Content uploaded by Anne-Sophie Deborde
Author content
All content in this area was uploaded by Anne-Sophie Deborde on Oct 11, 2017
Content may be subject to copyright.
Content uploaded by Mario Speranza
Author content
All content in this area was uploaded by Mario Speranza
Content may be subject to copyright.
Content uploaded by Mario Speranza
Author content
All content in this area was uploaded by Mario Speranza
Content may be subject to copyright.
676
Journal of Personality Disorders, 26(5), 676–688, 2012
© 2012 The Guilford Press
ALEXITHYMIA AS A MEDIATOR BETWEEN
ATTACHMENT AND THE DEVELOPMENT
OF BORDERLINE PERSONALITY DISORDER
IN ADOLESCENCE
Anne-Sophie Deborde, PhD, Raphaële Miljkovitch, PhD,
Caroline Roy, PhD, Corinne Dugré-Le Bigre, PhD,
Alexandra Pham-Scottez, MD, Mario Speranza, MD, PhD,
and Maurice Corcos, MD, PhD
Insecure attachment and the inability to identify emotions have both
been put forward as possible explanations for dysfunction of the emo-
tional system in borderline personality disorder (BPD). This study
aimed to test a model according to which the influence of attachment
on the development of BPD in adolescence is mediated by alexithymia.
Borderline severity was assessed by means of the Structured Interview
for DSM-IV Personality Disorders. Attachment and alexithymia were
measured respectively with the Relationship Styles Questionnaire and
the Toronto Alexithymia Scale. Mediation analyses conducted on 105
participants (54 with BPD and 51 matched controls) suggest that the
role of security and negative model of self (i.e., preoccupied and fearful
attachment styles) in the development of BPD symptoms are mediated
by alexithymia.
Borderline personality disorder (BPD) is characterized by intense and
labile emotions, significant conflict in interpersonal relationships, and
extreme behavioral impulsivity. These features often break out during
adolescence (Roberts, Attkisson, & Rosenblatt, 1998). Authors such as
Linehan, Heard, and Amstrong (1993), Corrigan, Davidson, and Heard
This article was accepted under the editorship of Paul S. Links.
From CRAC-Laboratoire Paragraphe EA 349, Paris 8 University (A.-S. D., R. M.); Psychiatry
Department, Cochin Hospital, Paris (A.-S. D., R. M.); Université Laval, Québec, Canada
(C. R.); Department of Psychiatry for Adolescents and Young Adults, Institut Mutualiste
Montsouris, Paris (A.-S. D., C. D., M. C.); University René Descartes-Paris V, Faculty of
Medecine, psychology laboratory (C. D., M. S., M. C.); Inserm U 669 (C. D., A. P., M. S.,
M. C.); Versailles Hospital, Child Psychiatry Department (M. S.); Paris XI University (M. S.);
and CMME, Sainte-Anne Hospital, Paris (A. P.).
Research supported by WYETH Foundation for Child and Adolescent Health and the Lilly
Foundation.
Address correspondence to Raphaële Miljkovitch, Université Paris 8, UFR Psychologie
Pratiques Cliniques et Sociales, 2 Rue de la Liberté, 93200 Saint-Denis, France; E-mail:
raphaele.miljkovitch@iedparis8.net
ALEXITHYMIA AS MEDIATOR BETWEEN ATTACHMENT AND BPD 677
(2000), and Silk (2000) posit that borderline patients suffer extreme dis-
turbances in mood regulation. Linehan (1987) describes this emotional
dysregulation as great sensitivity to emotional stimuli, great emotional in-
tensity, and slow return to emotional baseline (see also Crowell, Beau-
chaine, & Linehan, 2009, and Kuo & Linehan, 2009). Studies have indeed
evidenced associations between BPD and emotional dysregulation (e.g.,
Gratz, Rosenthal, Tull, Lejuez, & Gunderson, 2006), yet data on the devel-
opmental pathways leading to this dysregulation are still lacking.
The inability to identify emotions such as anger, fear, or shame has been
put forward as a possible explanation for emotional dysregulation (Line-
han et al., 1993). This inability relates to the concept of alexithymia. Alex-
ithymia is a personality construct characterized by a difficulty in identify-
ing and describing feelings, a lack of fantasy, and a concrete and
externally oriented thinking style (Sifneos, 1973; Taylor, Bagby, & Parker,
1997). According to Frijda (1986), identifying emotions usually serves an
adaptive function by providing information about a given situation and
the different actions that are possible in that situation. It is assumed that
borderline patients often cannot identify what emotions they feel and
hence what caused the emotions. This inability may increase distress and
trigger a range of dysfunctional behaviors characteristic of BPD aimed at
reducing negative affects (e.g., suicidal/self-harming behaviors). The in-
ability to identify feelings has been shown to be an important component
responsible for dysfunction of the emotional system in BPD (Wolff, Stigla-
mayr, Bretz, Lammers, & Auckenthaler, 2007). It thus seems reasonable
to consider alexithymia as a risk factor for BPD. Several studies show as-
sociations between alexithymia and BPD scores (Berenbaum, 1996; Mod-
estin, Furrer, & Malti, 2004; see also Bach, de Zwaan, Ackard, Nutzinger,
& Mitchell, 1994); however, these samples did not specifically include bor-
derline patients.
