lundh, johnsson et al (2002)
TRANSCRIPT
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 1/19
Alexithymia, Memory of Emotion, Emotional Awareness,and Perfectionism
Lars-Gunnar Lundh, Anders Johnsson, Karin Sundqvist, and Helen OlssonStockholm University
The Toronto Alexithymia Scale (TAS-20; R. M. Bagby, J. D. A. Parker, & G. J.Taylor, 1994) is a self-assessment instrument designed to measure deficits in meta-emotional functioning (e.g., difficulties in identifying and describing emotions).Four studies were carried out to examine the association between the TAS-20 and(a) performance measures of meta-emotional functioning (memory of emotion andemotional awareness; Studies 1–2) and (b) measures of perfectionistic standardsthat may possibly be involved in the self-assessment of abilities–difficulties (Stud-ies 3–4). The TAS-20 failed to correlate in the predicted direction with the per-formance measures but showed sizable correlations with measures of perfection-ism. Moreover, perfectionism was found to predict TAS-20 scores independent of depression, anxiety, and somatic complaints. The results are discussed in terms of
the TAS-20 measuring primarily certain aspects of meta-emotional self-efficacy.
The psychology of emotion is a complex field,which is only incompletely understood. One basic dis-tinction that can be made within this field is betweenemotional processing and meta-emotional processing.
Most research on emotion so far has focused on theformer (i.e., how various events are processed emo-tionally, that is, physiologically [e.g., in terms of au-tonomic arousal], cognitively [e.g., in terms of pri-mary and secondary appraisals], and behaviorally[e.g., in terms of approach or avoidance]). But emo-
tions themselves also undergo cognitive processing(i.e., an individual’s emotional responses at thesevarious levels constitute information that may un-dergo more or less additional cognitive processing).This meta-emotional processing involves identifying,labeling, and describing emotions; remembering emo-tions; reasoning about the emotions that one may feelin various hypothetical situations; analyzing the emo-tional consequences of various kinds of behaviors;empathizing with others’ emotional experiences, andso on. The development of this field of research re-quires both well-grounded theoretical constructs and
valid methods for measuring meta-emotional process-ing. The present series of studies1 focus on one of thefirst theoretical constructs to evolve in this area, thatis, alexithymia (Sifneos, 1973; Taylor, Bagby, &Parker, 1997), and the most well-established methodfor operationalizing this construct, that is, the TorontoAlexithymia Scale (TAS-20; Bagby, Parker, & Tay-lor, 1994).
The construct of alexithymia developed from clini-cal observations that patients with psychosomatic dis-
orders had difficulty talking about feelings and fanta-sies when assessed in psychodynamically orientedinterviews (Sifneos, 1973). This led to a definition of alexithymia that emphasized difficulties in identifyingand describing feelings (and in differentiating be-tween feelings and bodily sensations), in combinationwith an externally oriented thinking style, that is, arelative absence of internal fantasies and a corre-spondingly increased focus on details of externalevents and bodily symptoms (Nemiah, Freyberger, &Sifneos, 1976; Taylor et al., 1997). It is important tonote that the alexithymia construct from the beginninghas been tied to a deficit model; that is, an individu-
al’s “alexithymic communication” (i.e., talking littleabout feelings and fantasies and more about externalfacts and symptoms) has been attributed to a deficientability to identify and describe emotions and to a defi-cit in imaginal capacity, rather than to defensive cop-
Lars-Gunnar Lundh, Anders Johnsson, Karin Sundqvist,and Helen Olsson, Department of Psychology, StockholmUniversity, Stockholm, Sweden.
Correspondence concerning this article should be ad-dressed to Lars-Gunnar Lundh, who is now at the Depart-ment of Social Sciences, Mid Sweden University, Oster-sund, SE-831 25 Sweden. E-mail: [email protected]
1 We thank Peter Salovey and anonymous reviewers forvaluable comments on an earlier version of this article.
Emotion Copyright 2002 by the American Psychological Association, Inc.2002, Vol. 2, No. 4, 361–379 1528-3542/02/$5.00 DOI: 10.1037//1528-3542.2.4.361
361
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 2/19
ing or motivational reasons. In Taylor et al.’s (1997)formulation, for example, alexithymia is assumed toinvolve an impaired capacity to construct mental rep-resentations of emotions—a capacity that is neededboth for the cognitive processing of emotional expe-riences and for the verbal communication of emo-tional experiences to other people.
Problems in the Measurement of Alexithymia
Alexithymia is a theoretical construct, and one mayquestion whether any single measure is likely to serveas a sufficient index of this construct. We may there-fore need multiple measures. A number of instru-ments have been developed to measure alexithymia:(a) interview measures (e.g., the Beth Israel HospitalPsychosomatic Questionnaire; BIQ; Sifneos, 1973);(b) self-assessment measures (e.g., the Toronto Alexi-
tymia Scale; TAS-20; Bagby, Parker, & Taylor, 1994;Bagby, Taylor, & Parker, 1994); (c) observer mea-sures (e.g., the Observer Alexithymia Scale; OAS;Haviland, Warren, & Riggs, 2000). In addition, (d) anumber of performance measures have been devel-oped which index meta-emotional capacities morepositively, such as Levels of Emotional AwarenessScale (LEAS; Lane, Quinlan, Schwartz, Walker, &Zeitlin, 1990) and various measures of emotional in-telligence like the Multi-Factor Emotional Intelli-gence Scale (MEIS; Mayer, Caruso, & Salovey,1999), and the Mayer–Salovey–Caruso Emotional In-telligence Test (MSCEIT; Mayer, Caruso, & Salovey,
2000).All these measures are subject to various sources of error. For example, interview measures such as theBIQ rely on the interviewer’s observations of the per-son’s meta-emotional communication in the interview(i.e., the extent to which he or she describes and com-municates feelings and other internal experiences tothe interviewer, as compared with details of externalevents and physical symptoms), and people may showa lack of meta-emotional communication in the inter-view situation for a number of different reasons. Al-though such a lack of meta-emotional communicationmay be caused by alexithymia, that is, a deficient
ability to describe and communicate feelings, it mayalso appear, for example, (a) as a way of coping withmore or less severe forms of stress or (b) because of a lack of motivation to talk about feelings in the in-terview. The former alternative was recognized earlyunder the label of secondary alexithymia by Frey-berger (1977), who observed that some medically illpatients showed a constriction of emotional expres-
sion and imaginal activity as a result of their illness.An example of the latter possibility would be if pa-tients with somatic problems believe that their healthproblems have a medical rather than a psychologicalexplanation, and therefore are less motivated to speakabout feelings, dreams, and fantasies in the interviewsituation (Mortazavi, 2001).
Self-assessment methods also have their problems.The instrument that is generally acknowledged to bethe best-validated measure of alexithymia today is theTAS-20. The TAS-20 is a multidimensional, 20 iteminstrument with a three-factor structure that is as-sumed to capture three separate, though empiricallyrelated, facets of the alexithymia construct: (a) diffi-culty identifying feelings and distinguishing themfrom bodily sensations, (b) difficulty describing feel-ings to others, and (c) an externally oriented style of thinking. Whereas the first two TAS-20 subscales re-
quire the individual to assess his or her meta-emotional abilities–difficulties, the third factor asksabout characteristics of his or her style of thinking(e.g., being more focused on external facts than oninternal experiences). One problem is that self-assessment of traits in general relies on informationthat is retrieved from semantic memory (Klein, Ba-bey, & Sherman, 1997; Klein & Loftus, 1993; Klein,Sherman, & Loftus, 1996). What is measured by self-assessment instruments of alexithymia, therefore, isprimarily the individual’s beliefs, as formed by theaccessible memory information, about his or her func-tioning. This means that alexithymia can be measured
by self-assessment only to the extent that there is validmemory information of this kind available, at all lev-els of alexithymia. This seems particularly problem-atic with regard to the first two TAS-20 subscales,which require that individuals with alexithymia knowtheir difficulties in identifying and describing emo-tions (i.e., although their meta-emotional processingabilities are deficient, they must be meta-emotionallyaware of these deficiencies). This is the first potentialsource of error in using the TAS-20 as an operation-alization of alexithymia: To the extent that individualswith these kinds of meta-emotional difficulties lackknowledge about their difficulties, they may be ex-
pected to receive too-low scores on the TAS-20. And,conversely, an individual who has too little faith in hisor her meta-emotional abilities is likely to receivetoo-high scores on the TAS-20.
More generally, the first two TAS-20 subscales canbe said to measure aspects of a person’s perceivedmeta-emotional self-efficacy, or more correctly, his orher perceived lack of meta-emotional self-efficacy.
LUNDH, JOHNSSON, SUNDQVIST, AND OLSSON362
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 3/19
Self-efficacy (Bandura, 1986) refers to the person’sself-perceived competence in a certain area. A per-son’s self-efficacy beliefs may show a more or lessgood match with his or her real capacities but shouldnever be assumed to correspond perfectly with these.There are a number of different aspects of meta-emotional self-efficacy, for example, the perceivedability to identify emotions, to communicate emo-tions, and to manage upsetting emotions. The TAS-20Factors 1 and 2 can be seen as indexes of the first twoaspects of meta-emotional self-efficacy.
Another problem relates to mood-congruent re-trieval; that is, people are better able to retrieve in-formation that is of the same affective tone as theircurrent mood (Blaney, 1986; Ellis & Moore, 1999).Because beliefs about “being able” have a positiveaffective tone, whereas beliefs about “being unable”have a negative affective tone, it may be expected that
a positive mood may make it easier to retrieve beliefsabout positive abilities (e.g., the ability to identify anddescribe emotions), whereas a negative mood maymake it easier to retrieve information about difficul-ties and deficiencies in the same regard. This is asecond potential source of error in using the TAS-20to measure alexithymia: If a positive mood makes iteasier to retrieve positive information related to one’smeta-emotional skills, individuals in a positive moodare likely to receive too-low scores on the TAS-20.Conversely, if a negative mood makes it easier toretrieve negative information related to one’s meta-emotional skills, individuals in a negative mood are
likely to receive too-high scores on the TAS-20. Pre-vious research (e.g., Bagby, Parker, & Taylor, 1994;Lumley, Ovies, Stettner, Wehmer, & Lakey, 1996;Lundh & Simonsson-Sarnecki, 2001) has clearlyshown that the TAS-20 correlates positively withmeasures of negative affect, and negatively with mea-sures of positive affect. Although there are severalpossible explanations for these correlations, they arecompatible with the hypothesis that people who are ina negative emotional state judge their abilities to iden-tify and describe emotion more negatively, and peoplein a positive emotional state judge their abilities morepositively. Evidence clearly indicates that, although
self-assessed alexithymia is not simply a function of degree of depression (e.g., Luminet, Bagby, & Taylor,2001; Parker, Bagby, & Taylor, 1991), there is at leasta subgroup of depressed patients who score high onthe TAS-20 because of their depression (Honkalampi,Hintikka, Saarinen, Lehtonen, & Viinamaki, 2000).Similarly, large-scale follow-up studies have identi-fied subgroups of individuals in the general popula-
tion whose TAS-20 scores vary as a function of theirdegree of depression (Honkalampi et al., 2001).
