Explanation of differences regarding the acquisition of the Dutch language comprehension between Turkish and Moroccan Toddlers
J.J. Sluijmers, MSc.*, Robert
Lindeboom, Phd.**, Frans Pijpers, Phd.***,
*
**Academic Medical Centre of
***
****
e-mail: jsluijmers@ggd.amsterdam.nl
Postal
adress :
Pb. 2200
1000 CE AMSTERDAM
The Netherlands
Objective: The aim of this
retrospective trial is to investigate whether there are differences in the
development of the Dutch language comprehension between toddlers with a Turkish
(n = 105) or Moroccan (n = 126) migrant background, as well as to identify
demographic, social-pedagogic and medical factors related to the observed
differences.
Methods: The absolute differences in
the language comprehension was measured by using the Dutch version of the
Reynell Comprehension Language Scales, expressed in a Language Comprehension
Quotient (LCQ). All factors which were correlated to the LCQ-score, were being
further examined using a multiple linear regression technique.
Results: toddlers from a Turkish
background (mean age 41.5 months) showed a significant lower LCQ-score, both at
the first and second measuring point as compared to children from a Moroccan
background (mean age 41.8 months) (55 vs. 61 LCQ and 61 vs. 76 LCQ, p = <
0.01). The variability in LCQ at the second measuring point and the differences
between the first and second test could both not be explained by the Turkish or
Moroccan mother tongue (p = 0.81 and, p = 0.51 respectively). However, the
frequency of family, friends and neighbours speaking Dutch in the presence of
the children demonstrated an explained variability (r2) in LCQ of
50%.Conclusions: Turkish toddles have lower Dutch language comprehension skils
than Morrocan toddlers. Interventions such as improving the possibilities to
hear Dutch from family members and neigbours, as well as providing suitable
information regarding the bilingual education, could possibly contribute to a
decrease in these differences.
Background
Towards the end of the
nineties, a special kind of early and pre-school education was introduced
nationally in The Netherlands. The objective of this program is to avoid arrears
in education by offering children education while playing at an early age. The
early and pre-school education program has amongst other things positive
effects on language development [1,2]. Most of the children in the
Identification of demographic,
social-pedagogic and medical factors, which can explain possible causes, may
result in early intervention. The aim of this study was 1) to examine to what
extent there are differences between the level and development of Dutch
language comprehension between Turkish and Moroccan ‘early and pre-school
education’ toddlers, and 2) to identify demographic, social-pedagogic and
medical factors, which may contribute to possibly differences.
Study design
Retrospective
dossier research.
The study population consisted of all toddlers (N=245) in the age group
of 3.6 to 4 years with a Turkish or Moroccan background from the suburb
Geuzenveld/ Slotermeer, which attended the early and pre-school education
between 1997 and 2002. There were four participating early and pre-school
education programs.
Data collection
The toddlers were identified by means of the automatic registration
system at the schools. The required information was gathered from the intake
form of the school, a standardised questionnaire for the parents and test
results of the SLT. Six children were excluded because they were referred to
special education schools. For the remaining 239 children the respective
dossiers from their health clinic were collected with the database from the
child healthcare database. Eight dossiers were not recovered, four children had
moved, and four children could not be found. The information of 231 children
was recorded on an “anonymous case report form” and then entered twice into a
database (SPSS). In addition to the Reynell score was the following data which
may have some influence on language development according to literature
recorded: sexe, age in months, period in early and pre-school education in
months and language background of the most important educator (Turkish,
Moroccan or a Berber language). Social-pedagogic data includes: Language
amongst the children; language during upbringing and contacts within the family
[7]; the age when speaking the first words; [8] to be read to and watching
television; unemployment of father or mother; education of the parents; age of
the parents; family composition [9]; asylum seeker status; length of stay of
the mother in the Netherlands during first and second measurement; advise from
the child health clinic to consult an upbringing support facility (or
participate in a upbringing program). Medical data includes: familiar disorders
[10]; consanguinity [11]; progress of pregnancy
including duration and birthweight[12]; smoking; ‘APCAR’1 score;
bottle/breast feeding and possible eating disorders; results of hearing and
development screenings like the CAPAS2 hearing screening and the Van
Wiechenscheme for development (communication items) [13]; visus, otitis media
[14] and ENT operations (tubes, adeno and/or tonsillectomy).