In turn, alexithymia may stem from untoward attachment experiences.
According to Linehan’s etiological model for borderline pathology (Linehan
et al., 1993), invalidating environments where the expression of private
emotional experiences is not tolerated impede the understanding and la-
beling of emotions (see also Fonagy, Target, Gergely, Allen, & Bateman,
2003). Several studies show negative associations between alexithymia
and attachment security (Hexel, 2003; Meins, 2008; Montebarocci, Codis-
poti, Baldaro, & Rossi, 2004; Troisi, D’Argenio, Peracchio, & Petti, 2001;
Wearden, Lamberton, Crook, & Walsh, 2005). Also, experiences of child-
hood maltreatment and inadequate parenting are common among border-
line patients (see Widom, Czaja, & Paris, 2009, for a review). According to
Bateman and Fonagy (2006), both insecure attachment and dysfunctional
affect regulation constitute vulnerability factors for the development of
BPD. In the present model, we propose more specifically that alexithymia
mediates the effect of attachment on the development of BPD.
Although many studies show associations between attachment and
alexithymia (Hexel, 2003; Meins, 2008; Montebarocci et al., 2004; Troisi
678 DEBORDE ET AL.
et al., 2001; Wearden et al., 2005) and between the inability to identify
feelings and BPD symptoms (Berenbaum, 1996; Modestin et al., 2004;
Wolff et al., 2007), none examine the link between attachment and BPD
via alexithymia. In addition, previous research on attachment, emotional
regulation, and BPD has mostly been conducted on adult samples (see
Miller, Muelhenkamp, & Jacobson, 2008). Research findings on adults
cannot be transposed to adolescents because their emotional conscious-
ness is only emergent (Lerner & Steinberg, 2004) and because it is a time
when support from parents decreases and autonomous emotion regula-
tion is just beginning to be expected (Allen & Manning, 2007). Adolescence
is also a key period because the struggle for autonomy is likely to reacti-
vate unresolved attachment-related issues (Allen & Miga, 2010) and be-
cause it is the time when borderline symptoms often appear.
The following hypotheses were therefore tested on an adolescent sam-
ple: (1) BPD is associated with attachment; (2) BPD is associated with
alexithymia; and (3) alexithymia mediates the association between attach-
ment and BPD.
METHOD
PARTICIPANTS AND PROCEDURE
To test these hypotheses, one group of borderline patients and one group
of nonclinical adolescents were recruited. The samples were drawn from a
European longitudinal research project (European Research Network on
BPD [EURNET BPD]) investigating the diagnostic stability of BPD from
adolescence to young adulthood (13–18 years). The research network in-
volved five academic psychiatric departments specializing in treating ado-
lescents and young adults in France, Belgium, and Switzerland. This
study was approved by the French Ethical Committee (Comité de Protec-
tion des Personnes) and data were collected in an anonymous database
accepted by the French National Committee for Personal Freedoms (Com-
mission Nationale Informatique et Libertés).
Borderline participants were recruited in adolescent psychiatry de-
partments. Patients were considered for inclusion when they presented
at least five of the nine DSM-IV borderline criteria according to their
psychiatrist. Patients with psychotic disorders were excluded from the
study for feasibility reasons. Among the patients selected, BPD diagno-
sis was verified after administration of a semistructured interview
(SIDP-IV) confirming DSM-IV criteria. Psychiatric comorbidity was ex-
plored using a semistructured interview assessing DSM-IV criteria (Kid-
die-SADS).
The diagnostic interviews were conducted by a team of five clinical psy-
chologists and psychiatrists experienced in the assessment of DSM-IV Axis
I and II disorders in adolescents. To obtain high levels of reliability, the
research team participated in several training sessions, including the
ALEXITHYMIA AS MEDIATOR BETWEEN ATTACHMENT AND BPD 679
commented scoring of videotaped interviews and a training session con-
ducted by the developers of the Kiddie-SADS (Boris Birmaher and Mary
Kay Gill). Final research diagnoses were established by the best-estimate
method on the basis of the interviews and any additional relevant data
from the clinical record according to the LEAD standard (Pilkonis, Heape,
Ruddy, & Serrao, 1991). The interrater reliability for SIDP-IV was calcu-
lated from independent ratings of 10 videotaped interviews. The kappa
coefficient for agreement on the presence or absence of a BPD was very
high (0.84), and the values for the presence/absence of the other person-
ality disorders ranged from 0.54 to 1.00.
Because the number of male participants was too small (N = 15) and
because borderline symptomatology varies according to gender (Johnson
et al., 2003), only female participants were included. All participants, and
at least one of their parents for those under 18 years of age, gave their
written informed consent. After this screening procedure, all patients filled
out self-report questionnaires in their respective psychiatry departments.
Ninety-five female adolescents with a DSM-IV clinical diagnosis of BPD
were referred to the study by their psychiatrists. The formal diagnosis of
BPD according to SIDP-IV criteria was confirmed for 74 participants.
