Attachment & Human Development Vol 2 No 3 December 2000 328-346

Internal working models of

attachment during late
childhood and early
adolescence: an exploration of
stability and change
MASSIMO AMMANITI, MARINUS H. VAN
IJZENDOORN, ANNA MARIA SPERANZA and
RENATA TAMBELLI

ABSTRACT This article addresses the question of how the transition from late
childhood to early adolescence influences the organization of attachment. The
applicability of a measure for attachment representations in early adolescence,
the Attachment Interview for Childhood and Adolescence (AICA), was
explored. The AICA is based on the Adult Attachment Interview, which was
adapted in minor ways to the early adolescent age-group. It was hypothesized
that attachment shows considerable stability from late childhood to early
adolescence, although some changes might become manifest especially because
distancing mechanisms toward the parents may be activated in this penod. Also,
stability may be different for the various secure and insecure attachment clas-
sifications. Lastly, because gender differences become larger dunng the tran-
sition from childhood to adolescence, attachment differences between boys and
girls were explored. The same 31 Italian participants (14 girls, 17 boys)
completed the AICA at 10 years and at 14 years of age. The AICA attachment
classification distributions did not differ from Adult Attachment Interview
(AAI) distributions in comparable but older adolescent or young adult samples.
The stability of attachment security was considerable: 74% (k = .48). The
stability of the dismissing and secure categories was somewhat higher than the
stability of the (small) preoccupied and unresolved categories. The participants
tended to show more dismissing strategies across the four years, and to report
more rejection from their parents. It was suggested that the activation of
dismissing defense mechanisms might be necessary to keep parental figures at
some distance in order to achieve a more definite personal identity. Finally, no
significant gender differences in attachment emerged during the transition from
late childhood to early adolescence.

Correspondence to: Professor Massimo Ammaniti, Dipartimento di Psicologia dei Processi di

Sviluppo e Socializzazione, Universitä di Roma 'La Sapienza", Via dei Marsi, 78 - 00185 Rome,
Italy.

Attachment & Human Development ISSN 1461-6734 print/1469-2988 online © 2000 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI:110.1080/14616730010001587
 A M M A N I T I ET AI..·. IWMS IN LATE C H I L D H O O D / E A R L Y A D O L E S C E N C E 329

Keywords: attachment - stability - adolescence

Attachment theory, conceived originally äs a general theory of personality

development (Bowlby, 1969), has stressed the importance of the nature and
development of human attachment relationships. Attachment not only
enhances the survival during infancy, it also promotes adaptive development
during the life span (Bowlby, 1969, 1977). Although attachment processes
have been studied primarily in infancy, they remain critical also in other
phases of the life span, even though they may change in organization (Sroufe,
1979; Cicchetti, Cummings, Greenberg, & Marvin, 1990).
From a developmental perspective attachment theory postulates the
following processes:
1 the development and stabilization of different attachment patterns during
the first five years or so after birth are based on the children's genetic con-
stitution and their relational experience with their caregivers;
2 the dyadic attachment interactions are internalized by the children in their
representational System äs internal working models;
3 the transitions from infancy to childhood, adolescence, adulthood and old
age may require specific attachment strategies and may lead to develop-
mental discontinuities;
4 intergenerational transmission of attachment strategies from parents to
children may occur frequently äs parental attachment representations
translate into parenting behaviors.
In recent years attachment theory and research have advanced in several
new directions (Bretherton, 1992). One of the most promising is the study of
attachment across the life span. Now that substantial progress has been made
towards understanding the developmental process of attachment in infancy,
there is a growing interest in the transformation, development, function and
outcomes of different attachment organi/ations across the life span
(Ainsworth, 1985, 1991; Main, Kaplan, & Cassidy, 1985; Kobak & Sceery,
1988; Greenberg, Siegel, & Leitch, 1983; Kotler & Omodei, 1988; Greenberg,
Cicchetti, & Cummings, 1990; Cassidy & Shaver, 1999). In the current
paper, we study attachment äs an organizational construct in the
Bowlby-Ainsworth tradition.
The affectional bonds of attachment relevant for the motivational System
(Sameroff, 1983; Maslin-Cole & Spieker, 1990) amplify during the life span,
including not only the attachment of children to their parents, but also the
bonds of parents to their children, bonds with other kin, sexual partners, and
the bonds between friends. Of course, these classes of relationships may play
a different role in the attachment System and especially in regulating the feit
security, considering also their interplay with other behavioral Systems
(Ainsworth, 1991). Once the attachment bonds are stabilized during infancy
and early childhood, they tend to persist in the subsequent years especially
330 MIACHMLNl & HUMAN DEVELOPMENT VOL. 2 NO. 3

in stable child-reanng environments (Bowlby, 1988), although they may

undergo transformations and reintegrations during later years. Thus for
every new developmental phase, children are continually renegotiating the
Integration and the Balance between being m connection with significant
others and being autonomous. The 'goal-corrected partnership' (Bowlby,
1969), äs well the growing confidence in mutual understanding, enables chil-
dren in the pre-school years to tolerate the Separation from parents for longer
periods and with less distress, and to amplify the exploration of their
expanded world and their connecting with more people, especially peers.
From this perspective on attachment in early adolescence we may question
whether the growing detachment from parents could change the organization
of attachment bonds, if parents become more distant and do not sustain the
reciprocal attachment äs it was in the past. A related question deals with the
differential influence of puberty changes and other developmental processes
on the various attachment organizations, äs they have been assessed in adult-
hood using the Adult Attachment Interview (Main & Goldwyn, 1998; Hesse,
1999). Individuais with secure Interviews tend to maintam a coherent dis-
cussion of their personal attachment history and its influence on personality
development, being particularly able to integrale positive with negative feel-
ings and using their self-reflective functioning (Fonagy, Steele, Steele, Moran,
& Higgit, 1991) in understanding, m a more complex way, their own personal
mental states äs well their parents' states. For these reasons, during ado-
lescence they might adjust in a better way and understand the personal and
the relational changes involving their parents, while mamtaining strong
bonds with them.
In the case of individuals with dismissing Interviews, äs they tend to min-
imize the importance and impact of attachment relationships in their own
lives, through idealization of the parents or devaluing of them, the adolescent
processes might increase further these modes of distancing from parents.
Finally, individuals with preoccupied Interviews, who are not able to describe
their attachment biography coherently and who show an inability to move
beyond an excessive preoccupation with or sense of involvement in attach-
ment relationships, may be stimulated by the adolescent processes, especially
cognitive changes, to distance themselves from the parents and to face in a
better way contradictory feelings and thoughts about them.
Considering the developmental tasks of adolescence we suggest that during
this period the models of attachment, developed during infancy and child-
hood, might be revised and re-adjusted, but the models built during infancy
and childhood may be mamtained m stable conditions. The few studies about
attachment in adolescence (van IJzendoorn & Bakermans-Kranenburg, 1996;
Kobak & Sceery, 1988; Allen & Hauser, 1991; Hesse, van IJzendoorn, &
Main, 1993; Sagi et al., 1994) show that the distribution of attachment is not
significantly different from the normal adults' attachment distributions. It
should be stressed that the samples in these studies included young males and
females in late adolescence and young adulthood. Not one of the studies
 A M M A N I T I rr A I . · IWMS IN I A I F c n i L D i i o o D / t ARLY ADOLFSCI-NCL· 331

