wunderlich R. A. (1980). Mood states and acquiescence. Psychological Reports,
46, 191-195.
Mandler G. (1984). Mind and body: Psychology of emotion and stress. New York, Norton.
Mayer J. D. & Gaschke Y. N. (1988). The experience and meta-experience of mood. Journal
or Personality and Social Psychology, 55, 102-111.
McKinnon N. J. (1994). Symbolic Interactionism as Affect Control. Albany: State University
of New York Press.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 37/56
The affect structure revisited 37
McNair D. M. & Lorr M. (1964). An analysis of mood in neurotics. Journal of Abnormal and
Social Psychology, 69, 620-627.
Matlin M. & Stang D. (1978). The Pollyanna Principle: Selectivity in Language, Memory,
and Thought. Cambridge, MA: Schenkman Publishing Company.
Matlin, M., & Gawron, V. (1979). Individual differences in Pollyannaism. Journal of
Personality Assessment, 43(4), 411-412.
Matthews G., Jones D. M. & Chamberlain A. G. (1990).Refining the measurement of mood:
The UWIST Mood Adjective Checklist. British Journal of Psychology, 81, 17-42.
Mehrabian A. (1972a). Nonverbal communication. Chicago, IL: Aldine-Atherton.
Mehrabian A. (1972b). Nonverbal communication. In: Ed° Cole J.K., Nebraska symposium
on motivation, 19, 107-161, Lincoln, NE: University of Nebraska Press.
Mehrabian A. & Ksionzky S. (1974). A theory of affiliation. Lexington, MA: D.C. Heath.
Mehrabian A. (1980). Basic Dimensions for a General Psychological Theory. Cambridge,
MA: Oelgeschlager, Gunn & Hain.
Meyer G. J. & Shack J. R. (1989). Structural convergence of mood and personality: Evidence
for old and new directions. Journal of Personality and Social Psychology, 57, 691-
706.
Morgan R. L. & Heise D. R. (1988). Structure of emotions. Social Psychology Quarterly, 51,
19-31.
Morris W. N. (1989). Mood: The frame of mind . New York, Springer-Verlag.
Morris W. N. (1992). A functional analysis of the role of mood in affective systems. In: Ed°.
Clark, Review of personality and social psychology, 13, 256-293, Newbury Park, CA:
Sage.
Muthén L. K. & Muthén B. O. (1998–2004). Mplus user’s guide (3rd ed.). Los Angeles:
Muthén & Muthén.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 38/56
The affect structure revisited 38
Nemanick R. C. Jr. & Munz D. C. (1994). Measuring the poles of negative and positive mood
using the Positive Affect Negative Affect Schedule and Activation Deactivation
Adjective Check List. Psychological Reports, 74, 195-199.
Neufeld R. W. J. (1975). A multidimensional scaling analysis of schizophrenics' and normals'
perceptions of verbal similarity. Journal of Abnormal Psychology, 84, 498-507.
Neufeld R. W. J. (1976). Simultaneous processing of multiple stimulus dimensions among
paranoid and non-paranoid schizophrenics. Multivariate Behavioral Research, 11,
425-441.
Nowlis V. & Nowlis H. H. (1956). The description and analysis of mood. In: Annals of the
New York Academy of Sciences, Vol. 65, 345-355, New York, New York Academy of
Sciences.
Olsson U. (1979). On the robustness of factor analysis against crude classification of the
observations. Multivariate Behavioral Research, 14, 485-500.
Ortony A., Clore A. C. & Collins A. (1988). The cognitive structure of emotions. New York,
Cambridge University Press.
Osgood C. E. & Suci G. J. (1955). Factor analysis of meaning. Journal of Experimental
Psychology, 50, 325-338.
Osgood C. E., Suci G. J. & Tannenbaum P. H. (1957). The measurement of meaning. Urbana:
University of Illinois Press.
Osgood C. E. (1960). The Cross-Cultural Generality of Visual-Verbal Synesthetic
Tendencies. Behavioral Science , 5, 146-169.
Osgood C. E. (1963). An Exploration into Semantic Space. New York: Basic Books.