Twenty borderline patients had incomplete data on the self-report ques-
tionnaires and were excluded from the final sample of the study, which
was composed of 54 outpatients.
Because patients were mostly from the upper-middle classes and were
still studying, an advertisement for the study was placed in schools and
universities to recruit participants for the control group. The procedure
with the nonclinical sample was identical to that with the borderline sam-
ple. Control participants were screened in order to make sure they did not
have BPD (according to the SIDP-IV) or current or lifetime mental disor-
ders (according to the Kiddie-SADS). For better contrast, adolescents who
had consulted a psychiatrist or psychologist were also excluded from the
study. Fifty-one control participants matched for socioeconomic variables
were thus included.
There were no significant age differences between the two groups (Mpatients =
16.52; SD = 1.18; Mcontrol = 16.35; SD = 1.04; t = –0.76; NS). Regarding pa-
rental employment status, only three fathers (all in the clinical group) and
eight mothers (four in each group) were unemployed. Three levels of edu-
cation were considered: (1) some secondary education, (2) some postsec-
ondary education, and (3) higher education diploma. Most fathers had
higher education diplomas (67% among controls versus 57% among pa-
tients). Mothers from the control group had more often had postsecond-
ary education (49% versus 27% among mothers of patients) while moth-
ers of patients were more likely to have a higher education diploma (41%
versus 33% among controls). Nevertheless, there were no significant dif-
ferences between the control group and the clinical group concerning pa-
rental education (χ2 = 5.73, NS for the mother; χ2 = 3.96, NS for the fa-
ther).
680 DEBORDE ET AL.
MEASURES
The Structured Interview for DSM-IV Personality Disorders (SIDP-IV; Pfohl,
Blum, & Zimmerman, 1997) was used to confirm BPD diagnosis among
patients and screen for personality disorders among all participants. Bor-
derline severity for each of the nine criteria was coded as absent (0), sub-
liminal (1), present (2), or massive (3). Borderline severity scores thus var-
ied from 0 to 27. The SIDP-IV has shown good psychometric properties on
adolescent and young adult samples (Chabrol et al., 2002).
The Kiddie-SADS (Kaufman, Birmaher, & Brent, 1996; see Kaufman et
al., 1997, for data on psychometric properties) was used to verify the ab-
sence of psychiatric disorders among control participants and to assess
psychiatric comorbidity among patients. Diagnoses were established ac-
cording to DSM-IV criteria.
A brief ad hoc self report questionnaire was administered in order to
make sure control participants had never consulted for a psychiatric dis-
order, and to obtain sociodemographic data (e.g., parental employment
status and education).
The 20-item Toronto Alexithymia Scale (TAS-20; Bagby, Parker, & Tay-
lor, 1994a, 1994b) is a self-report scale composed of items ranging from 1
(strongly disagree) to 5 (strongly agree). The 20 items of the TAS are clus-
tered into three factors corresponding to the theoretical dimensions of
alexithymia: (F1) Difficulty Identifying Feelings, (F2) Difficulty Describing
Feelings, and (F3) Externally Oriented Thinking. TAS-20 scores are reli-
able, and the three-factor structure is replicable (Bagby et al., 1994a). The
TAS-20 is currently the most widely used measure of alexithymia, and
considerable work has gone into testing its reliability and validity (Bagby
et al., 1994b; Parker, Taylor, & Bagby, 2003; Taylor, Bagby, & Parker, 2003).
The Relationship Styles Questionnaire (RSQ; Bartholomew & Horowitz,
1991) is a self-report instrument with 30 items rated on a 5-point scale. It
is designed to measure a four-category model of adult attachment: secure,
fearful, preoccupied, and dismissing attachment. The mean rating for
each of the four subscales is computed, generating four continuous vari-
ables. Each attachment style is characterized by a particular underlying
model of self and others (Schafer & Bartholomew, 1994), which can be
scored as two separate continuous variables. Construct validity of the self
and others dimensions has been demonstrated (Bartholomew, 1990), as
well as convergent and discriminant validity (Griffin & Bartholomew, 1994).
In addition, moderate to high test–retest stability has been established
(Schafer & Bartholomew, 1994).
STATISTICAL ANALYSES
Descriptive analyses were conducted in order to examine the main char-
acteristics of both controls and patients. Independent sample t tests were
used to explore differences between the two groups.
ALEXITHYMIA AS MEDIATOR BETWEEN ATTACHMENT AND BPD 681
Associations between attachment, alexithymia, and borderline severity
were examined using Pearson correlations. To test the hypothesis that
alexithymia mediates the association between attachment and borderline
severity, the model proposed by Baron and Kenny (1986) was used. Ac-
cording to this model, mediation can be established if four conditions are
met: (1) the independent variable (attachment) affects the dependent vari-
able (BPD); (2) the independent variable affects the mediator (alexithymia);
(3) the mediator affects the dependent variable after the effect of the inde-
pendent variable on the dependent variable is taken into account; and (4)
the effect of the independent variable on the dependent variable is reduced
when the effect of the mediator on the dependent variable is taken into ac-
count. The mediation model is a causal model: The mediator is assumed
to cause the outcome and not vice versa. The total population (54 BPD + 51
nonclinical) was used to perform these analyses.