A third problem is that the self-assessment of ca-pabilities and difficulties involves an element of evaluation. The evaluation of performance is gener-ally made relative to some kind of personal standards(e.g., Bandura, 1986), and a crucial question is there-fore how stringent are standards people have for judg-ing their performance? As there is no reason whyself-assessment of the ability to identify and describeemotions should be any exception, these personalstandards are likely to influence the individual’s re-sponses to TAS-20, and in particular the TAS-20 Fac-tors 1 and 2. On the one hand, individuals with highpersonal standards for what counts as being good atidentifying and describing emotions are likely to re-ceive too-high scores on the TAS-20; on the otherhand, individuals with low standards in this regard are
likely to receive too-low scores.Although performance measures like the LEAS andthe MEIS may be assumed to measure meta-emotional capacities in a more direct way, they arealso subject to various problems. One problem is howto operationalize “performance” in the field of meta-emotion, because it is difficult to construct test situ-ations where responses are unambiguously correct orincorrect. The LEAS solves this problem by definingperformance in terms of structural characteristics of performance (e.g., so that more differentiated descrip-tions of emotions are given higher scores than lessdifferentiated descriptions, and descriptions of blends
of emotions are given higher scores than descriptionsof simple emotions), whereas the MEIS relies on aconsensus definition of correctness (i.e., higher scoresare given to responses that are produced by a largernumber of individuals). Further, there is also the ques-tion as to what extent performance on these tests doesadequately reflect the person’s meta-emotional com-petence, as distinct from effects of situational contextsand temporary cognitive, emotional, and motivationalstates. Relatively little research has been carried outwith these instruments, and not much is known, forexample, about their test–retest reliability.
A reasonable hypothesis is that all these various
kinds of instruments measure different aspects of meta-emotional functioning, and that they are there-fore complementary. To increase our understandingof their respective contribution to the measurement of meta-emotional functioning, we need to study theirintercorrelations and their correlations with other vari-ables (e.g., personal standards) that may possibly in-fluence people’s responses to these instruments. The
ALEXITHYMIA, META-EMOTION, AND PERFECTIONISM 363
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 4/19
TAS-20 has so far been validated primarily in relationto other self-assessment measures. As can be ex-pected, the TAS-20 has been found to correlate mod-erately to highly with conceptually and operationallysimilar constructs (e.g., self-assessment measures of emotional intelligence, psychological mindedness, re-ceptivity to feelings; for reviews, see Taylor et al.,1997; Taylor & Bagby, 2000). Although this testifiesto a certain coherence of people’s beliefs about them-selves in the area of meta-emotional functioning, it isless informative about the validity of self-assessmentmeasures of meta-emotional functioning. What weneed here are studies that use multiple methods of measuring meta-emotional functioning, that is, bothself-assessment measures and instruments that do notrely on self-assessment (e.g., performance measures,observer measures, or interview measures).
Self-assessment measures and performance mea-
sures have shown low correlations both in traditionalresearch on intelligence (Paulhus, Lysy, & Yik, 1998)and in research on emotional intelligence (Davies,Stankov, & Roberts, 1998). The results with regard tothe TAS (TAS-20 or earlier versions of the TAS) arecontradictory. Whereas some studies have failed tofind a significant association between the TAS and theaccuracy of recognition of emotions depicted in hu-man faces, color swatches, and abstract designs(Mayer, DiPaolo, & Salovey, 1990), or the accuracyof recognition of emotional content in stories (Beren-baum, Davis, & McGrew, 1998; Berenbaum &Prince, 1994), other studies have reported a signifi-cant association between the TAS-20 and decreasedrecognition of facial expressions of emotion (Jessimer& Markham, 1997; Mann, Wise, Trinidad, & Kohan-ski, 1994; Parker, Taylor, & Bagby, 1993; Prkachin &Prkachin, 2001) and a decreased ability to match ver-bal and nonverbal emotional stimuli (Lane et al.,1996). Lumley et al. (2002) studied the correlationsbetween the TAS-20 and three other measures of emotional abilities (the OAS, the LEAS, and theMSCEIT) in a sample of young adults: Although theTAS-20 showed moderate correlations in the expecteddirection with both the OAS (as rated by the partici-
pants’ parents) and the MSCEIT, none of the mea-sures correlated significantly with the LEAS. Finally,with regard to interview measures of alexithymia, sig-nificant positive correlations between the TAS-20 anda modified version of the BIQ have been reported(e.g., Bagby, Taylor, & Parker, 1994). This suggests acomplex picture, with the TAS-20 showing clear evi-dence of being a useful measure of meta-emotional
functioning, although there are also a number of con-tradictory results that necessitate further research.
Purpose of the Present Studies
The purpose of the present studies was to contributeto the further understanding of the construct validityof the TAS-20 by studying its association (a) withother measures of meta-emotional functioning that donot depend on self-assessment (Studies 1–2) and (b)with measures of perfectionistic standards that maypossibly be involved in the self-assessment of abili-ties–difficulties (Studies 3–4).
According to Taylor et al. (1997), individuals withalexithymia “in most instances, are unable to link[feelings] with memories” (p. 29). No empirical studyon this subject has so far been reported in the litera-ture. The purpose of Study 1, therefore, was to test
how the TAS-20 is related to latency to retrieve emo-tional memories on an autobiographical memory test(AMT) that was originally designed by Williams andBroadbent (1986). In Study 2, the TAS-20 was stud-ied for its associations with the LEAS (Lane et al.,1990), which is a performance-based test that requiresthe individual to describe his or her own and anotherperson’s possible emotional responses to a number of different scenarios. Studies 3 and 4 focus on the re-lation between self-assessed alexithymia and perfec-
tionism, defined as the desire to achieve high stan-dards of performance, in combination with overlycritical evaluations of one’s performance (Frost, Mar-
ten, Lahart, & Rosenblate, 1990). Perfectionism andalexithymia may be associated because of a number of different reasons. Not only would such a correlationbe expected if the self-assessment of alexithymia isinfluenced by degree of perfectionism, but also if in-dividuals with alexithymia develop perfectionistictendencies to compensate for their shortcomings, or if the two traits are associated because both have someother kind of common causes (e.g., negative affect).To explore the association between perfectionism andalexithymia, Studies 3 and 4 capitalize on the three-factor structure of the TAS-20, where the first twofactors are designed to measure difficulties identify-
ing and describing emotions, and the third factor mea-sures an externally oriented style of thinking that doesnot rely on a self-assessment of abilities or difficul-ties. If perfectionism has a causal influence on theself-assessment of meta-emotional difficulties–abilities, then perfectionism should correlate posi-tively with TAS-20 Factors 1 and 2; this hypothesis,however, gives us no reason to assume that it should
LUNDH, JOHNSSON, SUNDQVIST, AND OLSSON364
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 5/19
correlate similarly with Factor 3. If alexithymia leadsto perfectionism, however, or if both are caused bysome third factor, all three facets of alexithymia (i.e.,all three TAS-20 factors) may be expected to correlatewith perfectionism. Studies 3 and 4 were thereforedesigned to test the hypotheses (a) that perfectionismcorrelates specifically with TAS-20 Factors 1 (diffi-culties identifying emotions and distinguishing themfrom bodily sensations) and 2 (difficulties describingemotions to others), but not with Factor 3 (an exter-nally oriented style of thinking); and (b) that this as-sociation holds when negative affect is controlled for.
Study 1
According to Taylor et al. (1997), alexithymia re-flects a deficit in emotional regulation that, amongother things, is associated with difficulties connectingemotions with memories and specific situations. In
fact, it may even be argued that alexithymia should,by definition, involve an impaired capacity to remem-ber emotions, because to the extent that a person hasdifficulties identifying and describing emotions, he orshe is likely to use less emotional concepts when en-coding emotional episodes in memory, and emotionwords will consequently serve as less-efficient cuesfor the retrieval of these episodes.
One way of studying the accessibility of emotionalmemories is to measure the latency to retrieve memo-ries to cue words designating emotions. From thisperspective, it may be hypothesized that the morealexithymic an individual is, the more difficult it will
be for him or her to retrieve memories of situationswhere various emotions were experienced, and thelonger it will take to retrieve such memories. Thelatency to retrieve emotional memories has been stud-ied in connection with depression and dysphoria.Lloyd and Lishman (1975) studied clinically de-pressed patients and found that the more severe thedepression, the quicker the patient retrieved an un-pleasant memory. Partly similar results were reportedin a nonclinical sample by Teasdale and Fogarty(1979), who found that latencies to remember positiveor negative personal events were biased by mood,although most of this was due to slowed recall of
positive material in depressed mood rather thanspeeded recall of negative material. In contrast to de-pression, alexithymia may be expected to be associ-ated with prolonged latencies to all kinds of emotionwords, independent of valence. To test this hypoth-esis, the present study made use of an autobiographi-cal memory test that was devised by Williams andBroadbent (1986).
Method
Participants
Eighty-eight individuals (67 women and 21 men)participated in the study. Of these, 49 (40 women and9 men) were undergraduate students at StockholmUniversity, and 39 (27 women and 12 men) wererecruited from a community sample of individualsliving in Stockholm that was drawn randomly fromthe Swedish Government’s Person and Address Reg-ister (SPAR). The mean age of the participants was29.1 (SD 10.3) years. The student participants re-ceived partial course credit for their participation, andthe external participants received 150 Swedish crowns(approximately $15) for their participation.
Materials
TAS-20. The TAS-20 (Bagby, Parker, & Taylor,
1994; Bagby, Taylor, & Parker, 1994) is a multidi-mensional instrument with a three-factor structure thatis assumed to capture three separate facets of thealexithymia construct: (a) difficulty identifying emo-tions and distinguishing them from bodily sensations,(b) difficulty describing emotions to others, and (c) anexternally oriented style of thinking. The presentstudy used a Swedish translation of the TAS-20,which has shown good internal reliability and hasbeen found to replicate the three-factor TAS-20 modelin a Swedish sample (Simonsson-Sarnecki et al., 2000).