included young males and females in late childhood and early and middle
adolescence, significant periods for affective and cognitive changes (Inhelder
& Piaget, 1958; Brooks-Gunn, Petersen, & Eichhorn, 1985; Dämon & Hart,
1988; Paikoff & Brooks-Gunn, 1991; Hauser & Smith, 1992).
In any case there is a lack of longitudinal studies describmg how attach-
ment may be influenced by adolescence, comparing different age periods
within the same sample. The continuity of attachment from infancy to late
adolescence has been investigated in at least two longitudinal studies: the
Adult Attachment Interview responses of 17-year-olds (Hamilton, 1995) and
of 21-year-olds (Waters, Merrick, Treboux, Crowell, & Albersheim, 1995)
were found predictable from Strange Situation behavior to the mother 16 and
20 years previously. In Hamilton's study 77% of adolescents retained their
infant attachment Status with respect to the secure vs. insecure distinction,
while in Waters and his colleagues' study the stability of attachment was 70%.
Some other longitudinal studies showed less stability (Grossmann, Gross-
mann, & Zimmermann, 1999) but it is unclear whether this should be at least
partially attributed to the use of non-standard attachment assessments tech-
niques or to environmental discontinuity (see Main and Hesse, in press).
Stability studies provided still limited Information on continuities and
changes of patterns of attachment during the different periods of life cycle
and in high-risk conditions. Investigations evidenced short-term test-retest
reliability: 78% stability (k = .63) across the three organized attachment cat-
egories with Interviews conducted two months apart (Bakermans-Kranen-
burg & van IJzendoorn, 1993), while an Israeli study of 59 College students
conducted three months apart yielded 90% test-retest stability (k = .79; Sagi
et αι., 1994). Stability was also tested across a longer period: 18 months in a
study conducted by Crowell et al. (1996) with 86% stability in three cat-
egories, and in Canadian mothers between a pre-birth interview and Inter-
views conducted at 11 months of infant age with 90% three-category stability
(Benoit & Parker, 1994). The outcome of the last study is particularly import-
ant since it might have been expected that the major life transition occasioned
by the birth of a first child would change a mother's state of mind with respect
to attachment, äs could happen with the transformations during adolescence.
In sum, we explore the following hypotheses:

1 the Attachment Interview for Childhood and Adolescence (AICA) shows

promising construct validity and other psychometric characteristics when
compared with the Adult Attachment Interview (AAI; Main & Goldwyn,
1998);
2 attachment shows considerable stability from the end of late childhood to
early adolescence, although some change may occur äs a consequence of
distancing mechanisms towards the parents during this period;
3 stability may differ for the vanous attachment categories: adolescents with
secure AAIs at 10 years may be more resilient to the challenges of
adolescence, adolescents with dismissing AAIs at 10 years may be
 332 ATTACHMENT & HUMAN DEVELOPMENT VOL. 2 NO. 3

strengthened in their distancing from the parents, while the adolescents

with preoccupied A Als at 10 years may show more instability because of
the increasing detachment.

METHOD

Sample
Thirty-one young adolescents (17 male, 14 female) were participants in this
study drawn from two classes of a junior high school in Rome. Only
participants with intact families were selected for the longitudinal study.
Participants were native Italians from middle- (40%) and working-class
(60%) families. The mean age of the participants at recruitment was 10 years
(M = 126.5 months, SD = 5.1). In the second assessment, participants were
re-interviewed four years after their initial Interviews (mean age, 14 years;
M = 171.4 months, SD = 2.8).

Procedure
Data were collected at an interval of four years. In both the first and the
second assessment AICA (Attachment Interview for Childhood and Ado-
lescence; Ammaniti et al., 1990) was administered individually to the partici-
pants in the school by different Interviewers, and the Interviews were
audiotaped. No other data were collected.

Measures
The Attachment Interview for Childhood and Adolescence (AICA; Amman-
iti et al., 1990) is a revised Version of AAI (Adult Attachment Interview) for
participants m late childhood and early adolescence. The structure of the
interview and the sequence of the questions were unchanged, but the lan-
guage was simplified and adjusted to age. Explanations were added to clarify
some questions, e.g. 'Teil me about your relationship with your parents äs a
httle child. What I mean by the word "relationship" is what you did together,
how you got along, what you feit for each other'; 'Do you think your past
relationship with your mother affects you now that you're older? I mean the
person you are, what you like, your relationships with others.' Moreover,
only two questions were removed äs they addressed issues relevant for
parents only (e.g. questions related to fear of loss of a child or wishes for a
child). However, these changes did not modify the structure of the interview,
providing enough Information to rate overall states of mind.
The AAI is a semi-structured interview that probes alternately for general
descriptions of relationships with the main attachment figures in childhood,
specific supportive or contradicting memories, and descriptions of current

