International Journal of Pediatric Otorhinolaryngology 80 (2016) 5360

Contents lists available at ScienceDirect

International Journal of Pediatric Otorhinolaryngology

journal homepage: www.elsevier.com/locate/ijporl

The Galker test of speech reception in noise; associations with

background variables, middle ear status, hearing, and language
in Danish preschool children
Maj-Britt Glenn Lauritsen a, Margareta Soderstrom b, Svend Kreiner c, Jens Drup d,1,
Jrgen Lous e,*
a

The Research Unit for General Practice, Centre of Health and Society, University of Copenhagen, Copenhagen, Denmark
Department of General Practice, University of Copenhagen, Copenhagen, Denmark
c
Department of Biostatistics, University of Copenhagen, Copenhagen, Denmark
d
Section for Health Informatics, University of Aarhus, Aarhus, Denmark
e
Research Unit of General Practice, Department of Public Health, University of Southern Denmark, Jagtvej 20A, DK-8270 Hjbjerg, Denmark
b

A R T I C L E I N F O

A B S T R A C T

Article history:
Received 30 August 2015
Received in revised form 8 November 2015
Accepted 13 November 2015
Available online 24 November 2015

Purpose: We tested the Galker test, a speech reception in noise test developed for primary care for
Danish preschool children, to explore if the childrens ability to hear and understand speech was
associated with gender, age, middle ear status, and the level of background noise.
Methods: The Galker test is a 35-item audio-visual, computerized word discrimination test in
background noise. Included were 370 normally developed children attending day care center. The
children were examined with the Galker test, tympanometry, audiometry, and the Reynell test of verbal
comprehension. Parents and daycare teachers completed questionnaires on the childrens ability to hear
and understand speech. As most of the variables were not assessed using interval scales, non-parametric
statistics (GoodmanKruskals gamma) were used for analyzing associations with the Galker test score.
For comparisons, analysis of variance (ANOVA) was used. Interrelations were adjusted for using a nonparametric graphic model.
Results: In unadjusted analyses, the Galker test was associated with gender, age group, language
development (Reynell revised scale), audiometry, and tympanometry. The Galker score was also
associated with the parents and day care teachers reports on the childrens vocabulary, sentence
construction, and pronunciation. Type B tympanograms were associated with a mean hearing 56 dB
below that of than type A, C1, or C2. In the graphic analysis, Galker scores were closely and signicantly
related to Reynell test scores (Gamma (G) = 0.35), the childrens age group (G = 0.33), and the day care
teachers assessment of the childrens vocabulary (G = 0.26).
Conclusions: The Galker test of speech reception in noise appears promising as an easy and quick tool for
evaluating preschool childrens understanding of spoken words in noise, and it correlated well with the
day care teachers reports and less with the parents reports.
2015 Elsevier Ireland Ltd. All rights reserved.

Keywords:
Speech recognition in noise test
Day care children
Language development
Otitis media
Hearing
Non-parametric statistics

1. Introduction
The effect of the hearing loss often accompanying otitis media
with effusion (OME) on childrens development has been a focus
of intensive research, and the results of this research are diverging
[14]. A minor hearing loss accompanying OME may be debilitating

* Corresponding author. Tel.: +0045 86273471.

E-mail address: jlous@health.sdu.dk (J. Lous).
1
Deceased.
http://dx.doi.org/10.1016/j.ijporl.2015.11.014
0165-5876/ 2015 Elsevier Ireland Ltd. All rights reserved.

to one child, but not to another. The effect apparently depends on

the presence of both protective factors and risk factors present in
the listening environment [5].
About 90% of Danish preschool children attend a day care center,
spending on average 35 h in that setting weekly. This environment is
often very noisy which puts strong demands on the childs auditory
ability, in particular during OME bouts [6]. In order to secure a true
and fair view of the nature of this problem, clinicians should appraise
these childrens ability to hear and understand speech under the
conditions prevail during daily communication which would
include measures of the noise pollution found in day care centers [5].

54

M.-B.G. Lauritsen et al. / International Journal of Pediatric Otorhinolaryngology 80 (2016) 5360

Research indicates that appraisal of speech reception in noise is

useful for evaluating the effect of OME on auditory function [7,8]
and for identifying those children who may benet from OME
surgery [4,9]. In the light of these circumstances and inspired by
the principles of the TADAST test, the Danish Galker test of speech
reception in noise was developed [10,11].
A previous paper [1] provided evidence for the construct
validity and the reliability of the Galker test for 35 items (Galker35, now just called the Galker test) using the Rasch model. The test
was well-accepted by preschool children attending the test panel
[1].
The aim of the present paper is to further explore the usefulness
of the Galker test by examining the effect of background noise on
the test results and the associations between test scores and
background noise, middle ear effusion, hearing, language comprehension, and the parents and the care takers evaluation of the
childrens hearing and language faculties.
2. Methods

2.4. Hearing
Conditioned play audiometry was employed to test air
conduction with pure tones at 500, 1000, 2000, and 4000 Hz via
earphones according to described techniques [12]. Testing was
performed in a quiet room with an AS216 Interacoustics screening
audiometer. The validity of each session was ranked on a 14-point
scale based on the childs concentration, cooperativeness, and
perceived ambient noise level with 1 representing high validity.
Sessions with validity rankings of 1 or 2 with at least 2 measured
frequencies in both ears were kept for data analysis. A summary
index of hearing level (HL) for each ear was calculated by averaging
thresholds. Children were categorized according to the average
threshold of the best ear. Four groups were dened: HL less than
22 dB, 2225, 2629 dB, and 30 dB or more. Children who were
unable to carry out audiometry were categorized in a separate
group and included as a separate group throughout analysis.
In the graphic analysis, some of the variables were collapsed to
even fewer groups to be able do a more robust analysis; for
example, hearing groups were reduced to only three groups.