Bec k Dep ressio n Inv ent ory (BD I). The BDI(Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) is
a 21-item self-report questionnaire that assesses thepresence and severity of cognitive, behavioral, affec-tive, and somatic symptoms of depression. Each itemconsists of a 0–3 self-rating, with increasing scoresindicating greater symptom severity. The BDI is in-ternally consistent, has high test–retest reliability, andhas been shown to be valid with both psychiatric andnormal samples (Beck, Steer, & Garbin, 1988).
Karolinska Scales of Personality (KSP) Anxiety
and Social Desirability Scales. The present studyused two measures derived from the KSP (Gustavs-son, Weinryb, Goransson, Pedersen, & Åsberg,1997): a combined trait anxiety score, based on two
10-item scales (Psychic Anxiety and Somatic Anxi-ety), and a 10-item version of the Marlowe–CrowneSocial Desirability Scale (Crowne & Marlowe, 1960).All items have a four-choice format. These scaleshave shown good reliability and stability (Gustavssonet al., 1997).
AMT. This is a Swedish version of a test that wasoriginally designed by Williams and Broadbent
ALEXITHYMIA, META-EMOTION, AND PERFECTIONISM 365
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 6/19
(1986). In this task, 10 emotion words (5 positive and5 negative) are presented successively to the indi-vidual, who is asked to report the first memory of asituation where this emotion was experienced. Thepositive and negative emotion words are presented ina given order, with positive and negative words alter-nating: “happy,” “sorry,” “safe,” “angry,” “inter-ested,” “clumsy,” “successful,” “hurt,” “surprised,”and “lonely.” Each word was written on a separatecard, and the experimenter pronounced the word atthe same time that the card was exposed to the par-ticipant. The participant then had 60 s to produce amemory. If the participant did not produce a suffi-ciently specific memory, the instruction was repeated.
The entire procedure was tape recorded, and la-tency of retrieval was measured by listening to thetape afterward. If no specific memories were pro-duced within 60 s, the participant was assigned a la-
tency of 60 s. The total response time to the 10 words,therefore, ranged from 0 to 600 s. The reliability of the latency ratings was studied by having two raters
judge the retrieval latencies for 23 of the 88 partici-pants. The correlation between the total responsetimes (i.e., to all 10 words) for the two raters was .99.The internal consistency of the retrieval latencies tothe 10 emotional words was acceptable (Cronbach’s .69).
Procedure
All participants were tested individually, and thevarious measures were administered in the same order
for all individuals: first the AMT, then the KSP scales,the TAS-20, and finally the BDI. Testing was done bytwo experimenters, who used a written manual foradministering the AMT and who trained together tomake sure that the procedure was administered in thesame way for all participants. The two experimentersrated the AMT latencies, and the interrater reliability(see above) was judged on the basis of their ratings.
Results and Discussion
Descriptive Data
The participants scored 39.6 (SD 8.0) on theTAS-20 and 6.5 (SD 5.6) on the BDI. The menscored significantly higher than the women on theTAS-20 (43.5 vs. 38.3), t (86) 2.65, p < .01,whereas there were no gender differences on any of the other measures. The student subsample scored sig-nificantly lower than the community subsample on theTAS-20 (37.5 vs. 42.1), t (86) −2.77, p < .01, butthere were no significant differences between the sub-samples on any other measure.
Associations Between the TAS-20 and theOther Variables
As can be seen in Table 1 the correlational analysisshowed no support for the hypothesis. Although the
TAS-20 correlated positively with depression, itshowed no significant correlation with latency to re-trieve autobiographical memories on the AMT. Noneof the measures showed any significant correlationwith social desirability. When latencies were com-puted separately for positive and negative emotionalwords, the same lack of associations appeared.
None of the three TAS factors showed any ten-dency to correlate with the AMT (correlations rangingfrom −.05 to .11). The BDI correlated at .49 withFactor 1 (Difficulty Identifying Emotions) and at .31with Factor 2 (Difficulty Describing Emotions), butdid not correlate (r −.13, ns) with Factor 3 (Exter-
nally Oriented Thinking).One possibility is that differences in emotional
memory are apparent only in participants with highTAS-20 scores. To test this possibility, the 11 partici-pants who scored 50 or more on the TAS-20 werecompared with the rest of the group (who scored 48or less). This high-TAS group showed slightly longerlatencies to retrieve emotional memories than the rest
Table 1 Zero-Order Correlations Between TAS-20, Depression, Social Desirability, and Latencies on the Autobiographical
Memory Test
Instrument 1 2 3 4 5
1. TAS-20 —2. Beck Depression Inventory .33* —3. KSP Trait Anxiety .40** .33* —4. KSP Social Desirability Scale .13 .02 .08 —5. Autobiographical Memory Test .08 .05 −.06 .01 —
Note. TAS-20 Toronto Alexithymia Scale (Bagby, Parker, & Taylor, 1994; Bagby, Taylor, & Parker, 1994); KSP Karolinska Scalesof Personality (Gustavsson, Weinryb, Goransson, Pedersen, & Åsberg, 1997).* p < .01. ** p < .001.
LUNDH, JOHNSSON, SUNDQVIST, AND OLSSON366
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 7/19
of the group, 185.3 s versus 161.3 for all ten words.
This effect, however, was far from significant
( p > .30).
The fact that there were very few participants who
scored high on the TAS-20 in the present study must
be taken into account when interpreting the results.
However, if Taylor et al. (1997) are right in their
assumption that alexithymia is a dimensional con-
struct, then the predicted associations between TAS-
20 and other measures may be expected to show up
also in normal, nonclinical samples. It should be noted
that the TAS-20 showed correlations with depression
and anxiety (r .33 and r .40) in the present study
that were of approximately the same magnitude as
found in other research (e.g., Bagby, Parker, & Tay-
lor, 1994, Lundh & Simonsson-Sarnecki, 2001), and
also showed the typical association with gender (i.e.,
men scoring higher than women) and education, that
is, university students scoring lower than others, ashas been found in other studies (e.g., Lane, Sechrest,
& Riedel, 1998; Salminen, Saarijarvi, Aarela, Toikka,
& Kauhanen, 1999). Because the TAS-20 was asso-
ciated in this highly typical manner with depression,
anxiety, gender, and education, it is unlikely that the
lack of an association with retrieval latencies of emo-
tional memories should be due to peculiarities of the
present sample.
Study 2
Another approach to the study of emotional aware-
ness, that is not dependent on self-assessment, is rep-resented by Lane et al.’s (1990) LEAS. The LEAS
differs from the TAS-20 in that it does not require the
respondents to rate their abilities, but involves a more
direct measure of performance. The LEAS instructs
the respondent to imagine a number of scenes, each of
which involves two persons (“you” and another per-
son) and is followed by two questions: “How would
you feel?” and “How would the other person feel?”That is, it operationalizes “emotional awareness” not
in terms of emotion perception, but in terms of the
ability to imagine hypothetical emotions. The LEAS
has been found to correlate with measures of ego de-
velopment and the cognitive complexity of descrip-tions of parents (Lane et al., 1990) as well as self-
restraint and impulse control (Feldman Barrett, 1999).
The validity of the LEAS, however, is also seen in its
being associated with the perception of emotion. Lane
et al. (1996) administered the TAS-20 and the LEAS
to 380 participants in a community survey, together
with a Perception of Affect Task (PAT; Rau, 1993), a
140-item measure of the ability to match verbal and
nonverbal emotional stimuli. They found that the
LEAS accounted for 18.4% of the variance in the
accuracy of recognition of verbal and nonverbal emo-
tional stimuli in the PAT, whereas the TAS-20 ac-
counted for 10.5% of the variance in these dependent
measures. That is, there is evidence that the LEAS
indexes not only the ability to reason in terms of
hypothetical emotions, but emotional awareness in a
broader sense, including emotion perception.
At the same time, Lane et al. (1996) found only a
weak correlation (r −.19) between the LEAS and
the TAS-20, which suggests that the two instruments
measure, at best, weakly overlapping constructs. A
similar correlation between the two scales (r −.19)
was also reported in a German study (Suslow,
Donges, Kersting, & Arolt, 2000), whereas Lumley et
al. (2002) found no significant correlation between
these two measures. One substantial difference be-tween the two instruments is that whereas the TAS-20
correlates positively with anxiety, depression, and
negative affectivity in general (Bagby, Taylor &
Parker, 1994; Lundh & Simonsson-Sarnecki, 2001),
the LEAS shows less evidence of correlating with
negative affect. For example, Lane et al. (1996) re-
ported that the Taylor Manifest Anxiety Scale (Tay-
lor, 1953) correlated positively (r .32, p < .001)
with the TAS-20, but showed no correlation with the
LEAS (r .07). And although Suslow et al. (2000)
found that the TAS-20 correlates substantially with
various aspects of shame (r s ranging from 0.40 to
0.69), the LEAS showed no significant correlationswith any Shame scale.
The purpose of the present study was to examine
the relation between the TAS-20 (and its three sub-
scales) and the LEAS in a Swedish sample, along with
their respective associations with various aspects of
negative affectivity: depression, trait anxiety, and
anxiety sensitivity. Anxiety sensitivity (i.e., fear of
anxiety-related body sensations; Reiss, Peterson, Gur-
sky, & McNally, 1986) has recently been shown
(Devine, Stewart, & Watt, 1999) to correlate with the
TAS-20 (and its first two subscales), but there is so far
no study on the relation between the LEAS and anxi-
ety sensitivity.
Method
Participants
Seventy-eight undergraduate students (65 women
and 13 men) participated in the study. Their mean age
ALEXITHYMIA, META-EMOTION, AND PERFECTIONISM 367
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 8/19
was 27.7 (SD 7.1) years, and they received partial
course credit for their participation.
Materials
The present study used the same versions of the
TAS-20, the BDI, and the KSP Social Desirability
scale as used in Study 1. In addition, it used the fol-
lowing instruments.
LEAS. The LEAS (Lane et al., 1990) consists of
20 scenes, each described in two–four sentences, and
each involving two persons. Each scene is presented
at the top of a separate page, followed by two ques-
tions: “How would you feel?” and “How would the
other person feel?” The participants are instructed to
write their responses on the remainder of the page and
to use as much or as little of the page as needed to
answer the two questions.
Responses are scored separately for each scene,
with separate scores for the emotions described forself and for other, in accordance with the LEAS Scor-
ing Manual and Glossary (Lane, 1991). This scoring
manual includes a glossary of words at each level and
was translated and adapted into Swedish by Lars-
Gunnar Lundh and Karin Sundqvist. According to this
manual, Level 0 is scored when no response is given
or when nonemotion responses are given where the
word “feel” is used to describe a thought or a cogni-
tive state (e.g., “puzzled” or “uncertain”) rather than a
feeling. Level 1 is scored when responses are given
that indicate an awareness of physiological cues (e.g.,
“I’d feel tired.”). Level 2 is scored when the partici-
pant uses words that convey relatively undifferenti-ated emotions (e.g., “I’d feel bad”), or an action ten-
dency (e.g., “I’d feel like punching the wall.”). Level
3 is scored when words are used that convey typical,
differentiated emotion (e.g., “happy,” “sad,” “angry”).