Osgood C. E. (1966). Dimensionality of the semantic space for communication via facial
expressions. Scandinavian Journal of Psychology, 7, 1-30.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 39/56
The affect structure revisited 39
Osgood C. E. (1969). On the whys and wherefores of E, P and A. Journal of Personality and
Social Psychology, 12, 194-199.
Osgood C. E., May W. H. & Miron M. S. (1975). Cross-cultural universals of affective
meaning. Urbana, University of Illinois Press.
Osgood C. E. & Hoosain R. (1983). Pollyanna II: Two types of negativity. Journal of
Psychology: Interdisciplinary & Applied , 113(2), 151-160.
Parkinson B., Briner R. B., Reynolds S. & Totterdell P. (1995). Time frames for mood:
Relations between momentary and generalized ratings of affect. Personality and
Social Psychology Bulletin, 21, 331-339.
Parsons T. & Shils E. (1951). Toward a General Theory of Action. Cambridge, Ma.: Harvard
University Press.
Reisenzein R. (1992). A structuralist reconstruction of Wundt's three-dimensional theory of
emotion. In: Ed° Westmeyer H. The structuralist program in psychology:
Foundations and applications, 141-189, Toronto, Ontario, Canada, Hogrefe & Huber.
Reisenzein R. & Schonpflug W. (1992). Stumpf's cognitive-evaluative theory fo emotion.
American Psychologist , 47, 34-45.
Reisenzein R. (1994). Pleasure-activation theory and the intensity or emotions. Journal of
Personality and Social Psychology, 67, 525-539.
Reisenzein R. & Shimmack U. (1999). Similarity judgments and covariations of affects:
Findings and implications doe affect structure. Personality and Social Psychology
Bulletin, 25, 539-555.
Revelle W. & Rocklin T. (1979).Very simple structure: An alternative procedure for
estimating the optimal number of interpretable factors. Multivariate Behavioral
Research, 14, 403-414.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 40/56
The affect structure revisited 40
Russell J. A. & Mehrabian A. (1974). Distinguishing anger and anxiety in terms of emotional
response factors. Journal of Consulting and Clinical Psychology, 42, 79-83.
Russell J. A. & Mehrabian A. (1977). Evidence for a three-factor theory of emotions. Journal
of Research in Personality, 11, 273-294.
Russell J. A. (1978). Evidence of convergent validity on the dimensions of affect. Journal of
Personality and Social Psychology, 36, 1152-1168.
Russell J. A. (1979). Affective space is bipolar. Journal of Personality and Social
Psychology, 37, 345-356
Russell J. A. (1980). A circumplex model of affect. Journal of Personality and Social
Psychology, 39, 1161-1178.
Russell J. A. & Steiger J. H. (1982). The structure in person's implicit taxonomy of emotions.
Journal of Research in Personality, 16, 447-469.
Russell J. A. (1989). Measures of emotion. In: Ed° Plutchik R. & Kellerman H., Emotion:
Theory, research and experience, Vol. 4, 83-111, San Diego, CA, Academic Press.
Russell J. A., Lewicka M. & Miit T. (1989). A cross-cultural study of a circumplex model of
affect. Journal of Personality and Social Psychology, 57, 848-856.
Russell J. A. (1991). Culture and the categorization of emotions. Psychological Bulletin, 110,
426-450.
Russell J. A. & Carroll J. M. (1999). On the bipolarity of positive and negative affect.
Psychological Bulletin, 125, 3-30.
Russell J. A. & Feldman Barrett L. (1999). Core affect, prototypical emotional episodes and
other things called emotion: Dissecting the elephant. Journal of Personality and
Social Psychology, 76, 805-819.
Russell J. A. (2003). Core affect and the psychological construction of emotion.
Psychological Review, 110 (1), 145-172.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 41/56
The affect structure revisited 41
Schachter S. & Singer J. E. (1962). Cognitive, social and physiological determinants of
emotional states. Psychological Review, 69, 379-399.
Scherer K. R. (1984a). On the nature and function of emotion: A component process
approach. In: Ed° Scherer K. R. & Ekman P. Approaches to Emotion, 293-318,
Hillsdale, New Jersey, Erlbaum.
Scherer K. R. (1984b). Emotion as a multicomponent process: a model and some cross-
cultural data. In: Ed° Shaver P., Review of personality and social psychology, Vol. 5,
37-63, Beverly Hills, CA, Sage.