2.1. Study population

2.5. Middle ear status
All 34-year-old children attending 20 ordinary day care
centers situated in the municipality of Hillerd, Denmark, were
invited to participate, in total 513 children. Informed consent was
received from parents of 388 children, and 376 of these children
accepted participation. Galker test results for 6 children were lost
for technical reasons. This paper presents results based on data
from the remaining 370 children who were all normally developed.
2.2. Examinations
The tests were performed in the mornings and included the
Galker test, play audiometry, otoscopy, and tympanometry
performed by one physician (MGL). Within 2 weeks, the Reynell
test was used by an authorized speech therapist to test the
participants verbal comprehension. Furthermore, completed
questionnaires were collected from the parents and the day care
teachers during the sessions.
The study was approved by the Research Ethics Committee of
Frederiksborg County, Denmark.
2.3. The Galker test
The test is an audio visual two-alternative word recognition test
consisting of 35 word pairs presented on computer by a visible
female speaker against a background of speech-shaped white noise
at a xed signal-to-noise ratio of
2 dB. Colorful pictures
representing the word pairs, predominantly minimal pairs, emerge
on the screen, and the child now has 5 s to point at the correct
picture. Testing time was about 6 min. The total score of the test is
calculated as the number of correct pictures (range 035). The test
was administered in a quiet room. The physician used the mouse to
click on the pictures chosen by the child, and answers were
registered automatically in the computer. The Galker test is
described in more detail elsewhere [1].
For investigation of the effect of background noise, 107 children
were randomly selected and tested twice with the Galker test. On
Test 1 (baseline), they were all tested with the procedure described
above. On Test 2, 67 of the children were retested with the same
procedure again (Noise group). The remaining 40 children were
retested with the background noise switched off (No-noise group).
For the nonparametric analysis of associations, the Galker
scores were classied into 6 ordinal categories with nearly equal
numbers of children: Correct answers were scored as 018, 1923,
2425, 2627, 2829, and 3035 correct answers.

To categorize middle ear status, tympanometry was conducted.

Plain otoscopy was performed in relation to tympanometry to
determine the presence of cerumen or grommets. The children
were examined with a MT10 (Interacustics, Denmark) hand-held
impedance audiometer coupled to an Interacoustics MTP10
thermal printer. A pressure range from +300 to 600 da Pa was
used. Blinded to any other information, another physician (JL)
classied the tympanograms according to the modied Jerger/
Nikolajsen classication [13,14]. Children were classied according to the tympanometric result of the best ear. In the graphic
analysis, children were placed in three groups: 1. Children with at
least one ear with a tympanometric curve of type A or C1, or at least
one functioning grommet; 2. Children with either C2/C2 or C2/B
curve congurations; and 3. Children with bilateral B curve
congurations.
2.6. Language development
The Danish version of the Reynell developmental scale of verbal
comprehension revised [15] was administered to assess receptive
language. In this test, the children have to follow verbal
instructions of rising complexity. Comprehension is scored as
the number of correct answers. Before testing, the six speech
therapists performed an adjustment study on some other children
to ensure that the raters would use the test in the same way [16].
Language scores (Reynell) were categorized into 4 groups of
approximately equal size (2550, 5155, 5660, and 6167 points)
for analysis of associations.
2.7. Parent and day care teacher evaluations
Questionnaires were constructed to examine parents and day
care teachers perception of the childs hearing and language
development and to collect information about any history of otitis
media and relevant background factors. A draft was constructed
and was discussed with experts in the eld of childrens language
and hearing and an expert in the development of questionnaires.
Pilot tests were done with 20 parents of preschool children and
2 day care teachers. Problems with wording or ambiguity were
discussed, and suggested changes were incorporated into the
version of the questionnaire presented to the test respondent.
In the WHO ICD-10 classication, disorders of speech and
language are classied as specic speech articulation disorder,

M.-B.G. Lauritsen et al. / International Journal of Pediatric Otorhinolaryngology 80 (2016) 5360

expressive language disorder and receptive language disorder.

Questions regarding language development were based on this
distinction. However, it proved difcult to separate the receptive
domains in specic questions. In the pilot tests, various wordings
of a question including the word understand were tested. This
question was persistently associated with ambiguity by the test
persons and was nally omitted. Table 1 shows the questions
regarding parents and day care teachers perceptions. The answers
were given on a 5-item Likert scale.
Familiarity with the Danish language of parents of bilingual
children was assessed. If language difculties were apparent, help
from a Danish-speaking family member or friend was arranged.
2.8. Blinding
Parents, day-care teachers, speech therapists, and the physician
were all blinded to the results that the others had registered.

55

block includes the Galker test and the Reynell test scores and the
parents and the day care teachers evaluations of childs hearing
and speech comprehension.
The partial g-coefcient (GoodmanKruskal gamma coefcient) for ordinal categorical variables was calculated as a measure
of conditional dependence [19]. The gamma coefcient is a
measure of correlation based on the difference between concordant pairs (C) and discordant pairs (D). Gamma is computed as:
gamma = (C D)/(C + D). Thus gamma is the surplus of concordant
pairs over discordant pairs, as a percentage of all pairs, ignoring
ties. Two ordinal variables are associated with the extent to which
a relatively high value on one variable can be predicted from a
relatively high value on the other variable. The gamma coefcient
variates between one and minus one; a high value means
concordant (relation) and a low value means discordant (no
relation). Gamma is equivalent to Kendalls Tau, but gamma
statistics is preferable to Kendalls Tau when the data contain many
tied observations.