The highest score for self and other (Level 4) is scored
when two or more Level 3 words are used to convey
greater emotional differentiation than either word
alone. Each participant thus receives one score for the
self response and one for the other response, from 0 to
4. In addition, a third total score is used, defined as
equal to the higher of these two scores—except in
cases where both self and other receives four scores.
Under these circumstances, a total score of 5 is givenfor the scene if the emotions for self and other can be
differentiated from each other.
In Lane et al.’s (1990) original study, 20 protocols
were independently scored by two raters. Interrater
reliability for the total score on the LEAS was high,
r (20) .84, as was internal consistency (Cronbach’s
.81). The Swedish version of the LEAS showed
an equally good reliability: interrater reliability was r
.88, and Cronbach’s alpha was .85. Two partici-
pants failed to complete one scene with a scorable
response; the mode values of these 2 participants were
imputed for these scenes.
Beck Anxiety Inventory (BAI). The BAI (Beck,
Brown, Epstein, & Steer, 1988) is a 21-item self-
report questionnaire, with good reliability and high
test–retest stability, that assesses the presence and se-
verity of various symptoms of anxiety. Each item con-
sists of a 0–3 self-rating, with increasing scores indi-
cating greater symptom severity.
Anxiety Sensitivity Index (ASI). The ASI (Reiss et
al., 1986) is a reliable and valid 16-item self-report
questionnaire that assesses an individual’s fear of
anxiety-related body sensations (e.g., rapid heartbeat,
nausea, shortness of breath) on the basis of his or her
belief that these symptoms have negative conse-
quences. Respondents rate the degree to which theyagree or disagree with each item on a Likert scale
ranging from 0 (very little) to 4 (very much). Total
scores are computed by summing the ratings across
the 16 items.
Procedure
The participants were tested in groups, and the vari-
ous measures were administered in the same order for
all individuals: first the LEAS, then the TAS-20, the
KSP Social Desirability Scale, the ASI, the BDI, and
finally the BAI.
Results and Discussion
Descriptive Data
The participants scored 42.0 (SD 9.1) on the
TAS-20, 68.3 (SD 9.0) on LEAS, 7.3 (SD 5.5)
on the BDI, 9.8 (SD 6.7) on the BAI, and 18.7 (SD
9.2) on the ASI.
Associations Between the TAS-20, LEAS, and the Other Variables
As seen in Table 2, the TAS-20 showed a weak
positive correlation ( p < .05) with the LEAS (i.e., the
higher the participants scored on the TAS-20, the
higher they also scored on emotional awareness).As also seen in Table 2, Difficulty Identifying
Emotions was the only TAS-20 subscale to correlate
significantly with measures of negative affectivity
(depression, anxiety, and anxiety sensitivity). Neither
the LEAS nor any of the other TAS-20 subscales
showed any significant positive correlation with these
measures. The TAS-20 subscale Externally Oriented
LUNDH, JOHNSSON, SUNDQVIST, AND OLSSON368
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 9/19
Thinking even tended to correlate negatively with
both depression and anxiety.Inspection of the scatter plot of the correlation be-
tween the TAS-20 and the LEAS showed that the
positive correlation between these two measures was
due both to participants who scored low at both mea-
sures and to participants who scored high at both mea-
sures. To explore possible characteristics of individu-
als with such “contradictory” scores on the TAS-20
and the LEAS (i.e., “double-lows” and “double-
highs”), we classified the participants into three
equally large groups on the basis of their scores on
each of these instruments. This produced a TAS clas-
sification with a low-TAS group (TAS-20 < 38, n
27), a medium-TAS group (TAS-20 38–44, n 25),and a high-TAS group (TAS-20 > 44, n 26) and a
LEAS classification with a low-LEAS group (LEAS <
66, n 27), a medium-LEAS group (LEAS 66–72, n
25), and a high-LEAS group (LEAS > 72, n 26).
Cross-tabulation of these terciles resulted in the iden-
tification of 11 double-highs and 14 double-lows.
As seen in Table 3, the double-highs scored con-
sistently higher than the whole sample mean on the
BDI, the BAI, the ASI, and KSP Social Desirability;
although none of these comparisons approached sta-
tistical significance ( p s ranging between .10 and .20),
all effects were in the same direction and not negli-
gible (effect sizes ranging from z .43 to z .67).Although the double-lows scored lower than the
whole sample mean on all three measures of negative
affectivity, these effects were weak (effect sizes rang-
ing from z .06 to z .35), and the effect on social
desirability was not in the same direction.
To summarize, the present results replicated the
previous finding that the TAS-20, and particularly the
Difficulties Identifying Emotions subscale, correlates
with measures of negative affectivity, whereas theLEAS does not. The results did not, however, repli-
cate Lane et al.’s (1996) finding of a weak negative
correlation between the TAS-20 and the LEAS. On
the contrary, the present sample showed a weak posi-
tive correlation between these two measures. Further,
the results show clear evidence of subgroups that
score inconsistently on the TAS-20 and the LEAS—that is, respondents who score high on both alexi-
thymia and emotional awareness, and respondents
who score low on both alexithymia and emotional
awareness. In the present study, the former subgroup
tended to score higher on depression, trait anxiety,
Table 3
Comparison Between Groups With Contradictory Scores
on the TAS-20 and the LEAS, and the Total Sample
Instrument
Double-highs
(n 11)
Double-lows
(n 14)
Whole sample
(n 78)
M SD M SD M SD
TAS-20 53.0 05.7 32.8 3.0 42.0 9.1
LEAS 79.4 06.4 58.1 3.9 68.3 8.9
BDI 10.9 08.3 5.9 4.0 7.3 5.4
BAI 13.1 07.9 9.4 7.0 9.8 6.7
ASI 24.0 12.4 15.5 8.7 18.7 9.2
KSP-SD 26.8 02.9 25.4 4.0 25.1 4.0
Note. TAS-20 Toronto Alexithymia Scale (Bagby, Parker, &Taylor, 1994; Bagby, Taylor, & Parker, 1994); LEAS Levels of Emotional Awareness Scale (Lane, Quinlan, Schwartz, Walker, &Zeitlin, 1990); BDI Beck Depression Inventory (Beck, Ward,Mendelson, Mock, & Erbaugh, 1961); BAI Beck Anxiety In-ventory (Beck, Brown, Epstein, & Steer, 1988); ASI AnxietySensitivity Index (Reiss, Peterson, Gursky, & McNally, 1986);KSP-SD Karolinska Scales of Personality (Gustavsson, Wein-ryb, Goransson, Pedersen, & Åsberg, 1997) and Social DesirabilityScales.
Table 2
Zero-Order Correlations Between the Toronto Alexithymia Scale (TAS-20), Emotional Awareness, Depression, Anxiety,
Anxiety Sensitivity, and Social Desirability
Instrument 1 2 3 4 5 6
1. TAS-20 — .24 .16 .15 .25 .08
F1: Difficulty Identifying Emotions .16 .41** .37* .30* .02F2: Difficulty Describing Emotions .17 .14 .07 .19 .06
F3: Externally Oriented Thinking .19 −.28 −.20 .01 .11
2. Levels of Emotional Awareness — .03 .03 .18 .02
3. Beck Depression Inventory — .61** .35* −.12
4. Beck Anxiety Inventory — .48** .00
5. Anxiety Sensitivity Index — .16
6. KSP-SD —
Note. KSP-SD Karolinska Scales of Personality (Gustavsson, Weinryb, Goransson, Pedersen, & Åsberg, 1997) and Social DesirabilityScales; F1, F2, F3 Factor 1, Factor 2, and Factor 3.* p < .01. ** p < .001.
ALEXITHYMIA, META-EMOTION, AND PERFECTIONISM 369
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 10/19
and anxiety sensitivity, which is consistent with the
hypothesis that people who are in a negative emo-
tional state will underestimate their ability to identify
and describe emotions, and therefore receive too-high
scores on the TAS-20. However, they also tended to
score higher on social desirability, which points to
other possible explanations involving some kind of
response bias; for example, a response bias in the
form of an eagerness to please the experimenter by
exerting themselves to perform well on the LEAS
(thus producing too-high scores on the LEAS, as com-
pared with other participants who do not care to per-
form equally well).
How should we understand the subgroup who
scored low on both alexithymia and emotional aware-
ness? The present results do not give much of a clue
to the understanding of this kind of inconsistency.
One possibility is that these individuals are relatively
carefree individuals who do not worry especiallyabout personal shortcomings (which may lead to too-
low scores on the TAS-20) and/or who do not exert
themselves in order to perform well on tests like the
LEAS (which may lead to too-low scores on the
LEAS). People with low personal standards for their
performance in emotional communication situations,
for example, may be expected to show this combina-
tion of test results.
Study 3
As the TAS-20 involves the self-assessment of
abilities, and abilities generally are judged relative tosome kind of personal standard (e.g., Bandura, 1986),
the nature of these personal standards are likely to
influence the individual’s responses to the TAS-20.
That is, individuals with high personal standards for
what counts as being good at describing and commu-
nicating emotions are likely to receive too-high scores
on the TAS-20, whereas individuals with low stan-
dards in this regard are likely to receive too-low
scores. One approach to measuring personal standards
in general is to use instruments for measuring perfec-
tionism. In the early 1990s, two multidimensional in-
struments for measuring perfectionism were pub-
lished, one by Frost et al. (1990) and one by Hewittand Flett (1991), both of which have been widely used
in research on personality and psychopathology. No
previous research, however, has been reported on the
possible association between TAS-20 and any of these
measures of perfectionism.
The present study used Frost et al.’s (1990) Multi-
dimensional Perfectionism Scale (MPS), which mea-
sures six different aspects of perfectionism: Personal
Standards (PS), Concern Over Mistakes (CM),
Doubts About Action (DA), Organization (O), Paren-
tal Expectations (PE), and Parental Criticism (PC).
Because the Organization subscale does not correlate
with the other subscales, however, it is not included
when a summary score on perfectionism is computed.
The MPS makes no distinction between perfectionism
in different life domains, but rests on the assumption
that perfectionistic traits generalize across life do-
mains. If this is true, perfectionism may cause a per-
son to (a) set high personal standards for what it
means to be good at identifying and describing emo-
tions, (b) be more concerned about making mistakes
in identifying and describing emotions, and (c) doubt
his or her ability to identify and describe emotions.