Scherer K. R. (1993). Neuroscience projections to current debates in emotion psychology.
Cognition and Emotion, 7, 1-41.
Scherer K. R. (2000). Psychological models of emotion. In: Ed° Borod J. C. The
neuropsychology of emotion, 137-166, Oxford & New York, Oxford University Press.
Schlosberg H. (1952). The description of facial expression in terms of two dimensions.
Journal of Experimental Psychology, 44, 229-237.
Schimmack U. & Reisenzein R. (1997). Cognitive processes involved in similarity judgments
of emotion concepts. Journal of Personality and Social Psychology, 73, 645-661.
Schimmack U. & Grob A. (2000). Dimensional models of core affect: A quantitative
comparison by means of structural equation modeling. European Journal of
Personality, 14, 325-345.
Shaver P., Schwartz J., Kirson D. & O’Connor C. (1987). Emotion knowledge: Further
exploration of a prototype aproach. Journal of Personality and Social Psychology, 52,
1061-1086.
Sjösberg L., Svensson E. & Persson L.-O. (1979). The measurement of mood. Scandinavian
Journal of Psychology, 20,1-18.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 42/56
The affect structure revisited 42
Spence S., Shapiro D. & Zaidel E. (1996). The role of the right hemisphere in the
physiological and cognitive components of emotional processing. Psychophysiology,
33, 112-122.
Steyer R., Schwenkmezger P. , Notz P. & Eid M. (1994). Testtheoretische Analysen des
Mehrdimensionalen Befindlichkeitsfragebogens (MDBF) [Test-theorical analyses of
the Multidimensional Mood Questionaire]. Diagnostica, 40, 320-328.
Stryker S. (1992). Symbolic Interaction Theory. In: Ed°. Borgatta E.F. & Borgatta M.L.,
Encyclopedia of Sociology. New York: Macmillan.
Tellegen A. (1985). Structures of mood and personality and their relevance to assessing
anxiety, with an emphasis on self-report. In: Ed° Tuma A. H. & Maser J. D., Anxiety
and anxiety disorders, 681-706, Hillsdale, NJ, Erlbaum.
Thayer R. E. (1967). Measurement of activation through self-report. Psychological Reports,
20, 663-678.
Thayer R. E. (1978). Factor analytic and reliability studies on the activation-deactivation
adjective check list. Psychology Reports, 42, 747-756.
Thayer R. E. (1986). Activation-Deactivation Check List: Current overview and structural
analysis. Psychological Reports, 58, 607-614.
Thayer R. E. (1989). The biopsychology of mood and activation. New York, Oxford
University Press.
Thayer R. E. (1996). The origin of everyday moods: Managing energy, tension and stress.
New York, Oxford University Press.
Thomas D. L. & Diener E. (1990). Memory accuracy in the recall of emotions. Journal of
Personality and Social Psychology, 59, 291-297.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 43/56
The affect structure revisited 43
Triandis H. C. & Osgood C. E. (1958) A comparative factorial analysis of semantic structures
of monolingual Greece and American students. Journal of Abnormal and Social
Psychology, 57, 187-196.
Tulving E. & Thomson D. M. (1973). Encoding specificity and retrieval processes in episodic
memory. Psychological Review, 80, 352-373.
Tulving E. (1979). Relation Between Encoding Specifity and Levels of Processing. In: Ed°
Cermak L.S. & Craik F.I.M. : Levels of Processing in Human Memory, 405-428,
Hillsdale, N.J.: Erlbaum.
Tulving E. (1983). Elements of episodic memory. Vol. 2, New York: Oxford University
Press.
Vanman E. J., Boehmelt A. H., Dawson M. E. & Shell A. M. (1996). The varying time
courses of attentional and affective modulation of the startle eyeblink reflex.
Psychophysiology, 33, 691-697.
Warr P. (1971). Pollyanna’s personal judgments. European Journal of Social Psychology,
1(3), 327-338.
Warr P. (1978). A Study of Psychological Well-Being. The British Journal of Psychology,
1978, 6, 111-121.
Warr P., Barter J. & Brownbridge G. (1983). On the independence of positive and negative
affect. Journal of Personality and Social Psychology, 55, 128-141.