? ' , "· ;»''.'l"44%^v5'4>«ysrilÄcTI

S v - ^ttwW^;^^^
 ΑΜΜΛΝΓΠ ET A L.·. IWMS IN LATE C H I L D H O O D / E A R L Y A D O L E S C E N C E 333

relationships with them. Adults are asked to retrieve attachment-related

autobiographical memories from early childhood and to evaluate these mem-
ories and their effects from their current perspective, so that the structural
dimension of the transcript is coded rather than its content. The AAI coding
System, developed by Main & Goldwyn (1998; see also Hesse, 1999), was
used to classify adolescents into one of four categories for overall state of
mind with respect to attachment: (1) Dismissing of attachment relationships;
(2) Secure, freely valuing of attachment; (3) Preoccupied with attachment
relationships; and (4) Unresolved with respect to past loss or trauma.
In addition to these classifications, the interview transcripts were scored
on 12 nine-point scales. Five scales assessed 'probable childhood experience'
and childhood relationship to each parent (loving, rejecting, neglecting,
involving, pushing to achieve), and seven ('representational') scales assessed
the present mental representation of attachment (idealization, anger, deroga-
tion, passivity, coherence of transcript, lack of memory, and meta-cognition).
Some of the scales were scored separately for mothers and fathers: idealiza-
tion, anger, and derogation. The separate scores were combined, and the
overall scores on these representational scales were used in the analyses (Main
& Goldwyn, 1998). The scale for meta-cognition did not show sufficient vari-
ance to be included in the analyses.
In terms of psychometric qualities of the AAI, predictive validity studies
revealed concordance between the mother's state of mind with respect to
attachment and her 1-year-old child's attachment classification obtained in
retrospective (five years later, Main & Goldwyn, 1989, 75%), concurrent
(Ainsworth & Eichberg, 1991, 80%) and antecedent assessments (during
pregnancy, Fonagy, Steele, & Steele, 1991, 75%). Concerning discriminant
validity, the AAI classifications turned out to be independent of memory,
verbal and performance intelligence, and social desirability (Bakermans-
Kranenburg & van IJzendoorn, 1993; Sagi et al., 1994). Short-term test-retest
reliability of the three major categories of the AAI ranged from 78% (k = .63)
to 90% (k = .79) (Bakermans-Kranenburg & van IJzendoorn, 1993; Sagi et
al., 1994).
The AICAs were transcribed verbatim and coded by three independent
raters following Main and Goldwyn's coding System (1998) for the AAI. In
order to adjust the coding System to children's narratives, a group of experts
trained in the AAI rating scales produced specific criteria for the crucial
coherence scale (Dazzi, De Coro, Ortu, & Speranza, 1999; Muscetta, Dazzi,
De Coro, Ortu, & Speranza, 1999). Coherence has been considered äs central
in the assessment of mental representations of attachment even in children
(Oppenheim, Nir, Warren, & Emde, 1997). Specifically, some violations of
Grice's maxims were considered only moderately incoherent äs they recurred
often in this age-group. The following problems were taken into account:
difficulty in differentiating between present and past experiences (Ί was
happy with him . . . he was really good with me . . . he give me everything I
want'; Ήε liked very much to go to the country, he still likes it and we go to

if%ff^ffi|w-'.;
334 Λ Ι Ι Λ Ο Ι Ι Μ ί Ν Ι & HUMAN VOL 2 NO 3

the country, but I didn't hke it so much'), difficulty in reaching an abstract

level of descnption ('With my fathei I have a close relationship because he
helps me to do my homework we go out together to see some place
he was good with me because he bought everythmg I wanted'), normaliza
tion äs expression of contmuity of the expenence rather than a dismissing
strategy (when asked if the mterviewee thought there was anythmg different
now in his or her relationship with the mother 'Always normal äs it was
when I was a child', 'Maybe my father was more severe than my mother, äs
all the fathers with their sons') All these examples were rated äs moderately
coherent
All transcnpts were scored by at least two raters, blind to the other assess
ment (first or second interview) Two coders had passed the 30-case rehabil
ity test provided by Main & Hesse and one the 18 case test The mean
concordance rate across all AAI classifications for all pairs of raters was 82%
(k = 64) for the four categones Dismissing, Secure, Preoccupied, and Un
resolved attachment representation Consensual classifications, achieved by
rater discussion, were used for analysis Intercoder agreement for the expen
ences scales was äs follows Loving mother r = 78, Lovmg father r = 60,
Rejectmg mother r = 83, Rejectmg father r = 71, Neglectful mother r = 84,
Neglectful father r = 82, Involving mother r = 82, Involvmg father r = 72,
Pushmg-to-achieve mother r = 63, Pushmg-to-achieve father r = 70 Inter
coder agreement for the representational scales was äs follows Ideahzation
mother (r = 66), Ideahzation father (r = 68), Derogation mother (r = 83),
Derogation father (r = l 00), Anger mother (r = 77), Anger father (r = 75),
Coherence transcript (r = 84), Passivity (r = 70), Lack of memory (r = 91)
The intercoder rehability of Derogation father reached the maximum size
because this variable was rather skewed more than 90% of the participants
at 10 years of age and two thirds of participants at 14 years of age did not
mdicate any derogation

RESULTS

The results of our study are presented in two parts As a first step we discuss
the attachment classifications distnbution of the sample and we compare it
with othcr adolescent or young adult samples As a second step we study
stabihty and change of the categoncal attachment representations and con-
tinuous expenential and representational scales across a four-year penod in
this adolescent sample

AICA distributions compared with other samples

In the first assessment at 10 years of age, the 31 Interviews were classified äs
follows 9 (29%) dismissing, 20 (64%) secure/autonomous, and 2 (7%) pre
occupied The distnbution that mcluded the unresolved category was äs
 AMMANITI ET A L.: IWMS IN LATE CHILDIIOOD/EARLY ADOLESCENCE 335

follows: 8 (26%) dismissing, 20 (64%) secure/autonomous, l (3%) preoccu-

pied, and 2 (7%) unresolved/cannot classify. In the second assessment the 31
Interviews were classified similarly: 11 (36%) dismissing, 16 (52%)
secure/autonomous and 4 (13%) preoccupied. With the unresolved category
the distribution was: 11 (36%) dismissing, 16 (52%) secure/autonomous, 3
(10%) preoccupied and l (3%) unresolved/cannot classify.
We compared the distribution of our sample at 10 years and at 14 years
with the combined distribution of comparable low-risk samples of non-
Italian adolescents and young adults (Allen, Häuser, & Borman-Spurrell,
1996; Fonagy et al., 1996; Hesse et al., 1993; Kobak & Sceery, 1988; Sagi et
al., 1994: see Table 1). The multinomial test for comparisons between inde-
pendent distributions was used. The 10-years-old distribution was not sig-
2
nificantly different from the combined distributions of young adults, χ (2,
2
N = 382) = 1.72, ns, for the three-way attachment distribution; and χ (3, 7V
= 341) = 5.53, ns, for the four-way attachment distribution. The 14-years-old
2
distribution did not differ significantly either, χ (2, 7V = 382) = 1.29, ns, for
2
the three-way; and χ (3, 7V = 341) = 6.72, ns, for the four-way classifications
distribution. Standardized residuals for the various classifications were not
significant either.
In order to control for cultural factors and language, we compared the
current distributions of young Italian adolescents with a sample of 38 middle-
class Italian mothers, interviewed when their infants were l year old
(Muscetta, Bovet, Candelori, Mancone, & Speranza, 1999). Distribution of
this adult sample was äs follows: 9 (24%) dismissing, 25 (66%) secure, and 4