2.9. Analysis
3. Results
In all analyses, total scores based on the Rasch-validated
35 items of the Galker test were used [1]. For examination of the
noise effect on the Galker test, the data of children tested twice
were analyzed using a variance analysis in which the difference
between the Galker scores on the rst and the second test was used
as response variable. Presence of noise, child age, Reynell score,
and sex were used as explaining variables.
For examination of associations with the Galker test scores, we
used a non-parametric graphical model from a subclass of loglinear statistical models for multidimensional contingency tables.
In the graphical model, any relation is analyzed as a conditional
relation. This implies that a relation between two variables is
revealed only if the association remains signicant after controlling for the effect of any preceding or intervening variables. An
advantage of graphical modeling is the possibility of simultaneous
control for all other variables [17].
In a recursive graphical model, variables have been ordered in a
structure representing temporal or causal assumptions about the
relation between the variables [18]. In this study, the variables
were ordered in three recursive blocks assuming a causal structure.
Hence, variables from the rst block are assumed to potentially
inuence variables in the second and third block, but not vice
versa. The rst block includes the following variables: the childs
sex, age, result of tympanometry, and the parents educational
level. The second block includes results of audiometry. The third
Table 1
Questions to parents and teachers included in the graphical model.
How does the/your childs pronunciation compare with that of other children
at the same age?
Much ahead of/slightly ahead of/corresponding to/slightly delayed/very delayed
How is the/your childs vocabulary compared with that of other children at
the same age?
Much larger/slightly larger/corresponding to/slightly smaller/much smaller
How is the/your child able to construct sentences compared with other
children at the same age?
Much ahead of/slightly ahead of/corresponding to/slightly delayed/very delayed
How much of a problem do you think difculty of hearing has been to
the/your child during the past 4 weeks?
No problem/minor/moderate/quite a bit of/very much of a problem
Does the child have difculty hearing messages given to children in a group?
(teachers only)
No, never/rarely/sometimes/yes, often/yes, very often
Does the child have difculty following instructions in group activities
or games? (teachers only)
No, never/rarely/sometimes/yes, often/yes, very often
Overall, how much has your child been bothered by ear-problems?
(parents only)
Not at all/a little/moderately/very much/extremely

3.1. General
The 370 children represented an unselected sample of normally
developed Danish preschool children. However, one fourth of the
invited families did not respond to the invitation to participate in
the study. The non-participants did not differ from the participants
with regard to the gender ratio, but they tended to be slightly
younger. The day care teachers general impression was that the
non-participating children did not differ from the participating
children with regard to health, development, or behavior.
However, specic data on the non-participants could not be
collected.
The inclusion criterion was 34 years of age, but 30 children
turned 5 years during the intake procedure. They were kept in the
study, and the participating childrens mean age was thus 4.1 years
(standard deviation, SD = 0.65).
Table 2 presents background information on the participating
children and their families.
In general, the children had good health according to the
parents. About 28% were rated to have been bothered by moderate
to extreme ear problems during childhood. Almost just as many
(25%) had been treated with ventilation tubes at some point.
Results of tympanometry showed a high frequency of middle ear
pathology. Thus, only 178 (48%) of the children had normal
tympanometric curves (A or C1 curves) in both ears, 82 (22%) had
middle ear effusion in at least one ear (B curve) and 102 (28%) had
at least one ear with a C2 curve or a ventilation tube inserted.
Bilateral B curves were observed in 24 (6%) (Table 3).
Peripheral hearing acuity was measured by conditional play
audiometry down to a screening level of 20 dB. Although the
children were carefully conditioned to the procedure, overall 19%
remained unable to accomplish the procedure. As expected, this
was a particular problem among 3-year-olds (42%). Among 33% of
the children who had bilateral B curves, a hearing test was missing.
Among the children with bilateral B curves who successfully
completed audiometry, the hearing threshold was 30 dB or worse
on the best ear in 38% of the children. Overall, for children with a
complete examination including audiometry, the hearing level was
25 dB or worse bilaterally in 23 (6%) (Table 3).
3.2. The effect of background noise on the Galker test results
A subgroup of 107 children was retested with the Galker test.
Sixty-seven of these children were retested with the normal
background noise (Noise group) and 40 children were tested

M.-B.G. Lauritsen et al. / International Journal of Pediatric Otorhinolaryngology 80 (2016) 5360

56

Table 2
Characteristics of participants, N = 370.

Fig. 1 shows Test 1 scores plotted against Test 2 scores. As can be

seen, there is a clear ceiling effect of Test 2 in the No Noise group
which implies that the test becomes considerably easier without
noise. Furthermore, the pattern of the test-retest results of the
Noise group seen in Fig. 1 support the reliability described in the
previous paper [1]. The small gain in the score of the Noise group is
ascribed to a learning effect. The gain in the score of the No Noise
group beyond the learning effect is ascribed to the noise effect,
which was 5 to 6 points on average.