RESULTS
PRELIMINARY ANALYSES
The majority of adolescents with BPD met the criteria for at least one cur-
rent Axis I disorder (n = 47; 87%). Eating disorders were the most fre-
quently observed comorbidity (n = 24; 44%), followed by mood disorders
(n = 21; 39%), substance use disorders (n = 9; 17%), anxiety disorders (n =
6; 11%), and disruptive behavior disorders (n = 6; 11%). Borderline ado-
lescents showed high rates of comorbid Axis II personality disorders: ob-
sessive-compulsive (n = 21; 39%), avoidant (n = 10; 19%), antisocial (n =
7; 13%), paranoid (n = 7; 13%), dependent (n = 5; 9%), histrionic (n = 2;
4%), schizotypal (n = 2; 4%), and narcissistic (n = 2; 4%) personality disor-
ders. No schizoid personality disorder was found, probably because psy-
chotic patients were excluded from the study.
Table 1 presents the main characteristics for each group concerning
borderline severity and mean scores on the self-report questionnaires
(TAS-20 and RSQ). Borderline severity scores in the control group ranged
from 0 to 9 (M = 2.24; SD = 2.59) whereas those of patients varied from 10
TABLE 1. Characteristics of Study Participants:
Borderline Severity, Alexithymia, and Attachment Styles
CONTROLS
Mean ± SD (Range) PATIENTS
Mean ± SD (Range) t
BPD 2.24 ± 2.59 (0–9) 16.61 ± 4.44 (10–27) −20.1****
TAS-20 49.78 ± 9.28 (30–68) 57.75 ± 11.24 (28–81) −4.09****
RSQ
Secure 3.33 ± 0.41 (2.40–4.20) 3.00 ± 0.51 (1.80–4.00) 4.33****
Fearful 2.63 ± 0.79 (1.25–4.75) 3.02 ± 0.89 (1.25–5) −4.49*
Dismissing 2.66 ± 0.64 (1.40–3.80) 2.68 ± 0.87 (1–5) −0.12
Preoccupied 3.13 ± 0.71 (1–4,75) 3.46 ± 0.79 (1.75–5) −2.26*
Model of self 1.47 ± 0.50 (1–2) 1.25 ± 0.44 (1–2) 2.41*
Model of other 1.81 ± 0.39 (1–2) 1,71 ± 0.46 (1–2) 1.25
Note. NSNonsignificant; *p < .05; **p < .01; ***p < .005; ****p < .001.
682 DEBORDE ET AL.
to 27 (M = 16.61; SD = 4.44). The alexithymia mean score in the clinical
group was 57.75, which is above the alexithymia cut-off score (56). The
alexithymia mean score in the control group was 49.78.
MAIN ANALYSES
A series of t tests was conducted to examine differences in attachment and
alexithymia between the two groups (Table 1). Analyses revealed that,
compared to controls, borderline adolescents had significantly higher
alexithymia scores. Borderline patients were also less secure, more fear-
ful, more preoccupied, and had a more negative model of self. There were
no significant differences for the dismissing style and the model of others.
Table 2 presents the correlation coefficients among the six attachment
dimensions, alexithymia, and borderline severity. All measures were inter-
correlated except for dismissing attachment and model of others, which
were not associated with BPD and TAS scores (see Table 2). The attach-
ment scales associated with the TAS-20 and with borderline severity (i.e.,
secure, preoccupied, fearful, and model of self) were retained for the sub-
sequent mediation analyses. Table 3 presents the four mediation analyses
linking these attachment dimensions with borderline severity scores.
A partial mediating effect of alexithymia was found between secure at-
tachment and borderline severity. Secure attachment predicted low levels
of both borderline severity (9.8%, β = –0.31; p < .001) (condition 1) and
alexithymia (6.2%, β = –0.25; p < .01) (condition 2). While alexithymia was
a significant predictor of borderline severity (β = 0.40; p < .001) when se-
cure attachment was taken into account (condition 3), secure attachment
was less related to borderline severity (β = –0.21; p < .05) when alexi-
thymia was taken into account (condition 4).
A complete mediating effect of alexithymia was found between fearful
attachment and borderline severity. Fearful attachment significantly pre-
dicted both borderline severity (5.4%, β = 0.23; p < .05) (condition 1) and
alexithymia (5.1%, β = 0.23; p < .05) (condition 2). But fearful attachment
no longer explained borderline severity once alexithymia was taken into
account, whereas alexithymia did (β = 0.43; p < .001) (conditions 3 and 4).
No mediating effect of alexithymia was found between preoccupied at-
tachment and borderline severity.