Table l Distributions of attachment classifications in young adolcscent and young

adult samples
Young
Attachment combined Italian
dassification AICA 10 AICA 14 adults' mothers2

Dismissing 9 (29) 11 (36) 93 (26) 9 (24)

Secure 20 (64) 16 (52) 206 (59) 25 (66)
Preoccupied 2 ( 7) 4 (13) 52 (15) 4 (10)

Dismissing 8 (26) 11 (36) 64 (21) 7 (19)

Secure 20 (64) 16 (52) 157 (51) 24 (63)
Preoccupied 1 (3) 3 (10) 39 (13) 4 (11)
Unresolved 2 ( 7) 1 ( 3) 50 (16) 3 ( 8)
Key: 'The combined young adult sample is a combination of samples reported in Allen et al.
(1996), Fonagy et al. (1996), Hesse et al. (1993), Kobak and Sceery (1988) and Sagi et al. (1994)
(age ränge: 18-28 years)
2
Dcrived from Muscetta et al. (1999).
 336 ATTACHMENT & HUMAN DEVELOPMENT VOL. 2 NO. 3

(10%) preoccupied Interviews. Our adolescent distributions were not differ-

ent from this adult distribution: [χ2 (2, N = 69) = .87, ns, for the three-way
distribution at 10; χ2 (2, N = 69) = 2.94, ns, for the three-way distribution at
14]. The distribution including the unresolved category was: 7 (19%) dis-
missing, 24 (63%) secure, 4 (11%) preoccupied, and 3 (8%) unresolved for
loss. Both adolescent distributions were not significantly different from this
adult sample [χ2 (3, N - 69) = 2.45, ns, for the four-way distribution at 10; χ2
(3, N = 69) = 2.85, ns, for the four-way distribution at 14].

Stability and change of adolescent attachment

The attachment classification distributions at 10 and 14-years old were rather
similar, and the general stability across four years was 71% (k = .48, p = .001).
Nine out of the 31 cases (29%) were not stable over time (see Table 2).
The various attachment classifications differed in stability across the four-
year period. Comparing the secure vs. insecure classifications across time we
found that 74% (k = .48) remained stable; 78% of the dismissing vs. non-
dismissing Interviews remained within the same category (k = .55); and only
50% of the preoccupied vs. non-preoccupied Interviews were stable (k = .27).
Moreover, Interviews of two subjects were classified äs Unresolved at 10
years of age, but only one remained Unresolved at 14 (50%).
In this adolescent sample, the stability and change of attachment experi-
ences äs reported in the interview might be relevant for interpreting the
developmental pattern of the attachment classifications. In Table 3, means and
Standard deviations for the experiences scales are presented, äs well äs stab-
ility figures.
Four experiences scales showed significant correlations across the four
years period: Loving father, Neglectful mother, Rejecting mother, and Reject-
ing father. The latter two scales for Rejection also showed significant change.
The participants reported more frequent or more intensive rejections by both
their parents across this four-year period. The increase in rejection experi-
ences fits well into the pattern of more dismissiveness dunng early ado-
lescence.

Table 2 Stability of attachment of classifications between 10 years and 14 years of age

(stability is indicated by bold numbers)
Attachment 14 years Total
classifications
10 years Secure Dismissing Preoccupied
Secure 14 4 2 64%
Dismissing 1 7 1 29%
Preoccupied 1 0 1 7%

Total 52% 36% 13% 100%

,\ v,-u
 A M M A N I T I ET AL.\: I W M S IN LATE C H I L D H O O D / E A R L Y A D O L E S C E N C E 337

Table 3 Stability and change in experiences scales from 10 to 14 years

Experiences scale M (SD) i

Loving mother
10 years 4.4 (1.7)
14 years 4.4 (1.8) .26 0.12
Loving father
10 years 4.7 (1.7)
14 years 4.2 (1.9) .47** 1.55
Neglectful mother
10 years 2.5 (1.9)
14 years 2.3 (2.1) .54** 0.65
Neglectful father
10 years 2.0 (1.5)
14 years 2.1 (1.5) .23 0.10
Push-to-achievc mother
10 years 1.8 (1.3)
14 years 1.6 (1.0) .08 0.73
Push-to-achieve father
10 years 1.4 (0.8)
14 years 1.5 (1.0) .18 0.57
Rejecting mother
10 years 1.9 (1.5)
14 years 2.9 (1.8) .66*** 4.03***
Rejecting father
10 years 1.9 (1.4)
14 years 2.6 (1.9) .49** 2.28*
Involving mother
10 years 1.6 (1.1)
14 years 2.0 (1.2) -.12 1.08
Involving father
10 years 1.3 (0.8)
14 years 1.8 (1.4) -.15 1.66

Key. * p < .05, **

In order to investigate the nature of the stability and change of the ado-
lescents' representations in more detail, we tested whether the most import-
ant representational scales - idealization, lack of recall, derogation, anger,
passivity, coherence - showed continuity across the four-year period. In
Table 4, means and Standard deviations for these scales are provided.
Two representational attachment scales showed significant change in
means: lack of recall; and derogation. Both scales point at a dismissing state
of mind and both scales showed an increase in lack of recall and derogation
respectively. That is, across the four years of early adolescence, participants
showed more lack of recall and more derogation of attachment; at the same
time, these indicators of dismissiveness were significantly stable across time
338 A I I A C H M E N T & HUMAN D I V h L O P M h N l VOL 2 NO 3