Watson D. & Tellegen A. (1985). Toward a consensual structure of mood. Psychological
Bulletin, 98, 219-235.
Watson D., Clark L. A. & Tellegen A. (1988). Development and validation of brief measures
of positive and negative affect: The PANA Scales. Journal of Personality and Social
Psychology, 54, 1063-1070.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 44/56
The affect structure revisited 44
Watson D. & Clark L. A. (1992). On traits and temperament: general and specific factors of
emotional experience and their relation to the five-factor model. Journal of
Personality, 60, 441-476.
Watson D. & Clark L. A. (1994). Manual for the Positive and Negative Affect Schedule
(Expanded Form). University of Iowa; available from the authors.
Watson D. & Clark L. A. (1997). The measurement and mismeasurement of mood: Recurrent
and emergent issues. Journal of Personality Assessment , 86, 267-296.
Watson D., Wiese D., Vaidya J. & Tellegen A. (1999). The two general activation systems of
affect: Structural findings, evolutionary considerations and psychobiological
evidence. Journal of Personality and Social Psychology, 76, 820-838.
Whissell C. M. (1981). Pleasure and activation revisited: Dimension underlying semantic
responses to fifty randomly selected "emotional" words. Perceptual and Motor Skills,
53, 871-874.
Whorf B. L. (1940). Science and linguistics. Technology Review, 42, 229-231 & 247-248.
Whorf B. L. (1956). Language, thought and reality. In: Ed° Carroll J. B., Language, Thought,
and Reality. Selected Writings of Benjamin Lee Whorf , 207-219, New York: MIT
Press; London: John Wiley.
Williams D. G. (1989). Personality effects in current mood: pervasive or reactive?
Personaltiy and Individual Differences, 10, 941-948.
Wundt W. (1896). Grundriss des Psychologie [Outlines of Psychology]. Leibzig :
Engelmann.
Wundt W. (1906). Vorlesungen über die Menschen- und Tierseele [Lectures on the human
and animal mind] (4th
). Hamburg: Voss.
Yeomans K. A. & Golder P. A. (1982). The Guttman-Kaiser criterion as a predictor of the
number of common factors. Statistician, 31, 221-229.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 45/56
The affect structure revisited 45
Zautra A. J., Potter P. T. & Reich J. W. (1997). The independence of affects is context-
dependent: An integrative model of the relationship between positive and negative
affect. In: Ed° Lawton M. P. & Schaie K. W., Annual Review of Gerontology and
Geriatrics, 17, 75-103.
Zevon M. A. & Tellegen A. (1982). The structure of mood change: an
idiographic/nomothetic analysis. Journal of Personality and Social Psychology, 43,
111-122.
Zillmann D. (1983). Transfer of excitation in emotional behavior. In: Ed° Cacioppo J. T. &
Petty R. E., Social psychophysiology: A sourcebook , 215-240, New York: Guilford
Press.
Zwick W. R. & Velicer W. F. (1986). Comparison of five rules for determining the number
of components to retain. Psychological Bulletin, 99/3, 432-442.
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 46/56
The affect structure revisited 46
Appendix
The following table shows the correspondence between the English and French semantic
items, by opposite pairs and according to their orientation on valence and activation
dimensions.
English French
valence+ valence - valence+ valence -
activation+ activation- activation+ activation-
alert* sleepy alerte somnolent
active* dull éveillé engourdi
elated depressed exalté déprimé
enthusiastic* lethargic enthousiaste léthargique
hopeful resigned optimiste résigné
cheerful sad joyeux triste
proud* guilty* fier coupable
inspired* worried inspiré soucieux
attentive* inattentive attentif distrait
interested* bored intéressé ennuyé
activation- activation+ activation- activation+
strong* afraid* impassible craintif
determined* scared* déterminé effrayé
confident ashamed* confiant honteux
pleased distressed* content désespéré
friendly hostile* amical hostile
peaceful irritable* paisible irritable
relaxed nervous* détendu nerveux
satisfied upset* satisfait fâché
calm excited* calme agité
serene jittery* serein anxieux
*indicates semantic items of the PANAS
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 47/56
The affect structure revisited 47
Authors Note
This research is supported by grants from the Swiss National Science Foundation
(FNRS 610-057883.99 and PP001-106465). We would like to thank Susanne Kaiser, Michael
Eid and Grazia Ceschi for their helpful comments on a draft of this article.