Another partial mediating effect of alexithymia was also found between
model of self and borderline severity. Model of self predicted both border-
TABLE 2. Correlation Coefficients Between Borderline
Severity, Alexithymia, and Attachment Styles
BPD TAS-20
RSQ
Fearful Dismissing Secure Preoccupied Model
of self Model
of other
BPD 1 .46*** .23* –.27 –.31**** .36**** –.30*** –.76
TAS-20 .46*** 1 .23* –.01 –.25* .22* –.20* –.06
Note. NSNonsignificant; *p < .05; **p < .01; ***p < .005; ****p < .001.
ALEXITHYMIA AS MEDIATOR BETWEEN ATTACHMENT AND BPD 683
line severity (9.1%, β = –0.30; p < .005) (condition 1) and alexithymia
(3.8%, β = –0.20; p < .05) (condition 2), but model of self was less related
to borderline severity (β = –0.22; p < .05) when alexithymia was taken into
account (condition 4).
To summarize, the protective effect of secure attachment and model of
self for borderline severity was only partially explained by low levels of
alexithymia. The association between preoccupied attachment and bor-
derline severity was not mediated by alexithymia. Conversely, the predic-
tive power of fearful attachment with regard to borderline severity is ex-
plained by alexithymia (i.e., complete mediation).
DISCUSSION
In accordance with previous findings on adults (Webb & McMurran, 2008),
these results show that borderline adolescents are more alexithymic than
their matched controls. As expected, patients were also significantly more
insecure than nonclinical participants. As in previous research (Lyons-
Ruth, 2008 ; Westen, Nakash, Thomas, & Bradley, 2006), secure attach-
ment was negatively associated with borderline severity.
Mediation analyses suggest that secure attachment is a protective factor
in the development of BPD. This is consistent with Bowlby’s (1969/1982)
view that secure attachment is central to personality development. Re-
sults further suggest that this protective effect is partly due to the ability
to identify and express emotions. Bowlby proposed that a secure base is
necessary for the exploration of internal states. This exploration allows
TABLE 3. Regression Analyses Testing the Mediating
Role of Alexithymia in the Relationship Between
Attachment Style and Borderline Severity
Predicted
variables Variables in equation R2
total β
BPD Secure Style 9.8% –.31****
Alexithymia Secure Style 6.2% –.25**
BPD Secure Style & Alexithymia 25.0%
Secure Style –.21*
Alexithymia .40****
BPD Preoccupied Style 12.7% .36****
Alexithymia Preoccupied Style 4.9% .22*
BPD Preoccupied Style & Alexithymia 27.6%
Preoccupied Style .27***
Alexithymia .40****
BPD Fearful Style 5.4% .23*
Alexithymia Fearful Style 5.1% .23*
BPD Fearful Style & Alexithymia 22.6%
Fearful Style .14NS
Alexithymia .43****
BPD Model of self 9.1% –.30***
Alexithymia Model of self 3.8% –.20*
BPD Model of self & Alexithymia 25.0%
Model of self –.22*
Alexithymia .41***
Note. NSNonsignificant; *p < .05; **p < .01; ***p < .005; ****p < .001.
684 DEBORDE ET AL.
better identification of emotions, which in turn promotes mature and effi-
cient affect regulation (see also Gergely & Watson, 1996). Conversely, be-
cause insecure attachment is associated with lack of sensitive caregiving
(see de Wolff & van IJzendoorn, 1997), the findings also corroborate Line-
han’s model (Linehan et al., 1993). According to this model, the develop-
ment of BPD occurs within an invalidating environment in which emo-
tional displays are considered unwarranted, thus compromising the
understanding and labeling of emotions. The person thus fails to learn
how to solve the problems contributing to these emotional reactions, giv-
ing rise to more extreme emotional, behavioral, and cognitive dysregula-
tion (see also Fonagy et al., 2003).
Among the insecure attachment dimensions considered, preoccupied
and fearful attachments were associated with BPD. Preoccupied and fear-
ful attachments both imply a negative model of self. Further analyses
showed that a part of the effect of the negative model of self on borderline
severity is direct, whereas another part is mediated by alexithymia. The
RSQ (Bartholomew & Horowitz, 1991) was developed according to Bowl-
by’s (1969/1982) concept of internal working model (IWM). Depending on
the quality of care, people develop IWMs of self as more or less lovable and
of others as more or less reliable and loving. Batholomew (1990) proposed
that the negative model of self (i.e., fearful and preoccupied styles) relates
to anxiety (whereas the negative model of others relates to avoidance).
Anxiety refers to high vigilance concerning caregiver availability, frequent
verbal or physical contact with the caregiver, intense distress during sepa-
ration, and anger and resistance at the caregiver’s return (Hazan & Shav-
er, 1987). Our findings concerning the negative model of self offer a better
understanding of many BPD symptoms, such as anxiety and anger. Bor-
derline adolescents’ attachment styles can account for the constant wor-
rying about caregiver availability and the anger this leads to, as well as
intense and labile affects and relationships. Also, people with a negative
model of self behave in accordance with caregivers’ expectations rather
than with their true self. Proper identification of feelings is thus compro-
mised, rendering them more vulnerable to borderline symptoms. This fits
nicely with the theory developed by Fonagy et al. (2003), according to
which an «alien self» (Fonagy, Target, & Gergely, 2000)—arising from in-
sensitive caregiving—added to limited reflective capacities, constitutes the
roots of BPD.