That is, if perfectionism affects the self-assessment of
alexithymia, the TAS-20 Factors 1 and 2 should cor-
relate positively with the subscales PS, CM, and DA.In the present study, it was therefore expected that (a)
perfectionism (and especially the subscales PS, CM,
and DA) would correlate specifically with TAS-20
Factor 1 (difficulties identifying emotions and distin-
guishing them from bodily sensations) and Factor 2
(difficulties describing emotions to others), but not
with Factor 3 (an externally oriented style of think-
ing), which does not require the individual to rate his
or her difficulties or abilities; and (b) perfectionism
would predict TAS-20 scores even when controlling
for negative affect and somatic complaints.
Method
Participants
Seventy-four individuals (44 women, 29 men, and
1 person who did not report his or her gender) par-
ticipated in the study. The participants were recruited
randomly from a community sample of 1,000 men
and women aged 25–50 years old, living in central
Stockholm, that was drawn randomly from the SPAR
register. The mean age of the participants was 39.3
(SD 12.2) years.
Materials and Procedure
A questionnaire containing self-assessment instru-
ments of perfectionism, alexithymia, negative affect,
and somatic complaints was sent to the participants.
The participants also answered questions regarding
their age, sex, and whether they were undergoing
treatment for some medical illness. In case of affir-
mative answers to this question, they were asked to
LUNDH, JOHNSSON, SUNDQVIST, AND OLSSON370
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 11/19
name the illness. The present study used the same
version of the TAS-20 as in Study 1 and Study 2, and
in addition used the following instruments.
MPS. This instrument, which was developed by
Frost et al. (1990), contains 35 items in the form of
statements with a Likert-type 5-point response format
ranging from strongly disagree to strongly agree. The
Swedish version of the MPS shows good reliability
and validity (Saboonchi, 2000). Following Frost et al.
(1990) and Frost, Heimberg, Holt, Mattia, and Neu-
bauer (1993), the O subscale was not used in the
calculation of total MPS scores.
Pennebaker’s Inventory of Limbic Languidness
(PILL). This is a symptom checklist of 54 common
physical symptoms and bodily sensations, developed
by Pennebaker (1982). Respondents indicated the fre-
quency with which they had experienced each symp-
tom during the last year, and a total score was cal-
culated. The Swedish translation of the PILL had aCronbach’s alpha of .92 in a randomly selected
community sample (Lundh & Simonsson-Sarnecki,
2001).
Negative Affect Index (NAI). This is a 10-item
checklist of various aspects of negative affect (sad-
ness, tiredness, lack of interest, problems of concen-
tration, low self-esteem, guilt, worries, anger, stress,
loss of appetite), which is a short version of a negative
affect scale used by Lundh and Simonsson-Sarnecki
(2001). Respondents indicated on a 4-point scale the
frequency with which they had experienced each item
during the last year, and a total score was calculated.
The NAI had a Cronbach’s alpha of .93 in the presentsample.
Results and Discussion
Descriptive Data
The participants scored 41.7 (SD 13.2) on the
TAS-20, 66.8 (SD 22.0) on the MPS, and 48.0 (SD
30.9) on the PILL. Of the 74 participants, 20 re-
sponded that they were undergoing some kind of
medical treatment, whereas 54 responded that they
were not. Participants’ reported reasons for treatment
were hypertension (4 individuals), back pain (3),asthma (2), diabetes (2), gastrointestinal problems (2),
climacterical problems (2), psychiatric problems (1),
disturbance of balance (1), sinusitis (1), and Borrelia
(1). There were no significant differences on any of
the psychometric measures between those participants
who were undergoing medical treatment and those
who were not.
Associations Between the TAS-20 and Other Variables
The TAS-20 showed a sizable positive correlation
with the MPS (r .50, p < .0001), but also with the
PILL measure of somatic complaints (r .47, p <
.0001) and with the NAI (r .44, p < .0001). TheMPS also showed positive correlations of similar
magnitude with the PILL (r .42, p < .0003) and
with the NAI (r .57, p < .0001).
Consistent with Hypothesis 1, Factor 1 (Difficulties
Identifying Emotions) and Factor 2 (Difficulties De-
scribing Emotions) both correlated positively with the
MPS (r .57, p < .0001; and r .43, p < .0002,
respectively), whereas Factor 3 (Externally Oriented
Thinking) showed no significant correlation with the
MPS (r .18). This pattern was replicated for five
MPS subscales: CM (r s .47, .36, and .10), PS (r s
.41, .33, and .08), DA (r s .59, .40, and .23), PE
(r s .35, .28, and .20), and PC (r s .51, .33, and.19). The O subscale did not show any significant
correlations with the TAS-20 or any of its factors.
A multiple regression with TAS-20 as the depen-
dent variable, where the MPS, the PILL, and the NAI
were entered simultaneously into the equation, pro-
duced an R2 of 0.33 and, in accordance with Hypoth-
esis 2, showed that the MPS predicted TAS-20 scores
even when negative affect and somatic complaints
were controlled for (see Table 4).
To summarize, the present study showed a consid-
erable positive correlation between self-assessed
alexithymia and perfectionism, and specifically with
Factors 1 and 2. Moreover, perfectionism predicted
TAS-20 scores even when negative affect and somatic
complaints were controlled for. This is consistent with
the assumption that self-assessed alexithymia is influ-
enced by degree of perfectionism. However, it is not
equally consistent with the alternative explanations
that individuals with alexithymia develop perfection-
istic tendencies to compensate for their shortcomings
Table 4
Perfectionism, Somatic Complaints, and Negative Affect
as Predictors of TAS-20
Variable t p
MPS perfectionism .367 3.07 .003
PILL somatic complaints .303 2.14 .036
NAI negative affect .016 0.10 ns
Note. TAS-20 Toronto Alexithymia Scale (Bagby, Parker, &Taylor, 1994; Bagby, Taylor, & Parker, 1994); MPS Multidi-mensional Perfectionism Scale (Frost, Marten, Lahart, & Rosen-blate, 1990); PILL Pennebaker’s Inventory of Limbic Languid-ness (Pennebaker, 1982); NAI Negative Affect Index.
ALEXITHYMIA, META-EMOTION, AND PERFECTIONISM 371
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 12/19
(because then, perfectionism should also be expected
to correlate positively with TAS-20 Factor 3), or that
the two traits are associated because both are influ-
enced by degree of negative affect (because then, per-
fectionism should not be expected to predict TAS-20
scores when negative affect is controlled for).
Study 4
In Study 3, perfectionism was found to predict the
TAS-20 scores significantly when controlling for
negative affect and somatic complaints. The measure
of negative affect that was used, however, was newly
designed, and this was one reason for replicating this
study with other more well-known measures of nega-
tive affectivity. One purpose of Study 4 was to repli-
cate the previous study, with the inclusion of the BDI,
the BAI, and Spielberger, Jacobs, Russell, and
Crane’s (1983) State–Trait Anger Scale (STAS) totest the hypothesis that perfectionism predicts self-
assessed alexithymia when measures of negative af-
fect are controlled for.
In recent research on perfectionism, there are in-
creasing indications that there may be both positive
(healthy) and negative (unhealthy) forms of perfec-
tionism. Frost et al. (1993), for example, in a factor
analysis of Perfectionism subscales, found two fac-
tors, Maladaptive Evaluation Concerns and Positive
Striving, of which the first correlated with depression
and negative affect and the second correlated with
positive affect. These results are consistent with
Hamachek ’s (1978) distinction between normal andneurotic perfectionism. According to Hamachek, nor-
mal perfectionism means to set high standards for
oneself and to “derive a very real sense of pleasure
from the labors of a painstaking effort,” and yet “feel
free to be less precise as the situation permits” (p. 27).
Neurotic perfectionism, however, means to set high
standards and allow little latitude for making mis-
takes. Other researchers have made similar distinc-
tions in terms of positive versus negative perfection-
ism (Slade & Owens, 1998; Terry-Short, Owens,
Slade, & Dewey, 1995), active versus passive perfec-
tionism (Adkins & Parker, 1996; Lynd-Stevenson &
Hearne, 1999), and adaptive versus maladaptive per-fectionism (Rice, Ashby, & Slaney, 1998). As Slade
and Owens (1998) put it, positive perfectionism is the
result of positive reinforcement; it therefore reflects
the pursuit of success and excellence and is likely to
be associated with positive emotional consequences
such as pleasure, satisfaction, and euphoria. Negative
perfectionism, on the other hand, is assumed to be the
result of negative reinforcement; it therefore reflects
the pursuit of perfection as the only way of avoiding
failure (i.e., being less than perfect is treated as
equivalent with failure) and is associated with nega-
tive affect.
Like Frost et al. (1990), Slade and Owens (1998)
made no distinction between perfectionism in differ-
ent life domains, but started from the assumption that
perfectionistic traits generalize across life domains. If
this is true, then (a) positive perfectionism may cause
a person to strive for success and excellence, that is,
perfection, in his or her ability to identify and describe
emotions, which is not by itself likely to be associated
with higher scores on self-assessed alexithymia; and
(b) negative perfectionism may cause him or her to
perceive anything less than perfection in the identifi-
cation and description of emotions as a failure. Be-
cause the latter will lead to a more harsh judgment of
one’s own ability to identify and describe emotions,negative perfectionism may be expected to correlate
positively with self-assessed alexithymia. The present
hypotheses, therefore, were that (a) self-assessed
alexithymia is associated with negative rather than
positive perfectionism, (b) negative perfectionism
predicts self-assessed alexithymia even when control-
ling for measures of negative and positive affect, and
(c) negative perfectionism is associated specifically
with TAS-20 Factors 1 and 2.
Method
Participants
Seventy-seven undergraduate students (62 women,
15 men) participated in the study. Their mean age was
27.8 (SD 7.8) years, and they received partial
course credit for their participation. None of these
participants had taken part in Studies 1 or 2.
Materials and Procedure
The present study used the same versions of the
TAS-20, the BDI, and the BAI as used in the previous
studies. In addition, the following instruments were
used.
Positive and Negative Perfectionism Scale (PANPS).This is a 40-item questionnaire (Terry-Short et al.,
1995), with two 20-item subscales: one for Positive
Perfectionism (PANPS–P) and one for Negative Per-
fectionism (PANPS–N). The Swedish translations
of both subscales showed good internal consis-
tency (Cronbach’s .79 for PANPS–P and .82 for
PANPS–N).
LUNDH, JOHNSSON, SUNDQVIST, AND OLSSON372
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 13/19
STAS – T. This instrument, which was developed
by Spielberger et al. (1983), contains 10 items regard-
ing the habitual experience of anger. The items are in
the form of statements with a Likert-type 4-point re-
sponse format ranging from almost never to almost
always. Saboonchi and Lundh (in press) found that
the Swedish translation of the STAS had good internal
consistency (Cronbach’s .84) in a sample of 186
randomly selected adult individuals from Stockholm.