Correspondence concerning this article should be addressed to Véronique Elefant-
Yanni, Department of Psychology, University of Geneva, Boulevard du Pont-d'Arve 40, 1205
Geneva, Switzerland, or Rue de Monthoux 38, 1201 Geneva, Switzerland. E-mail:
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 48/56
The affect structure revisited 48
Footnotes
1The denotative usage gives to the thing represented by the word its essential and objective
meaning.2
The connotative usage presents the thing enriched by affective associations which, though
intangible, are nonetheless real.3Wilhelm von Humboldt 1767-1835
4Edward Sapir 1884-1936 & Benjamin Lee Whorf 1897-1941
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 49/56
The affect structure revisited 49
Table 1
Mplus: Varimax factor loadings estimated with ULS
Mplus EFA 3F LP2bis: Varimax-rotation Factor loadings estimated by ULS(RMSR=0.0663)
Items Factor 1 Factor 2 Factor 3 Uniqueness
interested 0.126 0.665 -0.378 0.399
bored 0.144 -0.278 0.476 0.676
afraid 0.856 -0.068 -0.231 0.209
strong -0.346 -0.186 0.525 0.570
inattentive -0.208 0.081 0.599 0.592
attentive 0.600 0.367 -0.383 0.359
determined 0.585 0.504 -0.121 0.389
scared 0.886 -0.173 -0.206 0.142
inspired 0.108 0.750 -0.192 0.389
worried 0.889 -0.145 -0.080 0.182
ashamed 0.549 -0.372 -0.062 0.557
confident -0.268 0.735 0.270 0.314
guilty 0.400 -0.241 -0.225 0.731
proud 0.030 0.769 0.018 0.407
pleased -0.380 0.864 -0.065 0.104
distressed 0.799 -0.357 -0.001 0.234
cheerful -0.378 0.852 -0.016 0.131
sad 0.748 -0.286 0.026 0.358
hostile 0.606 -0.358 -0.003 0.504
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 50/56
The affect structure revisited 50
friendly -0.269 0.744 0.001 0.375
resigned 0.577 0.031 0.179 0.634
hopeful -0.176 0.829 0.045 0.280
peaceful -0.455 0.539 0.424 0.323
irritable 0.727 -0.335 0.131 0.341
enthusiastic -0.231 0.868 -0.112 0.180
lethargic 0.262 -0.100 0.790 0.297
nervous 0.815 -0.111 -0.225 0.273
relaxed -0.465 0.666 0.265 0.270
depressed 0.713 -0.225 0.218 0.393
elated -0.096 0.759 -0.214 0.369
satisfied -0.362 0.863 -0.001 0.124
upset 0.792 -0.423 0.014 0.193
active 0.343 0.584 -0.408 0.375
dull -0.080 0.031 0.737 0.449
excited 0.708 0.084 -0.311 0.395
calm -0.447 0.389 0.491 0.407
sleepy -0.078 0.039 0.868 0.239
alert 0.541 0.380 -0.393 0.408
serene -0.327 0.715 0.356 0.255
jittery 0.778 -0.179 -0.294 0.276
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 51/56
The affect structure revisited 51
Table 2
Polychoric correlations for the opposite semantic items
LP2bis
Items Polychoric
interested, bored -0.519
afraid, strong -0.435
inattentive, attentive -0.538
determined, scared 0.408
inspired, worried -0.025
ashamed, confident -0.564
guilty, proud -0.177
pleased, distressed -0.663
cheerful, sad -0.606
hostile, friendly -0.67
resigned, hopeful -0.042
peaceful, irritable -0.523
enthousiastic, lethargic -0.22
nervous, relaxed -0.591
depressed, elated -0.315
satisfied, upset -0.674
active, dull -0.394
excited, calm -0.616
sleepy, alert -0.429
serene, jittery -0.591
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 52/56
The affect structure revisited 52
Table 3
Mplus: Varimax factor loadings estimated with ULS