Although preoccupied and fearful attachment styles were both predic-
tive of borderline severity, alexithymia only mediated the link between
fearful attachment and borderline severity. What differentiates fearful at-
tachment from preoccupied attachment is the model of others, negative
for the former and positive for the latter. People with a negative model of
others expect rejection from others when seeking comfort and reassur-
ance. Avoidance of closeness to minimize disappointment logically limits
the sharing of emotions. This is also consistent with Linehan’s etiological
model of BPD (Linehan et al., 1993).
ALEXITHYMIA AS MEDIATOR BETWEEN ATTACHMENT AND BPD 685
The current study has certain limitations that must be considered.
First, because control participants were selected as having no psychiatric
disorder, representativeness of this sample is questionable. This proce-
dure was nevertheless preferred so as to obtain greater contrast between
the clinical and nonclinical groups and identify borderline specificity more
clearly. In future studies, comparisons with other disorders would also
enable understanding of the psychopathological pathways specific to bor-
derline adolescents.
Second, although the present borderline sample is quite large compared
to those of most studies on adolescents, only female participants were in-
cluded, thus limiting the generalizability of our results. Future research
examining the links between attachment, alexithymia, and BPD is still
needed to understand the development of BPD among male adolescents.
This seems particularly useful given that borderline symptomatology var-
ies significantly according to gender (Johnson et al., 2003).
Third, reservations can be made regarding the assessments. Concern-
ing alexithymia, even if the TAS-20 is the most widely used questionnaire,
there is a growing consensus that the way in which people represent and
regulate emotions is, in part, implicit and not accessible to self-knowledge
(e.g., Westen & Blagov, 2007). In this respect, self-reports, which call for
explicit self-knowledge, probably miss the crucial component of the way in
which people actually process affective states. The use of a more objective
measure (e.g., Observer Alexithymia Scale ; Haviland, Warren, & Riggs,
2000) could provide a more reliable assessment of affect regulation.
Attachment was also measured using only a brief self-report. Thus par-
ticipants’ responses reflected subjective evaluations of self rather than ac-
tual attachment strategies. Nevertheless, subjective representations may
be relevant in understanding the processes at work in psychopathology. It
should also be noted that it is avoidance—rather than anxiety—that is
likely to lead to biased self-reports (Cassidy, 1994). In the present study,
borderline patients proved to have preoccupied and fearful (i.e., anxious)
rather than avoidant attachment styles.
In other studies, disorganized attachment has been identified as a vulner-
ability factor for the development of BPD (Bateman & Fonagy, 2006; Lyons-
Ruth, 2008). A task for future research would be to examine mediation
between attachment, alexithymia, and BPD, including a measure of disor-
ganization (e.g., unresolved loss or trauma, hostile/helpless state of mind).
CONCLUSION
In short, this study provides support for an etiopathogenic model accord-
ing to which part of the association between attachment and borderline
severity is mediated by alexithymia. This process already seems to be at
work during adolescence. Because in this period of life emotional con-
sciousness is only emergent, the ability to reflect upon emotions could be
expected to be limited. Yet the present study suggests that interindividual
686 DEBORDE ET AL.
differences with respect to the ability to identify and express emotions are
already determining factors for borderline symptomatology. Therefore
therapies aiming to increase emotional consciousness (e.g., Dialectical Be-
havior Therapy, Linehan, 1987; Transference Focused Psychotherapy,
Yeomans, Clarkin, & Kernberg, 2002; Mentalization-Based Treatment,
Bateman, & Fonagy, 2006), which have been developed for adults, could
also reasonably be considered for treatment of borderline adolescents.
REFERENCES
Allen, J. P., & Manning, N. (2007). From
safety to affect regulation: Attachment
from the vantage point of adolescence.
New Directions for Child and Adoles-
cent Development, 117, 23–39.
Allen, J. P., & Miga, E. M. (2010). Attach-
ment in adolescence: A move to the
level of emotion regulation. Journal of
Social and Personal Relationships,
27(2), 181–190.
Bach, M., de Zwaan, M., Ackard, D., Nutz-
inger, D. O., & Mitchell, J. E. (1994).
Alexithymia: Relationship to personal-
ity disorders. Comprehensive Psychia-
try, 35(3), 239–243.
Bagby, R. M., Parker, J. D., & Taylor, G. J.
(1994a). 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.
(1994b). The twenty-item Toronto
Alexithymia Scale. II. Convergent, dis-
criminant, and concurrent validity.
Journal of Psychosomatic Research,
38, 33–40.
Baron, R. M., & Kenny, D. A. (1986). The
moderator–mediator variable distinc-
tion in social psychological research:
Conceptual, strategic and statistical
considerations. Journal of Personali-
ty and Social Psychology, 51, 1173–
1182.
Bartholomew, K. (1990). Avoidance of inti-
macy: An attachment perspective.
Journal of Social and Personal Rela-
tionships, 7, 147–178.
Bartholomew, K., & Horowitz, L. (1991). At-
tachment styles among young adults:
A test of a four-category model. Jour-
nal of Personality and Social Psycholo-
gy, 61, 226–244.