Positive Affect Index (PAI; Lundh & Simonsson-
Sarnecki, 2001). This instrument contains 18 items
that ask the respondents to rate on a 4-point scale how
much they usually experience various kinds of posi-
tive emotional experiences (joy, safe–calm, satisfied
with self, energy–lust for work, feeling creative,
success, feelings of togetherness, nice company with
others, laughter–happy moments, enjoying music–singing–dancing, enjoying good TV programs–films–
theatre, enjoying reading–writing, enjoying goodfood, enjoying the weather–nature, good conversa-
tions, loving feelings, sexual pleasure, relaxation).
The PAI was originally developed within the Health
Quality of Life project, which involves a sample of
202 randomly selected individuals from Stockholm
(Lundh & Simonsson-Sarnecki, 2001), where it was
found to show good reliability (Cronbach’s .86)
and validity (r .56, with a diary measure of positive
affect).
Results and Discussion
The participants scored 40.6 (SD 11.2) on the
TAS-20, 5.4 (SD 5.5) on the BDI, 8.0 (SD 4.6)
on the BAI, 17.3 (SD 4.2) on the STAS–T, and37.7 (SD 8.8) on the PAI. They scored 67.2 (SD
8.2) on Positive Perfectionism (PANPS–P), and 50.8
(SD 9.5) on Negative Perfectionism (PANPS–N).
As can be seen in Table 5 the TAS-20 showed a
sizable positive correlation with Negative Perfection-
ism (r .45, p < .0001), but did not correlate sig-
nificantly with Positive Perfectionism. The TAS also
correlated positively with depression and anxiety, and
negatively with positive affect, but showed no signifi-
cant correlation with anger ( p > .15). Negative Per-
fectionism showed a very similar pattern of correla-
tions with depression, anxiety, and positive affect, but
in addition also showed positive correlations with
both Positive Perfectionism and trait anger. Positive
Perfectionism correlated positively not only with
Negative Perfectionism but also with positive affect
and anger, but showed no significant correlations with
anxiety or depression.
To test Hypothesis 2, we computed a multiple re-
gression with TAS-20 as the dependent variable, in
which the BDI, the BAI, the PAI, and the PANPS-N
were entered simultaneously into the equation. Thisproduced an R2 of 0.34 and showed the PANPS-N
significantly predicted the TAS-20 scores even when
negative and positive affect were controlled for (see
Table 6).
In support of Hypothesis 3, Negative Perfectionism
correlated positively with both Factor 1 (Difficulties
Identifying Emotions; r .55, p < .0001) and Factor
2 (Difficulties Describing Emotions; r .44, p
.001), but not with Factor 3 (Externally Oriented
Thinking; (r .03, ns). Factor 1 also correlated posi-
tively with depression (r .50, p < .0001), anxiety (r
.49, p < .0001), and anger (r .29, p < .01) and
negatively with positive affect (r −.44, p < .0001).Factor 2 showed a similar pattern of correlations, al-
though somewhat weaker: depression (r .37, p <
.001), anxiety (r .28, p < .05), anger (r .22, p <
Table 5
Zero-Order Correlations Between the TAS-20, Positive Perfectionism Negative Perfectionism, Depression, Anxiety, Anger,
and Positive Affect
Instrument 1 2 3 4 5 6 7
1. TAS-20 —2. PANPS–P .02 —3. PANPS–N .45*** .32* —4. BDI .46*** −.02 .44*** —5. BAI .43** .09 .38** .49*** —6. STAS .16 .33* .30* .15 .19 —7. PAI −.42** .35* −.33* −.63*** −.31* −.02 —
Note. TAS-20 Toronto Alexithymia Scale (Bagby, Parker, & Taylor, 1994; Bagby, Taylor, & Parker, 1994); PANPS –P–N Positiveand Negative Perfectionism Scale—Positive Perfectionism—Negative Perfectionism (Terry-Short, Owens, Slade, & Dewey, 1995); BDI
Beck Depression Inventory (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961); BAI Beck Anxiety Inventory (Beck, Brown, Epstein, &Steer, 1988); STAS State–Trait Anger Scale (Spielberger, Jacobs, Russell, & Crane, 1983); PAI Positive Affect Index (Lundh &Simonsson-Sarnecki, 2001).* p < .01. ** p < .001. *** p < .0001.
ALEXITHYMIA, META-EMOTION, AND PERFECTIONISM 373
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 14/19
.10), and positive affect (r −.31, p < .01). As in
other studies, Factor 3 showed a very different pat-
tern, showing only a weak negative association with
trait anger (r −.23, p < .05). None of the TAS
factors correlated significantly with Positive Perfec-tionism.
To summarize, Study 4 corroborates the results
from Study 3 both by showing a sizeable positive
correlation between Negative Perfectionism and the
TAS-20 (specifically Factors 1 and 2), and by show-
ing that Negative Perfectionism predicted TAS-20
scores when measures of negative affect (BDI and the
BAI) were controlled for. Positive Perfectionism,
however, although showing moderate correlations
with Negative Perfectionism, trait anger, and positive
affect, showed no significant association neither with
self-assessed alexithymia nor with measures of anxi-
ety or depression.
General Discussion
To summarize, there are two aspects of the present
results that need to be discussed. First, the present
studies showed no significant correlations in the ex-
pected direction between a self-assessment measure
of alexithymia, the TAS-20, and performance mea-
sures of emotional awareness (LEAS) and memory of
emotions (AMT). Second, the present results showed
substantial positive correlations between self-assessed
alexithymia (TAS-20) and two different measures of
perfectionism (MPS and PANPS).
TAS-20 and Performance Measures of Meta-Emotional Competence
Studies 1 and 2 showed no significant correlations
between the TAS-20 and performance measures of
memory of emotion and emotional awareness. There
are a number of possible explanations for this lack of
effects. First, the TAS-20 primarily measures people’s
beliefs about their meta-emotional functioning, rather
than meta-emotional functioning as such, and it is
possible that these beliefs differ from people’s actual
performance in such a way that no correlation with
performance measures are to be expected. Against
this, however, speaks a number of previous studies
where significant effects in the predicted direction
have been obtained between the TAS-20 and perfor-
mance measures of emotion perception (Jessimer &
Markham, 1997; Lumley et al., 2002; Mann et al.,
1994; Parker et al., 1993; Prkachin & Prkachin, 2001).
Second, it is possible that there are problems with
the AMT and the LEAS, which explain the absence of
effects. With regard to the LEAS, however, research-
ers may argue that previous research has shown that it
correlates both with measures of emotion perception
(Lane et al., 1996) and other cognitive-emotional de-
velopmental measures (Feldman Barrett, 1999; Lane
et al., 1990). With regard to AMT, there is more un-certainty because it has not been used in this research
context before. For example, even if alexithymia
would by definition involve difficulties to remember
emotions, researchers may argue that the AMT mea-
sures only the latency for each person to retrieve those
emotional events that are actually remembered. Or in
other words, to the extent that alexithymia is associ-
ated with less encoding of events in emotional terms,
this is likely to lead to a smaller number of events that
are encoded in emotional terms, and consequently to
a smaller number of events that can be retrieved on
the basis of emotional cue words—but must it also
lead to longer latencies to retrieve these emotionalevents? Further, it is possible that both the LEAS and
the AMT are sensitive to temporary situational con-
texts or motivational states in such a way that they do
not serve as stable measures of meta-emotional com-
petence. What is required here is more research on the
validity and reliability (e.g., the test–retest reliability)
of the LEAS and the AMT.
Third, there is a possibility that alexithymia, as as-
sessed by the TAS-20, is associated with difficulties
to remember emotions, and low scores on the LEAS,
only at very high or clinical levels of alexithymia. It
should be noted that the mean scores on the TAS-20
were rather low in the samples that were studied, andvery few of the participants scored high on the TAS-
20. Although researchers have generally argued (e.g.,
Taylor et al., 1997) that alexithymia is a dimensional
construct, which would imply that the predicted asso-
ciations between TAS-20 and other measures may be
expected to also show up in normal nonclinical
samples, it is also important to remember that the
Table 6
Negative Perfectionism, Anxiety, Depression, and Positive
Affect as Predictors of TAS-20
Variable t p
PANPS–N .258 2.356 .021
BAI .211 1.877 .065BDI .120 0.864 ns
PAI −.189 −1.525 ns
Note. PANPS–N Positive and Negative Perfectionism Scale—Negative Perfectionism (Terry-Short, Owens, Slade, & Dewey,1995); BAI Beck Anxiety Inventory (Beck, Brown, Epstein, &Steer, 1988); BDI Beck Depression Inventory (Beck, Ward,Mendelson, Mock, & Erbaugh, 1961); PAI Positive Affect Index(Lundh & Simonsson-Sarnecki, 2001).
LUNDH, JOHNSSON, SUNDQVIST, AND OLSSON374
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 15/19
TAS-20 was developed to identify individuals with
obvious features of alexithymia and not to discrimi-
nate among individuals at the lower end of this con-
struct.
Fourth, it is possible that the ability to remember
emotions (as measured by the AMT) and the ability to
imagine hypothetical emotions (as measured by the
LEAS) represent other kinds of meta-emotional com-
petence than the abilities–difficulties to identify and
describe emotions that are measured by the TAS-20.
If these are separate meta-emotional capacities, then
we should not expect any correlation between the
TAS-20 and the AMT or the LEAS, even if all these
instruments were highly reliable. Further, even if
these meta-emotional capacities are not entirely sepa-
rate, the differences between self-assessment (TAS-
20) and performance (AMT and LEAS) modes of
measurement may make it difficult to detect existing
associations between these capacities.The results from previous research that has used
both the TAS-20 and various kinds of performance
measures is at least not entirely inconsistent with this
possibility. When photographs of human faces ex-
pressing different emotions have been used as stimuli
and the participants have been required to name the
emotion that is expressed, most earlier studies report
that people with high TAS scores are less able than
people with low TAS scores to accurately label the
emotion that is expressed (Jessimer & Markham,
1997; Mann et al., 1994; Parker et al., 1993; Prkachin
& Prkachin, 2001), although there are also some stud-
ies that have failed to find such effects (Berenbaum &Prince, 1994; Mayer et al., 1990). However, when
verbal vignettes or stories depicting emotional events
have been used as stimuli, and the participants have
been required to reason about these, there is less sup-
port for an affect-recognition deficit in participants
with high TAS scores. Berenbaum and associates
(Berenbaum et al., 1998; Berenbaum & Prince, 1994)
even reported a tendency for high-TAS scorers to be
more accurate in their interpretation of emotional situ-
ations than low-TAS scorers. In the present studies,
both of the performance measures that failed to cor-
relate with TAS-20 made use of verbal stimuli; the
AMT used verbal labels of emotions as stimuli andrequired the participant to retrieve memories where
these emotions were experienced, whereas the LEAS
used verbal vignettes describing emotional situations
and required the participant to imagine the emotions
that would be felt in these situations.