Mplus EFA 3F LP2bisVA: Varimax-rotation Factor loadings estimated by ULS
(RMSR=0.0739) Items Factor 1 Factor 2 Factor 3 Uniqueness
afraid 0.179 0.785 -0.343 0.234
strong 0.075 -0.217 0.590 0.599
inattentive -0.127 -0.050 0.635 0.578
attentive -0.230 0.483 -0.507 0.457
ashamed 0.519 0.496 -0.129 0.467
confident -0.843 -0.152 0.228 0.214
guilty 0.336 0.356 -0.267 0.689
proud -0.789 0.081 -0.116 0.358
hostile 0.430 0.550 -0.080 0.506
friendly -0.786 -0.158 0.010 0.357
resigned 0.024 0.598 0.100 0.632
hopeful -0.849 -0.068 0.016 0.274
nervous 0.184 0.806 -0.325 0.210
relaxed -0.710 -0.354 0.273 0.295
depressed 0.264 0.665 0.073 0.482
elated -0.722 -0.063 -0.292 0.390
excited 0.017 0.721 -0.371 0.343
calm -0.448 -0.358 0.486 0.434
sleepy -0.069 0.041 0.848 0.275
alert -0.308 0.509 -0.504 0.393
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 53/56
The affect structure revisited 53
Table4
Mplus: Varimax factor loadings estimated with ULS
Mplus EFA 3F LP2bisVB: Varimax-rotation Factor loadings estimated by ULS
(RMSR=0.0529) Items Factor 1 Factor 2 Factor 3 Uniqueness
interested 0.017 0.778 -0.305 0.301
bored 0.148 -0.326 0.449 0.670
determined 0.469 0.629 0.016 0.385
scared 0.926 0.047 -0.062 0.137
inspired -0.037 0.810 -0.116 0.329
worried 0.918 0.064 0.031 0.153
pleased -0.582 0.767 -0.031 0.072
distressed 0.858 -0.172 0.062 0.231
cheerful -0.575 0.742 0.008 0.119
sad 0.788 -0.117 0.076 0.360
peaceful -0.634 0.350 0.346 0.355
irritable 0.728 -0.176 0.207 0.396
enthusiastic -0.440 0.793 -0.051 0.176
lethargic 0.167 -0.117 0.839 0.255
satisfied -0.573 0.745 0.023 0.116
upset 0.844 -0.244 0.106 0.217
active 0.243 0.655 -0.302 0.420
dull -0.134 -0.050 0.687 0.507
serene -0.520 0.568 0.351 0.284
jittery 0.806 0.010 -0.195 0.312
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 54/56
The affect structure revisited 54
Table 5
Mplus: Varimax factor loadings estimated with ULS
Mplus EFA 2F LP2W: Varimax-rotation Factor loadings estimated by ULS
(RMSR=0.0777). Variables Factor 1 Factor 2 Uniqueness
interested -0.131 0.803 0.338
afraid 0.829 0.371 0.175
strong -0.286 -0.465 0.702
attentive 0.397 0.704 0.347
determined 0.297 0.647 0.493
scared 0.875 0.292 0.149
inspired -0.201 0.728 0.430
ashamed 0.654 -0.102 0.562
guilty 0.510 -0.012 0.740
proud -0.307 0.600 0.546
distressed 0.812 0.007 0.340
hostile 0.731 -0.093 0.457
irritable 0.801 -0.114 0.346
enthousiastic -0.555 0.679 0.231
nervous 0.829 0.269 0.241
upset 0.897 -0.099 0.185
active 0.122 0.770 0.392
excited 0.657 0.404 0.405
alert 0.385 0.673 0.398
jittery 0.804 0.271 0.280
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 55/56
The affect structure revisited 55
Figure Captions
Figure 1. Circumplex of semantic items according to valence and activation
8/4/2019 Paper Archives Pia Get
http://slidepdf.com/reader/full/paper-archives-pia-get 56/56
The affect structure revisited 56
Figure 1
activation+alertactive
elatedenthousiastic
hopeful
cheerful
proud
inspired
attentive
interested
valence+
strong
determined
confident
pleased
friendly
peaceful
relaxedsatisfied
calmsereneactivation-
sleepydulldepressed
lethargic
resigned
sad
guilty
worried
inattentive
bored
valence -
afraid
scared
ashamed
distressed
hostile
irritable
nervousupset
excitedjittery