1 APCAR= Activity (muscle tone), Pulse, Grimace (reflex irritabilaty), Appearance (skin color), Respiration.
2 Capas= Compact Amsterdam Pedo-audiometric Screening, a kind of Ewingtest.
Language comprehension here means the understanding of spoken Dutch. The
Dutch language comprehension was tested twice with the Reynell test for
language comprehension [15]. Initially two weeks after arrival at the ‘early
and pre-school education program’ and two weeks before continuing to primary
education with an interval of approximately half a year. The approximate test
results were converted to a Language Comprehension Quotient (LCQ) by using
tables in the test manual, age equivalence and a percentile score. The
indicator for the development of language comprehension was the difference
between first and second measurement results.
A multiple lineair regression technique (backwards) is used to analyse
which variables independently correlate (p <0.05) with the LCQ score of the
second measurement and the difference between first and second LCQ scores.
Results:
The characteristics of the studypopulation in relation to the language
background (Turkish of Moroccan), during the first test in the ‘early and
pre-school education program’, as well as the results of both tests, are
compiled in table 1.
Table 1. Characteristics of
the sample survey in relation to language background
|
|
Turkish language background N = 105 |
Moroccan language background N = 126 |
|
Sex:
male* |
56% |
49% |
|
Age in months (min - max)a |
41.5 (30.3 - 46.0) |
41.8 (38.2 - 47.2) |
|
Period in ‘early and pre-school education program’ in
months (min - max) |
6.7 (3 - 13.2) |
6.4 (3 -11.2) |
|
Both parents speak Berber |
Not applicable |
30% |
|
Reynell test: mean (interquartile range): |
|
|
|
LCQ 1 st measurement** |
55 (55-55) |
61 (55-72) |
|
LCQ 2 nd measurement** |
61 (57-67.5) |
76 (67-82) |
|
Age
equivalent 1st measurement** |
14 (14-19) |
23 (16-27) |
|
Age
equivalent 2nd measurement** |
26 (23-29) |
34 (28-38) |
|
Percentile
score 1st measurement** |
0 (0-0) |
0 (0-3) |
|
Percentile
score 2nd test** |
0 (0-2) |
5 (1-10) |
a) During the first test:.
*P > 0.05 (Chi square test)
**P < 0.01(Mann Whitney U-test)
There were no substantial differences between groups as far as age and
attendance of the ‘early and pre-school education program’ is concerned. In the
Turkish group there were relatively more boys, but this turned out not to be
significant. It was shown that there were substantial differences in language
comprehension between both groups. The mean LCQ score of the first measurement
was 55 LCQ for the Turkish children and 61 LCQ for the Moroccan children
(p<0.01). This difference was greater at the second measurement (61 LCQ
versus 76 LCQ, p<0.01).
The variables, which had significant correlation with the LCQ score of
the second measurement, are summarized in table 2.
Table 2.
Characteristics, which are significantly related (p< 0.05) to the LCQ-score
of the second test of the whole population (N=231).