One possibility, therefore, is that the TAS-20 mea-
sures people’s beliefs about their ability to perceive
emotion when it is experienced (by self or others),
rather than their ability to imagine, remember, or rea-
son about emotion, and that people’s beliefs about
their ability to perceive emotion are most often suffi-
ciently accurate to produce significant effects in the
predicted direction in studies of the association be-
tween the TAS-20 and emotion perception. The abil-
ity to imagine emotions in hypothetical situations (as
measured by the LEAS), and the ability to retrieve
concrete instances of emotions on the basis of their
verbal labels (as measured by the AMT), however,
may represent meta-emotional abilities that are rela-
tively independent from emotion perception, or at
least so distinct that the TAS-20 (or other self-
assessment measures of emotion perception) would
fail to correlate with them.
TAS-20 and Perfectionism
With regard to the other main finding, the presentstudies are the first to report substantial positive cor-
relations between the self-assessment of alexithymia
and measures of perfectionism. The fact that these
associations were found with two very different mea-
sures of perfectionism (MPS and PANPS) testify to
the robustness of these findings. There are, however,
some previous studies (Laquatra & Clopton, 1994;
Taylor, Parker, Bagby, & Bourke, 1996) that have
used the TAS-20 together with the Eating Disorders
Inventory (EDI; Garner, Olmstead, & Polivy, 1983),
which contains a small subscale for Perfectionism,
and that found no significant correlation between the
TAS-20 and EDI Perfectionism, neither in anorexicwomen nor in normal controls. The EDI Perfection-
ism subscale has shown moderate correlations with
the MPS (Frost et al., 1990), and the reason for these
discrepant results is unclear.
Although the present findings on self-assessed
alexithymia and perfectionism are correlational and
therefore open to different causal explanations, we
think that the correlational patterns that were found
point in some directions rather than in others. If de-
gree of perfectionism (e.g., personal standards) affects
how a person will respond to evaluational questions
concerning his or her abilities–difficulties (as assessed
by TAS-20 Factors 1 and 2), but not with regard tomore descriptive questions about his or her style of
thinking (as assessed by TAS-20 Factor 3), then per-
fectionism should be expected to correlate positively
with TAS-20 Factors 1 and 2 but not with TAS-20
Factor 3. However, if perfectionism is an effect of
alexithymia, or if both are the effects of some other
factors, and if all three TAS-20 factors are essential to
ALEXITHYMIA, META-EMOTION, AND PERFECTIONISM 375
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 16/19
the alexithymia construct, then all three factors of the
TAS-20 may be expected to correlate significantly
with perfectionism. The correlational patterns that
were found (i.e., perfectionism correlating signifi-
cantly with Factors 1 and 2 but not with Factor 3)
clearly seem to favor the first explanation rather than
the second. It is also interesting to note the magnitude
of the correlations that were found between Factor 1
(Difficulties Identifying Feelings) and perfectionism
in both studies (r .57 and r .55).
Nevertheless, the present results are not sufficient
to demonstrate that perfectionistic standards have a
causal effect on the self-assessment of difficulties to
identify and describe emotions. One aspect of this
issue concerns the generality of personal standards, or
perfectionism across various life domains. It is not
self-evident that perfectionism, as measured by the
MPS and the PANPS, generalizes to all areas of self-
evaluation, including the evaluation of one’s meta-emotional capacities. Ideally, we should have a mea-
sure of the person’s degree of perfectionism in the
area of meta-emotional functioning, particularly with
regard to the identification and communication of
emotions. That is, if Factors 1 and 2 of the TAS-20
would correlate positively with a measure of the de-
mands that a person has on-him or herself with regard
to being able to describe and communicate emotions,
then the present hypothesis would be considerably
strengthened. Further, the best possible test of the
hypothesis that perfectionistic standards influence the
TAS-20 scores would probably be an experimental
study where the strictness of the participants’ personal
standards with regard to emotion identification and
emotion communication were manipulated directly in
order to determine whether this manipulation would
have an influence on the TAS-20 scores. Although it
may be difficult to see how such a manipulation is
possible, there is some evidence that negative moods
tend to induce higher standards for performance (Cer-
vone, Kopp, Schaumann, & Scott, 1994). If this ap-
plies also to standards for the communication of emo-
tion, this may be one reason why depression is
associated with higher TAS-20 scores (i.e., the nega-
tive mood induces higher standards in this regard,which makes the individual assess his or her ability to
identify and describe emotions as less good, with re-
sulting higher scores on self-assessed alexithymia).
However, the present finding that perfectionism pre-
dicts TAS-20 scores even when negative affect is con-
trolled speaks against the possibility that this would
be a sufficient explanation.
Conclusion
It is possible that meta-emotional processing should
be conceived of as a broad spectrum of different pro-
cesses, of which some are intercorrelated and others
are not. For example, if TAS-20 Factors 1 and 2 mea-
sure something like the lack of meta-emotional self-efficacy concerning the identification and communi-
cation of emotion, this does not by itself render the
TAS-20 any less interesting. As with other kinds of
self-efficacy, meta-emotional self-efficacy may well
be an important aspect of individual functioning, even
if it is influenced by current emotional state, personal
standards, and information structures in semantic
memory, and even if it does not correlate with other
aspects of meta-emotional processing. The study of
meta-emotional processing is perhaps most likely to
benefit from a multidimensional approach combined
with a methodological pluralism that (a) attempts to
identify different dimensions of meta-emotional pro-
cessing and (b) does this with a variety of different
methods, such as self-assessment scales, performance
tests, observer instruments, structured interviews, etc.
An important limitation of the present studies is
that they all used nonclinical samples of volunteers
(undergraduate students and community samples).
One possibility is that the TAS-20 is associated with
difficulties to retrieve emotional memories at clinical
levels, although no association is found at nonclinical
levels. Further, it remains to be seen whether perfec-
tionism also predicts TAS-20 scores in clinical
samples with high alexithymia scores. Further re-
search is therefore needed to ascertain whether the
present results also generalize to TAS-20 alexithymia
in clinical samples with higher levels of alexithymia.
References
Adkins, K. K., & Parker, W. (1996). Perfectionism and sui-
cidal preoccupation. Journal of Personality, 64, 529–543.
Bagby, R. M., Parker, J. D. A., & Taylor, G. J. (1994). The
twenty-item Toronto Alexithymia Scale: I. Item selection
and cross-validation of the factor structure. Journal of
Psychosomatic Research, 38, 23–32.
Bagby, R. M., Taylor, G. J., & Parker, J. D. A. (1994). Thetwenty-item Toronto Alexithymia Scale: II. Convergent,
discriminant, and concurrent validity. Journal of Psycho-
somatic Research, 38, 33 – 40.
Bandura, A. (1986). Social foundations of thought and ac-
tion: A social cognitive theory. Englewood Cliffs, NJ:
Prentice Hall.
Beck, A. T., Brown, G., Epstein, N., & Steer, R. A. (1988).
LUNDH, JOHNSSON, SUNDQVIST, AND OLSSON376
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 17/19
An inventory for measuring clinical anxiety. Journal of
Consulting and Clinical Psychology, 56, 893–897.
Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psycho-
metric properties of the Beck Depression Inventory:
Twenty-five years of evaluation. Clinical Psychology Re-
view, 8, 77–100.
Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Er-
baugh, J. (1961). An inventory for measuring depression.
Archives of General Psychiatry, 4, 561–571.
Berenbaum, H., Davis, R., & McGrew, J. (1998). Alexi-
thymia and the interpretation of hostile-provoking situa-
tions. Psychotherapy and Psychosomatics, 67, 254 –258.
Berenbaum, H., & Prince, J. D. (1994). Alexithymia and the
interpretation of emotion-relevant information. Cognition
and Emotion, 8, 231–244.
Blaney, P. H. (1986). Affect and memory: A review. Psy-
chological Bulletin, 99, 229–246.
Cervone, D., Kopp, D. A., Schaumann, L., & Scott, W. D.
(1994). Mood, self-efficacy, and performance standards:Lower moods induce higher standards for performance.
Journal of Personality and Social Psychology, 67, 499–512.
Crowne, D. P., & Marlowe, D. (1960). A new scale of social
desirability independent of psychopathology. Journal of
Consulting and Clinical Psychology, 24, 349–354.
Davies, M., Stankov, L., & Roberts, R. (1998). Emotional
intelligence: In search of an elusive construct. Journal of
Personality and Social Psychology, 75, 989–1015.
Devine, H., Stewart, S. H., & Watt, M. C. (1999). Relations
between anxiety sensitivity and dimensions of alexi-
thymia in a young adult sample. Journal of Psychoso-
matic Research, 47, 145–158.Ellis, H. C., & Moore, B. A. (1999). Mood and memory. In
T. Dalgleish & M. Power (Eds.), Handbook of cognition
and emotion. New York: Wiley.
Feldman Barrett, L. (1999). [Personality correlates of the
LEAS]. Unpublished raw data.
Freyberger, H. (1977). Supportive psychotherapy tech-
niques in primary and secondary alexithymia. Psycho-
therapy and Psychosomatics, 28, 337–342.
Frost, R. O., Heimberg, R. G., Holt, C. S., Mattia, J. I., &
Neubauer, L. A. (1993). A comparison of two measures
of perfectionism. Personality and Individual Differences,
14, 119–126.
Frost, R. O., Marten, P., Lahart, C., & Rosenblate, R.
(1990). The dimensions of perfectionism. Cognitive
Therapy and Research, 14, 449– 468.
Garner, D. M., Olmstead, M. P., & Polivy, J. (1983). De-
velopment and validation of a multidimensional eating
disorder for anorexia nervosa and bulimia. International
Journal of Eating Disorders, 2, 15–34.
Gustavsson, P., Weinryb, R. M., Goransson, S., Pedersen,
N. L., & Åsberg, M. (1997). Stability and predictive abil-
ity of personality traits across 9 years. Personality and
Individual Differences, 22, 784 –791.
Hamachek, D. E. (1978). Psychodynamics of normal and
neurotic perfectionism. Psychology, 15, 27–33.
Haviland, M. G., Warren, W. L., & Riggs, M. L. (2000). An
observer scale to measure alexithymia. Psychosomatics,
41, 385–392.