Table 4 Stability and change representational scales from 10 to 14 years

Representational scale M (SD) t

Idcali?ation
10 years 29 (15)
14 years 30 (1 5) 36 -046
Lack of recall
10 years 30 (23)
14 years 39 (2 2) 45 -2 19
Derogation
10 years 14 (10)
14 years 21 (1 2) 46 -332
Anger
10 years 18 (10)
14 years 17 (1 0) 32 053
Passivity
10 years 24 (16)
14 years 28 (1 4) 23 -1 41
Coherencc
10 years 48 (16)
14 years 44 (1 3) 37 123

Key p < 05, p < 01

in tcrms of rank ordermg Although dismissmg attachment classifications

were rather stable, the means of the attachment representations scales indi-
cated that dismissiveness became mtensified across the four years. We did not
find other sigmficant changes m mean scores on the attachment scales. Non-
parametnc Spearman correlations showed the same pattern of results com-
pared with the Pearson correlations presented m Table 3 and 4

DISCUSSION

In the present study we introduce an Instrument for assessing attachment in

late childhood and early adolescencc (Attachment Interview for Childhood
and Adolescence, AICA) that we developed by revising and adaptmg the
Adult Attachment Interview (AAI, Main & Goldwyn, 1998) The study aims
to contnbute to the measurement of the nature and quahty of attachment
relationships at different stages of the hfe cycle Specifically, the AICA is a
research technique meant to assess 'mternal workmg models' or 'mental rep-
resentations of attachment relationships' (Main & Solomon, 1986, Main,
Kaplan, & Cassidy, 1985) in late childhood and early adolesccnce The Instru-
ment focuses upon the assessment of attachment through language and other
representational processes. We consider that AICA is particularly interestmg
for the reason that m late childhood and m early adolescence there are not
 Α Μ Μ Λ Ν Γ Π ET Λ L.·. I W M S IN Ι.ΛΤΕ C H I L D I I O O D / E A R L Y A D O L E S C E N C K 339

comparable Instruments to those available in infancy (Ainsworth, Blehar,

Waters, & Wall, 1978) and in early and middle childhood (Main & Cassidy,
1988; Bretherton, Ridgeway, & Cassidy, 1990). In this period of the life cycle
explorative techniques based on observation of attachment behavior and on
symbolic representation in doll play may not be adequate while Interviews
based on verbal interaction may stimulate better the co-operation of the-sub-
jects and the collection of interesting Information about their mental world.
In order to facilitate the understanding of the questions and the co-
operation of the subject we formulated the questions of the Interview in a
simplified way, maintaining, however, the sequential order of the AAI and
removing the questions which explore the adult's relationship with his or her
child. Also the coding System based on the one used by Main and Goldwyn
(1998) was adapted considering the narrative competence in late childhood
and early adolescence. As we have evidenced in our Interviews, the coherence
of transcripts in children and early adolescents must be carefully assessed
because sometimes the 'narrative diachronicity' (Bruner, 1994) is not
respected. These violations happen when children speak about their relation-
ship with the parents during infancy oscillating from the past to the present
and vice versa.
Another feature of the adolescents' narrative about the relationship with
the parents is the particular emphasis on what they do and how they act with
their parents instead of valuing the mental interchange. While in adults the
emphases on 'fun', on 'activities', or even on materialism or material objects
(Main & Goldwyn, 1998) are indicators of the dismissing category of attach-
ment, in early adolescence this emphasis reflects a particular interest in activi-
ties and behavior sequences. For this reason the scale of coherence should in
this period consider the characteristics and the qualities of the narrative from
a developmental perspective. The developmental perspective does not imply
the necessity of a special training for researchers who want to use AICA.
They should, however, consider the specificity of co-operation at this age,
which is still influenced by 'linguistic egocentricity' (Piaget, 1923), a special
tendency of children to neglect some essential Information in the narrative in
order to let the listener understand the message.
Our findings confirm the first hypothesis about the use of the AICA in
late childhood and early adolescence. Results show that the distributions of
attachment categories in our 10- and 14-year-old participants are not signifi-
cantly different from those collected in other studies using the AAI with ado-
lescent and young adult samples. The adolescent attachment distribution
does not differ from the attachment distribution found in a sample of Italian
adults either. As in other studies, the secure category is modal in the distri-
bution of attachment categories at 10 years (64%) äs well äs at 14 years old
(52%) (van IJzendoorn & Bakermans-Kranenburg, 1996; van IJzendoorn &
Kroonenberg, 1988).
In general, the AICA Interviews do not produce more problems in con-
ducting or coding them with participants äs young äs 10 years of age than the