Bateman, A., & Fonagy, P. (2006). Mentaliza-
tion-based treatment: A practical guide.
Oxford: Oxford University Press.
Berenbaum, H. (1996). Childhood abuse,
alexithymia and personality disorder.
Journal of Psychosomatic Research,
41(6), 585–595.
Bowlby, J. (1982). Attachment and loss: Vol
1. Attachment (2nd ed.). New York: Ba-
sic Books. (Original work published in
1969)
Cassidy, J. (1994). Emotion regulation: In-
fluences of attachment relationships.
Monographs of the Society for Research
in Child Development, 59, 228–283.
Chabrol, H., Chouicha, K., Montavany, A.,
Callahan, S., Duconge, E., & Szul-
man, H. (2002). Personality disorders
in a nonclinical sample of adolescents.
Encéphale, 28, 520–524.
Corrigan F. M., Davidson, A. & Heard, H.
(2000). The role of dysregulated amyg-
dalic emotion in borderline personality
disorder. Medical Hypotheses, 54(4),
574–579.
Crowell, S. E., Beauchaine, T. P., & Linehan,
M. M. (2009). A biosocial developmen-
tal model of borderline personality:
Elaborating and extending Linehan’s
theory. Psychological Bulletin, 135(3),
495–510.
De Wolff, M. S., & van IJzendoorn, M. H.
(1997). Sensitivity and attachment: A
meta-analysis on parental anteced-
ents of infant attachment. Child Devel-
opment, 68, 571–591.
Fonagy, P., Target, M., & Gergely, G. (2000).
Attachment and borderline personali-
ty disorder: A theory and some evi-
dence. Psychiatric Clinics of North
America, 23, 103–122.
Fonagy, P., Target, M., Gergely, G., Allen,
J. G., & Bateman, A. (2003). The de-
velopmental roots of borderline per-
sonality disorder in early attachment
ALEXITHYMIA AS MEDIATOR BETWEEN ATTACHMENT AND BPD 687
relationships: A theory and some
evidence. Psychoanalytic Inquiry, 23,
412–459.
Frijda, N. H. (1986). The emotions. New York:
Cambridge University Press.
Gergely, G., & Watson, J. S. (1996). The so-
cial biofeedback theory of parental af-
fect-mirrorring: The development of
emotional self-awarness and self-
control in infancy. International Jour-
nal of Psycho-analysis, 77, 1–31.
Gratz, K. L., Rosenthal, M. Z., Tull, M. T.,
Lejuez, C. W., & Gunderson, J. G.
(2006). An experimental investigation
of emotion dysregulation in borderline
personality disorder. Journal of Abnor-
mal Psychology, 115, 850–855.
Griffin, D. W., & Bartholomew, K. (1994).
Model of the self and others: Funda-
mental dimensions underlying mea-
sures of adult attachment. Journal of
Personality and Social Psychology, 67,
430–445.
Haviland, M. G., Warren, W. L., & Riggs,
M. L. (2000). An observer scale to
measure alexithymia. Psychosomatics,
41(5), 385–392.
Hazan, C., & Shaver, P. R. (1987). Romantic
love conceptualized as an attachment
process. Journal of Personality and So-
cial Psychology, 59, 270–280.
Hexel, M. (2003). Alexithymia and attach-
ment style in relation to locus of con-
trol. Personality and Individual Differ-
ences, 35, 1261–1270.
Johnson, D. M., Shea, M. T., Yen, S., Battle,
C. L., Zlotnick, C., Sanislow, C. A., et
al. (2003). Gender differences in bor-
derline personality disorder: Findings
from the Collaborative Longitudinal
Personality Disorders study. Compre-
hensive Psychiatry, 44(4), 284–292.
Kaufman, J., Birmaher, B., & Brent, D.
(1996). Kiddie Schedule for Affective
Disorders and Schizophrenia—Present
and Lifetime Version (K-SADS-PL). Pitts-
burgh, PA: University of Pittsburgh
School of Medicine.
Kaufman, J, Birmaher, B., Brent, D., Rao,
U., Flynn, C., Moreci, P., et al. (1997).
Schedule for Affective Disorders and
Schizophrenia for School-Age Chil-
dren—Present and Lifetime Version
(Kiddie-SADS-PL): Initial reliability and
validity data. Journal of the American
Academy of Child and Adolescent Psy-
chiatry, 36, 980–988.
Kuo, J., & Linehan, M. (2009). Disentangling
emotion processes in borderline per-
sonality disorder: Physiological and
self-reported assessment of biological
vulnerability, baseline intensity, and
reactivity to emotionally evocative
stimuli. Journal of Abnormal Psychol-
ogy, 118(3), 531–544.
Lerner, R. M., & Steinberg, L. (2004). Hand-
book of adolescent psychology (2nd
ed.). Hoboken, NJ: Wiley.
Linehan, M. (1987). Dialectical behavior
therapy for borderline personality dis-
order: Theory and method. Bulletin of
the Menninger Clinic, 51, 261–276.