N (%)
Age
68 (18)
3.0<3.5
3.5<4.0
80 (22)
4.0<4.5
90 (24)
4.5<5.0
102 (28)
5.0<6.0
30 (8)
Sex
Female
186 (50)
Male
184 (50)
Education
Mothers: Less than high school
57 (17)
Fathers: Less than high school
85 (23)
Siblings
One sibling
226 (61)
80 (22)
More than one sibling
Bilingual
35 (9)
Ever received speech therapy
38 (10)
Childs general health, rated by parents
Good or very good
345 (93)
Less good
15 (4)
10 (3)
Question unanswered
Child been bothered by ear problems, rated by parents
Not at all or a little
253 (68)
Moderately
62 (17)
Very much or extremely much
41 (11)
Question unanswered
14 (4)
Diagnosed with OME by doctor, ever
150 (41)
Ever treated with ventilation tubes
94 (25)

3.3. Associations with the Galker test

without noise (No noise group). Both groups performed better on

Test 2. However, the difference between Test 1 and Test 2 was
signicantly higher in the No noise group (6.9 points) than in the
Noise group (1.3 points). Variance analysis showed that this
difference was accounted for only by the noise factor (p < 0.0001)
and not age, sex, or Reynell test performance (Table 4).

Table 3
Descriptive results of examinations, N = 370.

Otoscopy
Ventilation tubes present
Tympanometry in best ear
A
C1
C2
B/B
Grommets or perforation bilateral
Missing
Audiometry, hearing in best ear
2021 dB
2225 dB
2630 dB
30 dB or more
Missing
Reynell score by age groups
3.0<3.5
3.5<4.0
4.0<4.5
4.5<5.0
5.0<6.0
Missing
Galker score by age groups
3.0<3.5
3.5<4.0
4.0<4.5
4.5<5.0
5.0<6.0

3.4. Relations between the Galker test and language-related variables

Number (%)

Mean (SD)

370

4.1 (0.65)

The graphical model (Fig. 2) shows a partial, highly signicant

relation between the Galker and the Reynell test (g = 0.35). A

52 (14)
165
92
58
24
25
6

(45)
(25)
(15)
(6)
(7)
(2)

119
120
39
31
61

(32)
(32)
(11)
(8)
(17)

40

Test 2 Galker score

Age, years

Table 5 lists the variables examined for associations with the

Galker test together with the marginal (unadjusted) relation.
Signicant relations are found between the Galker score and all
variables except parent-rated and teacher-rated hearing, patentrated bother from ear problems, and mothers education. For
comparison, ANOVA analysis was included in Table 5. As can be
seen, the results of the two types of analysis correspond very well.
The variables of Table 5 were then analyzed with the graphical
model (Fig. 2). This model produced a visual output of the
associations, characterized by independence graphs where nodes
represent variables and edges and arrows represent direct,
independent relation. The arrows also symbolize the order of
the recursive blocks.
The partial g-coefcients, which represent the independent
relations adjusted for the effect of the other variables, are placed
within Fig. 2 [19]. Only signicant relations are indicated by edges
and arrows, whereas the marginal g-coefcients (Table 5), which
represent unadjusted relation, disappear after adjustment, and
these coefcients are not included in Fig. 2.
As the partial g-coefcient is used for categorical data and also
adjusted for confounders, it is usually not as high as unadjusted
correlations calculated on the basis of continuous data. A partial gcoefcient between two variables exceeding 0.3 is considered a
very strong association [19].

Number (%)

Mean (SD)

63
77
87
101
29
13

(17)
(21)
(24)
(27)
(8)
(3)

47
51
55
57
61

(8.3)
(6.8)
(5.7)
(5.9)
(3.4)

68
80
90
102
30

(18)
(22)
(24)
(28)
(8)

17
20
24
25
26

(7.3)
(6.6)
(3.7)
(4.6)
(4.3)

30

20

10

10

20

30

Test 1 Galker score

No Noise on test 2

Noise on test 2

Fig. 1. Effect of noise and retesting on Galker scores

40

M.-B.G. Lauritsen et al. / International Journal of Pediatric Otorhinolaryngology 80 (2016) 5360

57

Table 4
Description of the effect of noise on Galker test scores.
Group

With noise
Test 2
No noise
Test 2
Noise group
compared with
No noise group
p-values
*

Girls%

67

60

40

53

Age,
years (SD)

4.1
(0.6)
4.6
(0.6)

Reynell score
Age-adjusted*
Mean (SD)

0.9
(5.8)
0.5
(5.0)
0.31

Galker score Test 1

Galker score Test 2

Galker score Delta*

Mean (SD)

Median (IQR)

Mean (SD)

Median (IQR)

Mean (SD)

Median (IQR)

22.8
(5.2)
23.9
(4.5)
0.26

24
(6)
24
(6)

24.1
(5.8)
30.8
(3.4)
<0.0001

25
(6)
32
(4)

1.3
(3.0)
6.9
(3.4)
<0.0001

1
(4)
6
(3.5)

Difference between Test 1 and Test 2.

partial relation was also seen between the Galker test and the day
care teachers ratings of the childrens vocabulary attainment
(g = 0.26). This relation was not explained by the Reynell test,
whereas the Galker test explained the associations between the
Reynell test and teacher-rated vocabulary attainment. In other
words, even when information about performance on the Reynell
test was at hand, the Galker test still contributed more information
than the day care teachers vocabulary attainment rating.
Strong, unadjusted relations (Table 5) were also seen between
the Galker test and the day care teachers answers to all the other
language-related questions, in particular their rating of the childs
ability to follow instructions during group activities. These
associations were mediated in particular through the Reynell test
(g = 0.24) (Fig. 2). The analysis indicates that the Galker test and
the Reynell test measure two different, but associated phenomena.
Associations between all language-related questions and both
the Galker test and the Reynell test were closer for the teachers
ratings than for the parents ratings. The parents report on
language achievements added no value to the analyses when
teachers rating are known (Fig. 2).
3.5. Relations between the Galker test, ear status, and hearing-related
variables
Due to the missing data on audiometry for 61 children (17%),
any interpretation of the importance of this variable must be very
cautious. As noted in the Methods section, children with missing
data on audiometry were included as a separate group for analysis.