4
^^κΤ. Λ* jt ' ü 1 ** ", ""l li
* ι Λ*·«·
340 ATTACHMENT & HUMAN DEVEIOPMENI VOL. 2 NO. 3

AAI used in normal adult samples. The AAI coding System (Main &
Goldwyn, 1998) appears to be reliably adaptable to the AICA material, and
a high degree of intercoder agreement can be reached with well-trained AAI
coders.
Considering our second hypothesis about stability and change of attach-
ment representations during early adolescence, we found considerable stab-
ility of attachment security from 10 to 14 years of age: 74% (k = .48). The
stability across four years is about the same äs found in some short-term test-
retest studies (Bakermans-Kranenburg & van IJzendoorn, 1993; Sagi et al.,
1994). This finding is impressive when we take into account that in the latter
studies the time interval was extremely brief - about two to three months -
and that the participants were adult women or young adults. The stability of
attachment across four years in our study is particularly cogent not only for
the long test-retest period but also because it refers to the period of late child-
hood and early adolescence when major changes take place in body and
sexual functioning äs well äs in the affective and cognitive fields.
Concerning our third hypothesis about the different attachment categories,
we found a rather high stability for the dismissing category (78%, k = .55),
and for the secure category (74%, k = .48), whereas the preoccupied category
(50%, k = .27) and the unresolved state of mind (50%) classifications were less
stable. Of course, the preoccupied and unresolved categories consisted of only
a few participants, and we should be careful in drawmg defimte conclusions
on the basis of few observations. Furthermore, the coding of the AICA cannot
be done without some coding errors, äs is indicated by the high but not perfect
intercoder agreement. Therefore, a ceiling effect for stability can be expected,
äs coding errors will cause at least some instability. Nevertheless, the stability
of the representational AICA scales also points in the direction of somewhat
more stable scales for idealization, lack of recall, and derogation, which indi-
cate dismissing attachment representations.
The dismissing category, therefore, seems to display somewhat more stab-
ility. The vicissitudes of young adolescent development could affect working
models of attachment, broadening dismissing strategies äs they are connected
to detachment processes to parental figures. From this perspective dismiss-
ing attachment strategies, internalized during childhood, may stabilize to a
higher degree during adolescence (Weiss, 1991). Activation of dismissing
defense mechanisms (Fonagy, Moran, Steele, & Steele, 1992) may in this
period be adaptive, äs Main and Hesse (1990) pointed out, because they
permit the maintaining of a close relationship with parents and relieve the
anger or anxiety typical of this phase. In our study we found that from 10 to
14 years old the adolescents intensified their dismissing tendency of deroga-
tion and lack of recall, and they perceived their parents äs more rejectmg.
Similar detachment processes may destabilize the participants with pre-
occupied Interviews who may be stimulated to revise their internal working
models of attachment. As the Grossmanns pointed out (Grossmann &
Grossmann, 1991), changes in adaptive behavioral strategies could occur
 A M M A N I T I ET AL.. IWMS IN LATE C H I L D H O O D / E A R L Y A D O L E S C E N C E 341
throughout the whole life cycle, and durmg adolescence these changes may
be especially affected by the need to keep parental figures at a distance in
order to achieve a more definite individual identity. In particular, adolescents
with preoccupied Interviews may feel the largest discrepancy between their
past attachments and their current strivmg for more autonomy.
According to our hypothesis we expected the secure attachment caiegory
to be the most stable because it would allow for better adaptive capacity and
resiliency. Nevertheless, even participants with secure Interviews do not
always remain on the same developmental trajectory: of 10-year-olds with
secure Interviews only 14 out of 20 had also secure Interviews at 14 years of
age, and 6 participants moved mto an insecure category. Only two of the 10-
years-old children with insecure Interviews became secure at 14 years of age.
Unfortunately, our study does not address the issue of the causes for change
of attachment representation. Waters et al. (1995) found that major negative
life-events such äs divorce of the parents caused their participants to deviate
from their original developmental pathway. What factors may stimulate inse-
cure children to become secure are studied in expenmental Intervention
studies in early childhood (e.g. Bakermans-Kranenburg, Juffer, & van Ijzen-
doorn, 1998). If parents become more sensitive to the children's attachment
signals the children may develop secure attachments somewhat more fre-
quently. Further (Intervention) study of similar processes in late childhood
and early adolescence is badly needed.
While psychodynamic studies, based upon climcal observations, tend to
underline adolescent crisis with identity transformation (Enkson, 1968),
attachment theory, on the contrary, emphasizes stability of attachment
models developed during infancy if environmental conditions remain the
same. Although it could be possible to hypothesize that affective, cognitive
and physical changes allow a revision of mternal working models durmg
adolescence (e.g. the metacognitive ability to reflect on one's own mental
processes; see Flavell, 1979, and Dämon and Hart, 1988), our data tend to
emphasize the contmuity of attachment organization during adolescence
confirming the fmdings of Offer, Ostrov and Howard (1981). During early
adolescence internal working models of attachment may already have incor-
porated a self-perpetuating tendency in organizing relational expenences
consistent with the existing attachment representations (Zeanah & Anders,
1987), and even the growing number of more important peer relationships
may become molded to the current attachment models.

In sum, despite the small sample size we found significant stability of attach-
ment representations from late childhood to early adolescence. The most
important change of attachment across four years consisted of an increase in
dismissiveness, which we related to the increasing need of adolescents to
become independent of their parents. The Attachment Interview for Child-
hood and Adolescence (AICA) shows considerable promise for further
research on changing attachment relationships of adolescents with their
342 AlIACHMtNT & HUMAN DEVELOPMENT VOL 2 NO 3

parents which imply 'transformations m understanding of seif, fnends, and

parents leadmg to recogmtion and expression of a greater ränge of feelmg'
(Hauser & Smith, 1992) For this reason we recommend that other
researchers should go on and Start collectmg data on attachment from 10 until
14 years old, usmg such Interviews äs AICA or the autobiographical narra-
tives used, for example, by Mary Main (Main, Kaplan, & Cassidy, 1985) Of
course, it would be mteresting to take mto account other variables, too
attachment of the parents, for instance, or biological variables, such äs the
timing and the charactenstics of biological changes associated with puberty
(Allen & Land, 1999, Spangler & Zimmermann, 1999)

ACKNOWLEDGEMENTS

We would hke to thank Mary Main, Erik Hesse, Peter Fonagy and Stuart
Hauser for their msightful comments on our paper We also thank Sergio
Muscetta, Carla Candelon, Gmliana Carosi, Alessandra De Coro, Angela
Mancone, Francesca Ortu and Laura Vismara who worked in the current
project