For example: Within the researched group there were 105
children who spoke Dutch within the family with each other as a second
language. They have an average LCQ of 76.
|
|
Nr |
LCQ-score Characteristic present |
N r |
LCQ-score Characteristic absent |
P (T-test) |
|
Demographics: |
|
|
|
|
|
|
Sex male |
121 |
68 |
110 |
72 |
<0.03 |
|
Social-pedagogic: |
|
|
|
|
|
|
L2 amongst the children b |
105 |
76 |
91 |
64 |
<0.01 |
|
L2 family contacts b |
120 |
76 |
95 |
62 |
<0.01 |
|
L2 during upbringing b |
110 |
76 |
105 |
64 |
<0.01 |
|
1st words < 19 months |
191 |
70 |
37 |
65 |
0.01 |
|
Being read to > 2x a week |
50 |
74 |
73 |
68 |
0.01 |
|
Only child |
47 |
65 |
181 |
71 |
<0.01 |
|
Mother >5 years in NL |
157 |
71 |
14 |
60 |
<0.01 |
|
Advice upbringing support facility |
35 |
61 |
173 |
71 |
<0.01 |
|
Medical c |
|
|
|
|
|
|
Pregnancy <38 weeks |
21 |
64 |
182 |
71 |
0.01 |
|
Smoking of mother during pregnancy |
21 |
65 |
183 |
71 |
0.03 |
|
Birthweight < 2500 gr. |
13 |
64 |
195 |
70 |
0.03 |
|
Capas hearing screening passed |
153 |
71 |
41 |
66 |
0.01 |
|
Van Wiechen communication 1.9 years
passed |
137 |
71 |
24 |
65 |
0.01 |
|
Van Wiechen communication 3.9 years
passed |
163 |
71 |
30 |
65 |
<0.01 |
|
ENT intervention |
25 |
64 |
183 |
71 |
0.01 |
a) The numbers don’t always add up to 231,
because the information was not recorded for all of the children, or the
children did not appear or were not called.
b) Nt2 = Dutch as second language in combination
with speaking the own language.
c) Definitions
and stop points conform Platform child healtcare [24]
A higher LCQ score (10 points or more) as opposed to a lower LCQ score
(see opposite column) is strongly related to Dutch as second language (L2)
during family contact; L2 during upbringing; speaking L2 amongst the children
within the family and the duration of stay from the mother in the Netherlands
for longer than 5 years. Variables, which were associated with a lower LCQ
score, are amongst others: advice to consult an upbringing support facility or
program; duration of pregnancy of less than 38 weeks; smoking of the mother
during pregnancy; child birthweight of less than 2500 gram; an insufficient
CAPAS hearingscreening and an insufficient result during the Van Wiechenscheme
communications items around the age of 1.9 and 3.9 years.
Variables, which after the multiple linear regression technique,
significantly contributed to the level of Dutch language comprehension are
compiled in table 3 (With exception of ‘language background’; this factor is,
although not significant, included in the table, due to the fact that most language
tests are related to language groups).
Table 3: Variables which multivariate
(significant) contribute to the LCQ-score of the second test and the difference
between the LCQ result of the first and second test (development of Dutch
language comprehension)
|
|
Explained
variance (R2) LCQ-score
2nd test |
Explained
variance (R2) Difference
LCQ 1st and 2nd test |
|
Nt2 family
contacts |
38.4% |
9.0% |
|
LCQ-score 1st
test |
17.7% |
NA |
|
Nt2 amongst the
children |
7.2% |
NS* |
|
Advice upbringing
support facility |
3.2% |
2.3% |
|
Language
background |
NS* |
NS* |
|
|
|
|
|
R2 |
67% |
11.3% |
NA = Not applicable
NS = not significant
*> 1%
Concerning the level of language comprehension during the second test,
67% of the variance in the score can be explained by variables in the model.
The individual contribution of the variables to the explained variance in the
model were: speaking L2 during family contacts 38.4%; LCQ score during first
test 17.7%; speaking L2 amongst the children 7.2%; advise upbringing support
facility 3.2%. The language background does not contribute significantly to the
language comprehension level of the second test, or to the development of
language comprehension between the two test points (<1%)
Recommendations
Above results indicate that it may be advantageous to explore the possibility
to conduct an extensive queastionair for (expectant) parents at an early stage,
to determine the language situation and their attitude and ideas concerning the
bilingual education of there baby or toddler. Before the doctor or nurse at the
childhealth clinic – or preferably the obstetrician, like at the new
‘parent-child-centre’ in
Table 4:
Questionair in relation to exploring the multi language situation and
possibilities of the parents.*
Who spend most time with the child?