The graphical analysis revealed that important relations were

linked to that particular group of children (Fig. 2).
Among children with a complete hearing test (N = 309), a
marginally independent association between hearing level (four
groups) and the Galker test (six groups) was found (ANOVA, F = 2.8,
p = 0.02). Age modied the association between the Galker test and
the audiometry results. Among children younger than 4 years, a
direct, adjusted association (g = 0.20) between the audiometry and
the Galker test results was found (data not shown in Fig. 2). Further
scrutiny of this relation revealed that the association was caused
predominantly by low performance on the Galker test among the
children who were unable to carry out audiometry. Low
performance in this group was also the main cause of the direct
relations between the audiometry test result and both the Reynell
test and the teachers reports on the childrens ability to follow
instructions given to the group (Fig. 2).
A signicant relation was found between the results of
tympanometry and the Galker test (Fig. 2). The graphical analysis
showed that this relation was mediated through audiometry.
Furthermore, the independent relation seen in Fig. 2 between
audiometry and tympanometry was explained both by worse
hearing among the children with bilateral B curves and by a higher
occurrence of bilateral B curves in children who were unable to
carry out the hearing test.
Examining the reports from day care teachers and parents
about the childs hearing problems and parent-reported ear
problems, we found no signicant, partial, or marginal associations
with the Galker test.

Table 5
Marginal (unadjusted) relation between variables of the graphical analysis and the Galker test groups.
Variable (number of groups)

Gamma coefcient

p-value

ANOVA F-value

p-value

Gender, girl (2)

Age (5)
Education, mother (4)
Education, father (4)
Tympanometry (5)
Audiometry (4 + 1)
Reynell scale (4)
Teacher-rated hearing (5)
Teacher-rated ability to hear messages given to groups (5)
Teacher-rated ability to follow instructions during group activities (5)
Teacher-rated vocabulary (5)
Teacher-rated sentence construction (5)
Teacher-rated pronunciation (5)
Parent-rated hearing (5)
Parent-rated bother from ear problems through childhood (5)
Parent-rated vocabulary (5)
Parent-rated sentence construction (5)
Parent-rated pronunciation (5)

0.19
0.49
0.08
0.11
0.20
0.30
0.59
0.11
0.26
0.35
0.37
0.34
0.34
0.09
0.03
0.15
0.15
0.13

0.005
<0.0001
0.116
0.034
0.011
<0.0001
<0.0001
0.13
<0.0001
<0.0001
<0.0001
<0.0001
<0.0001
0.26
0.32
0.01
0.01
0.02

6.6
27.0
2.4
2.8
2.2
12.6
57.4
2.7
5.8
11.0
14.1
11.7
9.3
1.1
1.7
6.5
4.5
2.4

0.01
<0.0001
0.07
0.04
0.06
<0.0001
<0.0001
0.048
<0.0001
<0.0001
<0.0001
<0.0001
<0.0001
0.36
0.14
<0.0001
0.001
0.05

58

M.-B.G. Lauritsen et al. / International Journal of Pediatric Otorhinolaryngology 80 (2016) 5360

Fig. 2. The graphical model of the relation between the Galker test, the Reynell test, audiometry, tympanometry, parents reports, day care teachers reports, and background
variables.
Partial (adjusted) g-coefcients are placed on the gure.

3.6. Relation between the Galker test, the childs age and gender, and
the parents educational level
A partial, highly signicant association between the Galker test
and the age of the child was demonstrated (gamma = 0.33) (Fig. 2).
A marginally, signicantly higher Galker score was seen among
girls; and a marginally, just signicant, higher score was seen
among children of fathers with a higher educational level (Table 5).
4. Discussion
4.1. The main ndings
The Galker test of speech reception in noise is a rating scale
designed to serve as a simple method for use in primary care to
quantify the impact of hearing problems often caused by middle
ear effusions in preschool children. No true gold standard exists for
this method, and its validity and characteristics have been
investigated from various angles. In a previous paper [1], the
internal construct validity was examined with the Rasch model.
Construct validity and reliability were demonstrated for 35 items
of the scale.
The present paper further explored the validity of the Galker
test by examining its associations with measures of middle ear

effusion, hearing, and language development and by examining the

effect of noise on the test results. We found that the Galker test
measured both linguistic auditory function and hearing impairment.
The study was conducted in Danish day care centers. In a
sample of normally developed Danish preschool children, we
found a high percentage of children with current OME. We found
that 16% of the children had unilateral type B tympanograms and
6% had bilateral type B tympanograms; 14% had grommets, and
25% of the children had been treated with ventilation tubes.
4.2. The effect of noise on the Galker test and relation with languagerelated variables
A relation between the Galker test and language development
was consistently found, whether reported by day care teachers or
parents or measured with the Reynell test of verbal comprehension. We found a particularly strong, independent relation
between the Galker test and the Reynell test. Some degree of
association between the two tests was expected because of the
linguistic contents of both, and because both are impressive
language tests. However, the words and pictures of the Galker test
are simple, and our results demonstrated that exclusion of noise
resulted in a strong ceiling effect which reduced the tests