Hewitt, P. L., & Flett, G. L. (1991). Perfectionism in the self
and social contexts: Conceptualization, assessment, and
association with psychopathology. Journal of Personality
and Social Psychology, 60, 456 – 470.
Honkalampi, K., Hintikka, J., Saarinen, P., Lehtonen, J., &
Viinamaki, H. (2000). Is alexithymia a permanent feature
in depressed patients? Psychotherapy and Psychosomat-
ics, 69, 303–308.
Honkalampi, K., Koivumaa-Honkanen, H., Tanskanen, A.,
Hintikka, J., Lehtonen, J., & Viinamaki, H. (2001). Why
do alexithymic features appear to be stable? A 12-monthfollow-up of a general population. Psychotherapy and
Psychosomatics, 70, 247–253.
Jessimer, M., & Markham, R. (1997). Alexithymia: A right
hemisphere dysfunction specific to recognition of certain
facial expressions? Brain and Cognition, 34, 246–258.
Klein, S. B., Babey, S. H., & Sherman, J. W. (1997). The
functional independence of trait and behavioral self-
knowledge: Methodological considerations and new em-
pirical findings. Social Cognition, 15, 183–203.
Klein, S. B., & Loftus, J. (1993). The mental representation
of trait and autobiographical knowledge about the self. In
T. K. Srull & R. S. Wyer, Jr. (Eds.), The mental repre-
sentation of trait and autobiographical knowledge about the self. Advances in social cognition (Vol. 5, pp. 1– 49).
Hillsdale, NJ: Erlbaum.
Klein, S. B., Sherman, J. W., & Loftus, J. (1996). The role
of episodic and semantic memory in the development of
trait self-knowledge. Social Cognition, 14, 277–291.
Lane, R. D. (1991). LEAS scoring manual and glossary.
Tucson, AZ: Department of Psychiatry, University of
Arizona Health Sciences Center.
Lane, R. D., Quinlan, D. M., Schwartz, G. E., Walker,
P. A., & Zeitlin, S. B. (1990). The Levels of Emotional
Awareness Scale: A cognitive-developmental measure of
emotion. Journal of Personality Assessment, 55, 124 –
134.
Lane, R. D., Sechrest, L., & Riedel, R. (1998). Sociodemo-
graphic correlates of alexithymia. Comprehensive Psy-
chiatry, 39, 377–385.
Lane, R. D., Sechrest, L., Riedel, R., Weldon, V., Kaszniak,
A. W., & Schwartz, G. E. (1996). Impaired verbal and
nonverbal emotion recognition in alexithymia. Psychoso-
matic Medicine, 58, 203–210.
ALEXITHYMIA, META-EMOTION, AND PERFECTIONISM 377
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 18/19
Laquatra, T. A., & Clopton, J. R. (1994). Characteristics of
alexithymia and eating disorders in college women. Ad-
dictive Behaviors, 19, 373–380.
Lloyd, G. G., & Lishman, W. A. (1975). Effects of depres-
sion on speed of recall of pleasant and unpleasant expe-
riences. Psychological Medicine, 5, 173–180.
Luminet, O., Bagby, R. M., & Taylor, G. J. (2001). An
evaluation of the absolute and relative stability of alexi-
thymia in patients with major depression. Psychotherapy
and Psychosomatics, 70, 254 –260.
Lumley, M. A., Davis, M., Labouvie-Vief, G., Gustavson,
B., Clement, R., Barry, R., & Simon, T. (2002). Multiple
measures of emotional abilities: Their interrelationships
and associations with physical symptoms. Psychosomatic
Medicine, 64, 146.
Lumley, M. A., Ovies, T., Stettner, L., Wehmer, F., &
Lakey, B. (1996). Alexithymia, social support and health
problems. Journal of Psychosomatic Research, 41, 519–
530.
Lundh, L. G., & Simonsson-Sarnecki, M. (2001). Alexi-
thymia, emotion, and somatic complaints. Journal of Per-
sonality, 69, 483–510.
Lynd-Stevenson, R. M., & Hearne, C. M. (1999). Perfec-
tionism and depressive affect: The pros and cons of being
a perfectionist. Personality and Individual Differences,
26, 549–562.
Mann, L. S., Wise, T. N., Trinidad, A., & Kohanski, R.
(1994). Alexithymia, affect recognition, and the five-
factor model of personality in normal subjects. Psycho-
logical Reports, 74, 563–567.
Mayer, J. D., Caruso, D., & Salovey, P. (1999). Emotionalintelligence meets traditional standards for an intelli-
gence. Intelligence, 27, 267–298.
Mayer, J. D., Caruso, D., & Salovey, P. (2000). Selecting a
measure of emotional intelligence. The case for ability
scales. In R. Bar-On & J. D. A. Parker (Eds.), The hand-
book of emotional intelligence (pp. 320–342) San Fran-
sisco: Jossey-Bass.
Mayer, J. D., DiPaolo, M., & Salovey, P. (1990). Perceiving
affective content in ambiguous visual stimuli: A compo-
nent of emotional intelligence. Journal of Personality As-
sessment, 54, 772–781.
Mortazavi, M. S. (2001). Alexithymia. A psychologicalanalysis based on clinical and healthy individuals.
Uppsala, Sweden: Almqvist & Wicksell International.
Nemiah, J. C., Freyberger, H., & Sifneos, P. E. (1976).
Alexithymia: A view of the psychosomatic process. In
O.W. Hill (Ed.), Modern trends in psychosomatic medi-
cine (Vol. 3, pp. 430 – 439). London: Butterworths.
Parker, J. D., Bagby, R. M., & Taylor, G. J. (1991). Alexi-
thymia and depression: Distinct or overlapping con-
structs? Comprehensive Psychiatry, 32, 387–394.
Parker, J. D., Taylor, G. J., & Bagby, R. M. (1993). Alexi-
thymia and the recognition of facial expressions of emo-
tion. Psychotherapy and Psychosomatics, 59, 197–202.
Paulhus, D. L., Lysy, D. C., & Yik, M. S. M. (1998). Self-
report measures of intelligence: Are they useful as proxy
IQ tests? Journal of Personality, 66, 525–554.
Pennebaker, J. W. (1982). The psychology of physiological
symptoms. New York: Springer-Verlag.
Prkachin, G. C., & Prkachin, K. M. (2001). Alexithymia
and detection of facial expressions of emotion [Abstract]
Psychosomatic Medicine, 63, 135–136.
Rau, J. C. (1993). Perception of verbal and nonverbal af-
fective stimuli in complex partial seizure disorder. Dis-
sertation Abstracts International B, 54, 506B.
Reiss, S., Peterson, R. A., Gursky, D., & McNally, R. J.
(1986). Anxiety sensitivity, anxiety frequency, and the
prediction of fearfulness. Behaviour Research and Therapy, 24, 1–8.
Rice, K. G., Ashby, J. S., & Slaney, R. B. (1998). Self-
esteem as a mediator between perfectionism and depres-
sion: A structural equations analysis. Journal of Coun-
seling Psychology, 45, 304 –314.
Saboonchi, F. (2000). Perfectionism. Conceptual, emo-
tional, psychopathological, and health-related implica-
tions. Stockholm: Department of Psychology, Stockholm
University.
Saboonchi, F., & Lundh, L. G. (in press). Perfectionism,
anger, somatic health, and positive affect. Personality
and Individual Differences.
Salminen, J., Saarijarvi, S., Aarela, E., Toikka, T., & Kau-hanen, J. (1999). Prevalence of alexithymia and its asso-
ciation with sociodemographic variables in the general
population of Finland. Journal of Psychosomatic Re-
search, 46, 75–82.
Sifneos, P. E. (1973). The prevalence of “alexithymic” char-
acteristics in psychosomatic patients. Psychotherapy and
Psychosomatics, 22, 255–262.
Simonsson-Sarnecki, M., Lundh, L. G., Torestad, B.,
Bagby, R. M., Taylor, G. J., & Parker, J. D. A. (2000). A
Swedish translation of the 20-item Toronto Alexithymia
Scale: Cross-validation of the factor structure. Scandina-
vian Journal of Psychology, 41, 25–30.
Slade, P. D., & Owens, B. G. (1998). A dual process model
of perfectionism based on reinforcement theory. Behavior
Modification, 22, 372–390.
Spielberger, C., Jacobs, G., Russell, S., & Crane, R. S.
(1983). Assessment of anger: The State-Trait Anger
Scale. In J. N. Butcher & C. D. Spielberger (Eds.), Ad-
vances in personality assessment (Vol. 2, pp. 16–190).
Hillsdale, NJ: Erlbaum.
LUNDH, JOHNSSON, SUNDQVIST, AND OLSSON378
7/21/2019 Lundh, Johnsson Et Al (2002)
http://slidepdf.com/reader/full/lundh-johnsson-et-al-2002 19/19
Suslow, T., Donges, U. S., Kersting, A., & Arolt, V. (2000).
20-item Toronto Alexithymia Scale: Do difficulties de-
scribing feelings assess proneness to shame instead of
difficulties symbolizing emotions? Scandinavian Journal
of Psychology, 41, 329–334.
Taylor, G. J., & Bagby, R. M. (2000). An overview of the
alexithymia construct. In R. Bar-On & J. D. A. Parker
(Eds.), The handbook of emotional intelligence. San Fran-
cisco: Jossey-Bass.
Taylor, G. J., Bagby, R. R., & Parker, J. D. A. (1997). Dis-
orders of affect regulation: Alexithymia in medical and
psychiatric illness. Cambridge, England: Cambridge Uni-
versity Press.
Taylor, G. J., Parker, J. D. A., Bagby, R. M., & Bourke,
M. P. (1996). Relationships between alexithymia and
psychological characteristics associated with eating dis-
orders. Journal of Psychosomatic Research, 41, 561–568.
Taylor, J. A. (1953). A personality scale of manifest anxi-
ety. Journal of Abnormal and Social Psychology, 48,
285–290.
Teasdale, J. D., & Fogarty, S. J. (1979). Differential effects
of induced mood on retrieval of pleasant and unpleasant
events from episodic memory. Journal of Abnormal Psy-
chology, 88, 248–257.
Terry-Short, L. A., Owens, R. G., Slade, P. D., & Dewey,
M. E. (1995). Positive and negative perfectionism. Per-
sonality and Individual Differences, 18, 663– 668.
Williams, J. M. G., & Broadbent, K. (1986). Autobiographi-
cal memory in attempted suicide patients. Journal of Ab-
normal Psychology, 95, 144 –149.
Received August 23, 2001
Revision received June 26, 2002
Accepted June 29, 2002
ALEXITHYMIA, META-EMOTION, AND PERFECTIONISM 379