The mother/ father/ caretaker/ equal.
How well does the mother (father, caretaker) command
the own language (Dutch)?**
The mother (father, caretaker) commands the own
language (Dutch) well/ a little/ not at all.
The mother (father, caretaker) speaks the mother
tongue (Dutch) outside the home, in presence of the child, during contact with
relatives/ friends/ neighbours/ acquaintances/ colleagues.
During which situation?
(For example: shopping/ playground/ trips/ work)
During which situation at home does the mother
(father, caretaker) speak the mother tongue (Dutch) in presence of the child?
**
(For example: upbringing (consoling/ being upset/
washing/ taking to bed/ playing and cuddling) / reading aloud or telling
stories/ singing songs.)
Which language(s) are the parents (caretakers)
speaking with each other?
Which language(s) are the (elder) children in the
family speaking with each other mostly?
How much contact does the child has with Dutch
speaking peers?
In what situation?
What are the parents (caretakers) doing to improve
their Dutch (if needed)?
Do they let the child know?
Which language do the parents (caretakers) find
important for their child?
Why?
*With
permission from the author extracted and adapted from Blumenthal e.a. 1999
[21].
**This also
needs to be asked in relation to the father (or other caretakers) instead of
the mother and Dutch instead of the mother tongue.
Depending on results of the questionair and considering the situation of
the parents; ‘one place, one language strategy’ may be recommended for example.
This means that at home the mother tongue is spoken and outside Dutch [22]. If
parents are not fluent in Dutch, they can be advised to show their children that
they are doing their best to acquire the Dutch language in order to give their
child a good example and a possitive attitude towards the second language. If
one of the parents is fluent in Dutch then, ‘one person, one language’ strategy
can be considered [22]. If both parents are fluent in the Dutch language, then
is converting totally to this second language not recommended. Earlier research
showed that the possibility exists that both languages will develop
deficiencies [23].
The influence of the factors: ‘Speaking Dutch during contacts with
friends, family and neighbours (L2 family contacts)’ and ‘speaking Dutch
amongst the children within the family’, on language comprehension and
development, can be called severe. A possible underlying explanation why the
Moroccan group more often use Dutch during family contacts, is that in
It may be possible that differences in attitude,about the importance of
the mother tongue compared to the L2 between Moroccan and Turkish parents, play
a role [7]. The term ‘L2’ in this research does not relate to the quantity and
quality of the language exposure in relation to language comprehension.
Regarding the influence of quality of language exposure on development of Dutch
language comprehension, further research is required. The influence of an
‘upbringing support facility’ on language comprehension may possibly show the
relation between language development and behaviour, as was found in other
research [16, 17]. This suggests the importance of early identification of both
language and behaviour problems [18].
The determined differences in language comprehension may affect future
results at school [19]. Our study results however relate to a period of 6.5
months. Whether an early Reynell test indeed can predict school results in the
future needs to be further examened. Considering that the target group
originated from a particular area and consisted of participants of the ‘early
and pre-school education program’, it is not possible to generalise the results
for other geographical areas or other (language) groups. In addition it may be
possible that, due to the retrospective character of this study, some bias was
introduced. Therefore one must be cautious to apply the conclusions and
recommendations.
The Dutch language comprehension amongst Turkish children is lower and
develops slower than that of their Moroccan peers. Explanations are mainly
found in Dutch family contacts in presence of the child and speaking of Dutch
amongst the siblings. The Turkish or Moroccan language background is not an
explanation in its own right. It is advised that at the childhealth clinic, or
better still during pregnancy, the language situation of the parents is
analysed. Appropriate advice, suitable to the possibilities of the parents, may
allow for both languages to develop fully. Before steps are taken towards
implementation, further research is essential.
References