M.-B.G. Lauritsen et al. / International Journal of Pediatric Otorhinolaryngology 80 (2016) 5360

discriminatory power. Therefore, it is assumed that the association

between the Galker test and the Reynell test cannot be explained
solely by the fact that they measure linguistic competence, but
rather by the combination of linguistic and auditory function. The
results are interpreted on the basis of Tallals theory which argues
that language difculties are linked to auditory processing difculties [20]. Her studies suggest that language difculties do not arise
primarily from inability to analyze linguistic components of
language, but from the inability of the auditory system to analyze
the rapid stream of acoustic information that characterizes speech.
There may be several explanations for the demonstrated
stronger association of the day care teachers than the parents
assessment of the childrens speech and hearing faculties with the
Galker test results as well as the Reynell test results. The day care
teachers all had received authorized pedagogic training; from their
work, they had experience with the observation of children and
therefore had a comprehensive frame of reference; and they
observed the children in a demanding, noisy environment which
may have brought the childrens difculties out into the open.
4.3. Relation between the Galker test and hearing, tympanometry,
and history of otitis media
No clear conclusions can be drawn about the relation between
hearing and Galker scores as the results were awed by some
missing hearing tests, and the screening audiometry was down to
20 dB only. Being unable to carry out audiometry was related to
low performance on the Galker test and was explained not only by
young age, but also by factors related to middle ear effusion
(tympanometry) and language comprehension (Reynell).
Besides problems with missing data, the screening procedure
used may not have been sufciently sensitive to detect the often
very small differences in hearing related to OME. Many children
showed some degree of middle ear pathology, and about 19%
presented bilateral thresholds of 25 dB or worse (Table 3). In
addition, the relation between pure tone audiometry and methods
of speech audiometry depends very much on the contents of the
items. The more linguistic information is incorporated together
with features like background noise and lip reading, the less
predictable the test results from the pure-tone audiogram [21
24]. Blandy & Lutman found no relation between hearing threshold
levels and speech recognition in noise in 7-year-olds [25]. In our
study, it was difcult to obtain valid hearing tests for all children.
One explanation could be the fact that the tests were done at
random locations at the different day care centers rather than
within the stable environment of an auditory clinic. Other studies
have reported unfeasible examination with audiometry in 1068%
of children aged 2 to 4 years [2628].
We observed a small, marginal effect of middle ear effusion on
the Galker results (Table 5). This effect derived from a lower score
in children with bilateral B curves, but no effect was detected in
children with C2 curves. Similar results have been reported
previously [7,11].
An effect of long lasting OME on auditory processing [29] and
speech perception has been described [3033]. A study by Gravel
et al. did not conrm this effect [34]. We found no relation between
the Galker test and the parents ratings of the degree to which the
child had been bothered by ear problems. This does not preclude a
relation as earlier treatment with grommets could have remedied
possible difculties or else the retrospective information from
parents through questionnaires may have been too inaccurate [35].
4.4. Clinical implications
The overall purpose of this project was to provide the clinician
with a more real-life-like and appealing tool for evaluation of the

59

understanding of spoken words in preschool children, in particular

children with middle ear effusion and hearing loss.
For most children, OME is transient [13], and OME has
apparently no effect on the development in otherwise healthy
children [36]. However, children who already have difculties,
language difculties in particular, are expected to be more
vulnerable to the effect of OME [37]. Early intervention in this
group is recommended, and duration of OME is no longer
recommended as a sole criterion for intervention. In healthy
children, watchful waiting seems safe as long as hearing is not or
only very mildly affected and as long as they are social functioning
and the tympanic membrane is without structural damage
[38]. However, in this situation, the clinician is faced with the
challenge of identifying the child at risk of the negative effect of
OME and to undertake watchful monitoring of the child until the
hearing and the ear function eventually return to normal
[39]. Children with OME are now extensively managed and treated
with ventilation tubes on the basis of an assumed reduced access to
real life speech without these tubes. However, this approach
involves a risk of both over-treatment and under-treatment
[4]. The Galker test appears to be a promising, supplementary
tool to which resort may be made to inform the decision of how to
manage these children. However, more trials are needed to fully
document the usefulness of the Galker test in this respect.
4.5. Further research
In the present project, difculty of speech reception in noise
was assumed to indicate an increased risk of long-term developmental difculties. However, the study was cross-sectional, and
only follow-up examination of the children can clarify whether this
assumption holds true. Long-term follow-up would also help
determine which score levels of the Galker test indicate clinically
important intervention thresholds. Investigation of test results in
clinical subgroups with known difculties in general and hearing
loss and documented long-term OME in particular is needed as is
examination of the sensitivity of the Galker test to detect changes
in middle ear status and treatment effects.