REFERENCES
Amsworth, M D S (1985) Attachment across the life span Bulletin of the New York
Academy of Mediane, 61, 792-812
Amsworth, M D S (1991) Attachment and other affectional bonds across the life
cyclc In C M Parkes, J Stevenson Hmde & P Marns (Eds), Attachment across
the hfe-cycle (pp 35-51) London Routledge
Amsworth, M D S , Blehar, M C , Watcrs, E , & Wall, S (1978) Patterns of attach
ment A psychological study of the Strange Situation Hillsdale, NJ Erlbaum
Amsworth, M D S , & Eichberg, C (1991) Effects on mfant-mother attachment of
mother's unresolved loss of an attachment figure, or other traumatic expenence
In C M Parkes, J Stevenson Hmde & P Marns (Eds), Attachment across the
life cycle (pp 160-183) London Routledge
Allen,] P, & Häuser, S T (1991) Prediction of young adult attachment represen
tations, psychological distress, social competence and hard drug use from famüy
interactions m adolescence Paper presented at the 59th mecting of the Society for
Research m Child Development, Seattle, WA, April
Allen, J P, Hauser, S T, & Borman Spurrcll, E (1996) Attachment theory äs a
framework for understanding sequelac of severe adolescent psychopathology
An 11 year follow up study Journal of Consulting and Chnical Psychology,
64(2), 254-263
Allen, J P, & Land, D (1999) Attachment m adolescence InJ Cassidy & P R Shaver
(Eds), Handbook of attachment Theory, research, and chnical apphcations
(pp 319-335) New York Guilford Press
Ammaniti, M , Candelon, C , Dazzi, N , De Coro, A , Muscetta, S , Ortu, F, Pola,
M , Spcranza, A M , Tambelli, R, & Zampmo, F (1990) l A L Intervista
 A M M A N I T I ET Λ L.: IWMS IN LATE C H I L D H O O D / E A R L Y A D O L E S C E N C E 343
sull'attaccamento netto, latenza (A.I.C.A., Attachment Interview for Childhood
and Adolescence). Unpublished protocol, University of Romc.
Bakermans-Kranenburg, M. J., Juffer, F., & van IJzendoorn, M. H. (1998). Intcrven-
tions with video feedback and attachment discussions: Does type of maternal
insecurity make a difference? Infant Mental Health Journal, 19(2), 202-219.
Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H. (1993). A psychpmctric
study of the Adult Attachment Interview: reliability and discriminant validity.
Developmental Psychology, 29(5], 870-879.
Benoit, D., & Parker, K. C. H. (1994). Stability and transmission of attachment across
three generations. Child Development, 65, 1444-1456.
Bowlby, J. (1969). Attachment and loss, Voll. Attachment. New York: Basic Books
(2nd ed. 1982).
Bowlby, J. (1977). The making and breaking of affectional bonds. British Journal of
Psychiatry, 130, 201-210.
Bowlby,]. (1988). A secure base. London: Routledge.
Bretherton, I. (1992). The origins of attachment theory: John Bowlby and Mary
Ainsworth. Developmental Psychology, 25(5), 759-775.
Bretherton, L, Ridgeway, D., & Cassidy, J. (l990). Assessing internal working models
of the attachment relationship: An attachment story completion task for 3-years -
olds. In M. T. Greenberg, D. Cicchetti & E. M. Cummings (Eds), Attachment in
the preschool years: Theory, research, and Intervention (pp. 273-308). Chicago:
University of Chicago Press.
Brooks-Gunn, J., Petersen, A., & Eichhorn, D. (1985). The study of maturation
timing effects in adolescence. Journal of Youth and Adolescence, 14, 149-161.
B runer, J. (1994). The narrative construction of reality. In M. Ammaniti & D. N. Stern
(Eds), Psychoanalysis and development. Representations and narratives. New
York: New York University Press.
Cassidy, J., & Shaver P. R. (Eds) (1999). Handbook of attachment. New York:
Guilford Press.
Cicchetti, D., Cummings, E. M., Greenberg, M. T, & Marvin, R. S. (1990). An
organizational perspective on attachment beyond infancy: Implications for
theory, measurement, and research. In M. T. Greenberg, D. Cicchetti & E. M.
Cummings (Eds), Attachment in the preschool years: Theory, research, and Inter-
vention (pp. 3^-9). Chicago: University of Chicago Press.
Crowell, J. A., Waters, E., Treboux, D., O'Connor, E., Colon-Downs, C., Feider, O.,
Golby, B., & Posada, G. (1996). Discriminant validity of the Adult Attachment
Interview. Child Development, 67, 2584-2599.
Dämon, W., & Hart, D. (1988). Self-understanding in childhood and adolescence.
Cambridge, UK: Cambridge University Press.
Dazzi, N., De Coro, A., Ortu, F., & Speranza, A. M. (1999). L'intervista sull'attacca-
mento in preadolescenza: Un'analisi della dimensione della coerenza. Psicologia
Clinica dello Sviluppo, ///(l), 129-153.
Erikson, E. H. (1968). Identity: youth and crisis. New York: W. W. Norton.
Flavell, J. (1979). Metacognition and cognitive meaning. American Psychologist, 84,
906-911.
Fonagy, R, Moran, G., Steele, M., & Steele, H. (1992). L'integrazione della teoria
psicoanalitica e del lavoro sull'attaccamento: La prospettiva intergenerazionale.
In M. Ammaniti & D. N. Stern (a cura di), Attaccamento e psicoanalisi. Bari:
Laterza.
 344 ATTACHMENI & HUMAN DEVELOPMENT VOL 2 NO 3

Fonagy, P, Steele, H , & Steele, M (1991) Maternal representations of attachment

dunng pregnancy predict the organization of mfant-mother attachment at one
year of age Child Development, 62, 891-905
Fonagy, P, Steele, M , Steele, H , Leigh, T, Kennedy, R , Mattoon, G , Target, M , &
Gerber, A (1996) The relation of attachment Status, psychiatnc classification,
and response to psychotherapy Journal of Consulting and Chnical Psychology,
64(\), 22-31
Fonagy, P, Steele, M , Steele, H , Moran, G , & Higgit, A (1991) The capacity for
understandmg mental states The reflective seif m parent and child and its sigmfi
cance for secunty of attachment Infant Mental Health Journal, 12(3), 201-218
Greenberg, M T, Cicchetti, D , & Cummings, E M (Eds) (1990) Attachment m the
preschool years Theory, research, and Intervention Chicago Umversity of
Chicago Press
Greenberg, M T, Siegel, J M , & Leitch, C J (1983) The nature and importance of
attachment relationships to parents and peers dunng adolescence Journal of
Youth and Adolescence, 12, 373-386
Grossmann, K E , & Grossmann, K (1991) Attachment quahty äs an orgamzer of
emotional and behavioral responses m a longitudmal perspective In C M
Parkes, J Stevenson Hmde & P Marris (Eds), Attachment across the hfe-cycle
(pp 93-114) London Routledge
Grossmann, K E , Grossmann, K , 8c Zimmermann, P (1999) A wider view of attach
ment and exploration Stability and change dunng the years of immatunty In
J Cassidy & P R Shaver (Eds), Handbook of attachment New York Guilford
Press
Hamilton, C E (1995) Contmmty and discontinuity of attachment from infancy
through adolescence Paper presented at the bienmal meeting of the Society for
Research in Child Development, Indianapolis, IN, March
Hauser, S T, & Smith, H F (1992) The development and experience of affect in
adolescence In T Shapiro & R N Emde (Eds), Affect Psychoanalytic perspec
tives Madison, WI International Umversity Press
Hesse, E (1999) The Adult Attachment Interview historical and current perspec-
tives In J Cassidy & P R Shaver (Eds), Handbook of attachment New York
Guilford Press
Hesse, E , van IJzendoorn, M H ,& Main, M (1993) Unresolved Status in the Adult
Attachment Interview Related to dissociation, absorption, and anusual behefs
Leiden, the Netherlands Center for Child and Family Studies
Inhelder, B , & Piaget, J (1958) The growth of logical thmking from childhood to
adolescence New York Basic Books
Kobak, R R , & Sceery, A (1988) Attachment m late adolescence workmg models,
affect regulation, and representations of seif and others Child Development, 59,
135-146
Kotler, T, & Omodei, M (1988) Attachment and emotional health A hfe span
approach Human Relations, 41, 619-640
Main, M , & Cassidy, J (1988) Categories of response to reumon with the parent at
age 6 Predictable from infant attachment classifications and stable over a l
month pcnod Developmental Psychology, 24, 1-12
Main, M , & Goldwyn, R (1989) Interview based adult attachment classifications
Related to mfant-mother and mfant-father attachment Unpublished manu
scnpt Umversity of California, Berkeley