Linehan, M. M., Heard, H. L., & Armstrong,
H. E. (1993). Naturalistic follow up of
a behavioral treatment for chronically
parasuicidal borderline patients. Ar-
chives of General Psychiatry, 50(12),
971–974.
Lyons-Ruth, K. (2008). Contributions of the
mother–infant relationship to disso-
ciative, borderline, and conduct symp-
toms in young adulthood. Infant Men-
tal Health Journal, 29, 203–218.
Meins, E. H.-W. (2008). Understanding alexi-
thymia: Associations with peer attach-
ment style and mind-mindedness. Per-
sonality and Individual Differences,
45, 146–152.
Miller, A. L., Muehlenkamp, J. J., & Jacob-
son, C. M. (2008). Fact or fiction: Di-
agnosing borderline personality disor-
der in adolescents. Clinical Psychology
Review, 28, 969–981.
Modestin, J., Furrer, R., & Malti, T. (2004).
Study on alexithymia in adult non-
patients. Journal of Psychosomatic Re-
search, 56(6), 707–709.
Montebarocci, O., Codispoti, M., Baldaro,
B., & Rossi, N. (2004). Adult attach-
ment style and alexithymia. Personali-
ty and Individual Differences, 36(3),
499–507.
Parker, J. D., Taylor, G. J., & Bagby, R. M.
(2003). The 20-item Toronto Alexi-
thymia Scale. III. Reliability and facto-
rial validity in a community popula-
tion. Journal of Psychosomatic Research,
55, 269–275.
Pfohl, B., Blum, N., & Zimmerman, M.
(1997). Structured Interview for DSM-
IV Personality Disorders: SIDP-IV.
Washington, DC: American Psychiat-
ric Press.
Pilkonis, P. A., Heape, C. L., Ruddy, J., &
688 DEBORDE ET AL.
Serrao, P. (1991). Validity in the diag-
nosis of personality disorders: The use
of the LEAD standard. Psychological
Assessment: A Journal of Consulting
and Clinical Psychology, 3(1), 46–54.
Roberts, R. E., Attkisson, C. C., & Rosen-
blatt, A. (1998). Prevalence of psycho-
pathology among children and adoles-
cents. American Journal of Psychiatry,
155(6), 715–725.
Schafer, E., & Bartholomew, K. (1994). Reli-
ability and stability of adult attach-
ment patterns. Personal Relationships,
1, 23–43.
Sifneos, P. E. (1973). The prevalence of alex-
ithymic characteristics in psychoso-
matic patients. Psychotherapy and
Psychosomatics, 22, 255–262.
Silk, K. R. (2000). Borderline personality dis-
order: Overview of biological factors.
Psychiatric Clinics of North America,
23, 61–76.
Taylor, G. J., Bagby, R. M., & Parker, J. D.
(1997). Disorders of affect regulation:
Alexithymia in medical and psychiatric
illness. New York: Cambridge Univer-
sity Press.
Taylor, G. J., Bagby, R. M., & Parker, J. D.
(2003). The 20-item Toronto Alexi-
thymia Scale. IV. Reliability and facto-
rial validity in different languages and
cultures. Journal of Psychosomatic Re-
search, 55, 277–283.
Troisi, A., D’Argenio, A., Peracchio, F. & Pet-
ti, P. (2001). Insecure attachment and
alexithymia in young men with mood
symptoms. Journal of Nervous and
Mental Disease, 189, 311–316.
Wearden, A. J., Lamberton, N., Crook, N., &
Walsh, V. (2005). Adult attachment,
alexithymia, and symptom reporting:
An extension to the four category
model of attachment. Journal of Psy-
chosomatic Research, 58, 279–288.
Webb, D., & McMurran, M. (2008). Emotion-
al intelligence, alexithymia and bor-
derline personality disorder traits in
young adults. Personality and Mental
Health, 2, 265–273.
Westen, D., & Blagov, P. S. (2007). A clinical-
empirical model of emotion-regulation:
From defense and motivated reason-
ing to emotional constraint satisfac-
tion. In J. J. Gross (Ed.), Handbook of
emotion regulation (pp. 373–392). New
York: Guilford Press.
Westen, D., Nakash, O., Thomas, C., & Brad-
ley, R. (2006). Clinical assessment of
attachment patterns and personality
disorder in adolescents and adults.
Journal of Consulting and Clinical Psy-
chology, 74(6), 1065–1085.
Widom, C. S., Czaja, S. J., & Paris, J. (2009).
A prospective investigation of border-
line personality disorder in abused
and neglected children followed up
into adulthood. Journal of Personality
Disorders, 23(5), 433–446.
Wolff, S., Stiglmayr, C., Bretz, H. J., Lam-
mers, C.-H., & Auckenthaler, A.
(2007). Emotion identification and
tension in female patients with bor-
derline personality disorder. British
Journal of Clinical Psychology, 46(3),
347–360.
Yeomans, F. E., Clarkin J. F., & Kernberg,
O. F. (2002). A primer of transference-
focused psychotherapy for the border-
line patient. Northvale, NJ: Jason Ar-
onson.