5. Conclusion
The present paper demonstrates the signicant inuence
background noise has on the Galker test of speech reception in
noise. A strong, signicant relation was found between the Galker
test and language comprehension evaluated by speech therapists,
and a signicant association was found between the Galker test
and language development evaluated by teachers in day care
centers. A weak, but signicant association was found between the
Galker test and middle ear status (bilateral type B tympanometry)
and audiometry. The relation with peripheral hearing was awed
by 17% missing tests.
The Galker test of speech reception in noise appears promising
as a tool in primary care for evaluating the ability of the preschool
child to hear and understand speech. Further research of its
validity is needed, in particular long-term follow-up studies to
determine the predictive value of the test and to further examine
clinical subgroups with known difculties.
Acknowledgements
We wish to thank the participating families and the staff from
the day-care centers; speech and hearing therapist Eve Galker for
her year-long participation in developing the Galker test, and for
her pilot-testing of the Galker test in day-care centers around
Aarhus. Willy Karlslund is acknowledged for technical assistance

60

M.-B.G. Lauritsen et al. / International Journal of Pediatric Otorhinolaryngology 80 (2016) 5360

and Anne Birgitte Srensen, Eigil Laulund, Lotte Krause Slot,

Susanne Rosbk and Lisbeth Anse Petersen for performing
language assessments of the children.
The study received funding from: The Fund for Research in
General Practice [Fonden vedrrende nansiering af forskning I
almen praksis og sundhedsvsenet i vrigtForskningsfonden],
The Health Foundation [Sygekassernes Helsefond], The Hearing
Society [Landsforening for bedre hrelse], The Research Fund in
Frederiksborg County [Forskningspuljen i Frederiksborg Amt], The
Foundation for Medical Research [Fonden for lgevidenskabelig
fremme], The Education fund for General Practice [De praktiserende lgers uddannelsesfond], Magda og Svend Aage Friederichs
Memory Grand [Magda og Svend Aage Friederichs mindelegat].
References
[1] M.G. Lauritsen, S. Kreiner, M. Soderstrom, J. Drup, J. Lous, Validity, reliability, and
acceptance of the Galker test of speech reception in noise for Danish children, Int.
J. Pediatr. Otorhinolaryngol. 79 (2015) 16941701, PMID: 236260661.
[2] J. Roberts, L. Hunter, J. Gravel, R. Rosenfeld, S. Berman, M. Haggard, et al., Otitis
media, hearing loss, and language learing: controversies and current research, J.
Dev. Behav. Pediatr. 25 (2004) 110122, PMID: 15083134.
[3] G.G. Browning, M.M. Rovers, I. Williamson, J. Lous, M.J. Burton, GrommetsGrommets (ventilation tubes) for hearing loss associated with otitis media with
effusion in children, Cochrane Database Syst. Rev. 10 (2010), CD001801.
[4] J. Lous, Which children would benet most from tympanostomy tubes (Grommetsw)? A personal evidence-based review, Int. J. Pediatr. Otorhinolaryngol. 72
(2008) 731736, PMID: 18336921..
[5] L. Vernon-Feagans, A. Miccio, YontF K., Speech, language, pragmatitics and
attention, in: R.M. Rosenfeld, C.D. Bluestone (Eds.), Evidence-Based Otitis Media,
second ed., BC Decker Inc, Hamilton, ON, 2003, pp. 360382.
[6] M.L. Bistrup, Prevention of adverse effects of noise on children, Noise Health 5
(2003) 5964, PMID: 12804213..
[7] D.A. Adams, J.A. Birkin, R.J. Flanagan, Disability for speech-in-noise in children
with otitis media with effusion, in: M. Tos, J. Thomsen, V. Balle (Eds.), Otitis Media
Today. Proceedings of the Third Extraordinary Symposium on Recent Advances in
Otitis Media, Kluger Publications, The Hague, 1999, pp. 669676.
[8] A. Zumach, E. Gerrits, M.N. Chenault, L.J. Anteunis, Otitis media and speech-innoise recognition in school-aged children, Audiol. Neurootol. 14 (2009) 121129,
PMID: 18852485.
[9] MRC Multicentre Otitis Media Study Group, Speech reception in noise: an
indicator of benet from otitis media with effusion surgery, Clin. Otolaryngol.
Allied Sci. 29 (2004) 497504, PMID: 15373863.
[10] I. Williamson, C. Sheridan, The development of a test of speech reception
disability for use in 5- to 8-year-old children with otitis media with effusion,
Eur. J. Disord. Commun. 29 (1994) 2737, PMID: 8032104.
[11] I. Williamson, C. Sheridan, E. Galker, J. Lous, A video-based performance in noise
test for measuring audio-visual disability in young school children: test development, with validation by trained teachers, parents and audiometry as relative
standards for disability, Int. Pediatr J. Otorhinolaryngol. 49 (1999) 127133,
PMID: 10504019.
[12] M.P. Northern, J.L. Downs, Behavioral hearing testing, in: J.L. Northern, M.P.
Downs (Eds.), Hearing in Children, fth ed., Lippincott Williams & Wilkins,
Philadelphia, 2002.
[13] M. Fiellau-Nikolajsen, J. Lous, Prospective tympanometry in 3-year-old children.
A study of the spontaneous course of tympanometry types in a non-selected
population, Arch. Otolaryngol. 105 (1979) 461466, PMID: 464883.
[14] J. Lous, Use of tympanometry in general practice in Denmark, Int. J. Pediatr.
Otorhinolaryngol. 78 (2014) 124127, PMID: 24315212.
[15] D. Skovlund, Reynell Developmental Language Scales, Danish version, Dansk
Psykologisk Forlag, Herning, 1983.
[16] M.G. Lauritsen, Identication of the preschool child at risk of developmental
effects from otitis media with effusion, in: Evaluation and Further Development of
a Danish Speech-in-Noise Test. Research Unit for General Practice in Copenhagen,