Η „νϊ

ί. ^ ? "ν, l·t1 ~n Ί·*'

 AMMANITI ET A L.·. IWMS IN LATE CHILDHOOD/EARLY ADOLESCENCE 345
Main, M., & Goldwyn, R. (1998). Adult Attachment scoring and classification Systems.
Unpublished manuscript. University of California, Berkeley.
Main, M., & Hesse, E. (1990). Parents' unresolved traumatic experiences are related
to infant disorganized Status: Is frightened and/or frightening parental behavior
the linking mechanism? In M. T. Greenberg, D. Cicchetti & E. M. Curnmings
(Eds), Attachment in the preschool years: Theory, research, and Intervention
(pp. 161-182). Chicago: University of Chicago Press.
Main, M., & Hesse, E. (in press). Predicting Adult Attachment Interview response in
late adolescents from infant Strange Situation behavior with mother and father.
Manuscript in preparation. University of California, Berkeley.
Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood and
adulthood: A move to the level of representation. In I. Bretherton & E. Waters
(Eds), Growing points of attachment theory and research (pp. 66-104). Mono-
graphs ofthe Society for Research in ChildDevelopment, iO (1-2, Serial No. 209).
Main, M., & Solomon, J. (1986). Discovery of a new insecure-disorga-
nized/disoriented attachment pattern: Procedures, findings and implications for
the classification of behavior. In T. B. Brazelton & M. Yogman (Eds), Affective
development in infancy (pp. 95-124). Norwood, NJ: Ablex Press.
Maslin-Cole, C., & Spieker, S. J. (1990). Attachment äs a basis for independent moti-
vation: A view from risk and nonrisk samples. In M. T. Greenberg, D. Cicchetti
& E. M. Curnmings (Eds), Attachment in the preschool years: Theory, research
and Intervention (pp. 245-272). Chicago: University of Chicago Press.
Muscetta, S., Bovet, A. F., Candelori, C., Mancone, A., & Speranza, A. M. (1999).
Funzione riflessiva materna e Stile di attaccamento nei bambini. Psicologia Clinica
dello Sviluppo, 111(1), 109-128.
Muscetta, S., Dazzi, N., De Coro, A., Ortu, F., & Speranza, A. M. (1999). 'State of
mind with respect to attachment' and change in psychotherapeutic relationship:
A study of the coherence of transcript in a short-term psychotherapy with an
adolescent. Psychoanalytic Inquiry, 19(5), 885-921.
Offer, D., Ostrov, E., & Howard, K. I. (1981). The adolescent. A psychological self-
portra.it. New York: Basic Books.
Oppenheim, D., Nir, A., Warren, S., & Emde, R. N. (1997). Emotion regulation in
mother-child narrative co-construction: Association with children's narratives
and adaptation. Developmental Psychology, 33(2), 284-294.
Paikoff, R. L., & Brooks-Gunn, J. (1991). Do parent-child relationships change
during puberty? Psychological Bulletin, 110(1), 47-66.
Piaget, J. (1923). Le langage et la pensee chez Venfant. Neuchätel: Delachaux &
Niestle.
Sagi, A., van IJzendoorn, M. H., Sharf, M., Koren-Karie, N., Joels, T., & Mayseless,
O. (l 994). Stability and discriminant validity of the Adult Attachment Interview:
A psychometric study in young Israeli adults. Development Psychology, 30,
771-777.
Sameroff, A. (1983). Developmental Systems: Context and evolution. In W. Kessen
(Ed.), Mussen's Handbook of Child Psychology: Vol. 1. New York: Wiley.
Spangler, G., & Zimmermann, P. (1999). Attachment representation and emotion
regulation in adolescents: A psychobiological perspective on internal working
models. Attachment & Human Development, 1(3), 270-290.
Sroufe, L. A. (1979). The coherence of individual development: Early care, attachment
and subsequent developmental issues. American Psychologist, 34, 834-841.
346 ATTACHMENT & H U M A N D E V E L O P M E N T VOL. 2 NO. 3

van IJzendoorn, M. H., & Bakermans-Kranenburg, M. J. (1996). Attachment

representations in mothers, fathers, adolescems, and clinical groups: A meta-
analytic search for normative data. Journal of Consulting and Clinical Psychol-
ogy,64(\),%-2\.
van IJzendoorn, M. H., & Kroonenberg, P. M. (1988). Cross-cultural patterns of
attachment: A meta-analysis of the Strange Situation. Cbild Development, 59,
147-156.
Waters, E., Merrick, S. K., Treboux, D., Crowell, J., & Albersheim, L. (1995). Attach-
ment security from infancy to early adulthood: A 20-year longitudinal study.
Paper presented at the biennial meeting of the Society for Research in Child
Development, Indianapolis, IN, March.
Weiss, R. S. (1991). The attachment bond in childhood and adulthood. In C. M.
Parkes, J. Stevenson-Hinde & P. Marris (Eds), Attachment across the life-cycle
(pp. 66-76). London: Routledge.
Zeanah, C. H., & Anders, T. F. (1987). Subjectivity in parent-infant relationship: A
discussion of internal working models. Infant Mental Health Journal, 8(3),
237-250.