[17]

[18]

[19]
[20]

[21]
[22]

[23]

[24]

[25]
[26]

[27]

[28]

[29]

[30]
[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

University of Copenhagen, Maanedsskrift for Praktisk Lgegerning, Copenhagen,

2006 (Ph.D thesis).
S. Kreiner, An informal introduction to graphical modelling, in: H.C. Knudsen, G.
Thornicroft (Eds.), Mental Health Service Evaluation, Cambridge University Press,
Cambridge, 1996.
J.P. Klein, N. Keiding, S. Kreiner, Graphical models for panel studies, illustrated on
data from the Framingham Heart Study, Stat. Med. 14 (1995) 12651290, PMID:
7569487.
L.A. Goodman, W.H. Kruskal, Measures of associations for cross classications,
J. Am. Stat. Assoc. 49 (1954) 732764.
P. Tallal, Perceptual and linguistic factors in the language impairment of developmental dysphasics: an experimental investigation with the token test, Cortex
11 (1975) 196205, PMID: 1204360.
D.B. Fry, Word and sentence tests for use in speech audiometry, Lancet 2 (7195)
(1961) 197199, PMID: 13702318.
M. Nilsson, S.D. Soli, J.A. Sullivan, Development of the hearing in noise test for
measurement of speech reception thresholds in quit and in noise, J. Acoust. Soc.
Am. 95 (1994) 10851099, PMID: 8132902.
M.J. Middelweerd, R. Plomp, The effect of speechreading on the speech-reception
threshold of sentences in noise, J. Acoust. Soc. Am. 82 (1987) 21452147, PMID:
3429736.
M.P. Haggard, E.J. Wood, S. Carroll, Speech admittance and tone tests in school
screening. Reconciling economics with pathology and disability perspectives, Br.
J. Audiol. 18 (1984) 133153, PMID: 6487848.
S. Blandy, M. Lutman, Hearing threshold levels and speech recognition in noise in
7-year-olds, Int. J. Audiol. 44 (2005) 435443, PMID: 16149238.
MRC Multi-centre Otitis Media Study Group, Inuence of age, type of audiometry
and childs concentration on hearing thresholds, Br. J. Audiol. 34 (2000) 231240,
PMID: 10997452.
R.L. Allen, A. Stuart, D. Everett, S. Elangovan, Preschool hearing screening: pass/
refer rates for children enrolled in a head start program in eastern North Carolina,
Am. J. Audiol. 13 (2004) 2938, PMID: 15248802.
M.E. Roberts, Comparative study of pure-tone, impedance, and otoscopic hearing
screening methods. A survey of narrative Indian children in British Columbia,
Arch. Otolaryngol. 102 (1976) 690694, PMID: 985203.
J.W. Hall 3rd, J.H. Grose, H.C. Pillsbury, Long-term effects of chronic otitis media
on binaural hearing in children, Arch. Otolaryngol. Head Neck Surg. 121 (1995)
847852, PMID: 7619408.
S. Jerger, J. Jerger, B.R. Alford, S. Abrams, Development of speech intelligibility in
children with recurrent otitis media, Ear Hear. 4 (1983) 138145, PMID: 6862115.
R.L. Clarkson, P.D. Eimas, G.C. Marean, Speech perception in children with histories of recurrent otitis media, J. Acoust. Soc. Am. 85 (1989) 926933, PMID:
2926008.
S. Nittrouer, L.T. Burton, The role of early language experience in the development
of speech perception and phonological processing abilities: evidence from 5-yearolds with histories of otitis media with effusion and low socioeconomic status, J.
Commun. Disord. 38 (2005) 2963, PMID: 15475013.
S. Majerus, P. Amand, V. Boniver, J.P. Demanez, L. Demanez, M. Van der Linden, A
quantitative and qualitative assessment of verbal short-term memory and phonological processing in 8-year-olds with a history of repetitive otitis media, J.
Commun. Disord. 38 (2005) 473498, PMID: 15950984.
J.S. Gravel, J.E. Roberts, J. Roush, J. Grose, J. Besing, M. Burchinal, et al., Early otitis
media with effusion, hearing loss, and auditory processes at school age, Ear Hear.
27 (2006) 353368.
K.A. Daly, B. Lindgren, G.S. Giebink, Validity of parental report of a childs medical
history in otitis media research, Am. J. Epidemiol. 139 (1994) 11161121, PMID:
8192144.
J.E. Roberts, R.M. Rosenfeld, S.A. Zeisel, Otitis media and speech and language: a
meta-analysis of prospective studies, Pediatrics 113 (2004) e238e248, PMID:
14993583.
R.J. Ruben, Host susceptibility to sequelae, in: R.M. Rosenfeld, C.D. Bluestone
(Eds.), Evidence-Based Otitis Media, second ed., BC Decker Inc, Hamilton, ON,
2003, pp. 505514.
R.M. Rosenfeld, L0. Culpepper, K.J. Doyle, K.M. Grundfast, A. Hoberman, M.A.
Kenna, et al., Clinical practice guideline: otitis media with effusion, Otolaryngol.
Head Neck Surg. 130 (2004) S95S118, PMID: 15138413.
W. Lok, M.N. Chenault, L.J.C. Anteunis, C. Meesters, M.P. Haggard, Selecting infants
with OM that need referral and further assessment: creating a case-nding instrument, Int. J. Pediatr. Otorhinolaryngol. 75 (2011) 13011307, PMID: 21840067.