PDF hosted at the Radboud Repository of the Radboud University
Nijmegen
The following full text is a publisher's version.
For additional information about this publication click this link.
http://hdl.handle.net/2066/43784
Please be advised that this information was generated on 2022-05-30 and may be subject to
change.
An acoustic description of the vowels of northern and southern
standard Dutch II: Regional varieties
a)
Patti Adank
b兲
F.C. Donders Centre for Cognitive Neuroimaging, Kapittelweg 29, 6525 EN Nijmegen, The Netherlands
Roeland van Hout
Centre for Language Studies, Radboud University Nijmegen, Erasmuslaan 1, PO Box 9103, 6500 HD,
Nijmegen, The Netherlands
Hans van de Velde
Utrecht Institute of Linguistics OTS, University of Utrecht, Trans 10 3512 JK Utrecht, The Netherlands
共Received 3 May 2006; revised 20 November 2006; accepted 20 November 2006兲
An analysis is presented of regional variation patterns in the vowel system of Standard Dutch as
spoken in the Netherlands 共Northern Standard Dutch兲 and Flanders 共Southern Standard Dutch兲. The
speech material consisted of read monosyllabic utterances in a neutral consonantal context 共i.e.,
/sVs/兲. The analyses were based on measurements of the duration and the frequencies of the first two
formants of the vowel tokens. Recordings were made for 80 Dutch and 80 Flemish speakers, who
were stratified for the social factors gender and region. These 160 speakers were distributed across
four regions in the Netherlands and four regions in Flanders. Differences between regional varieties
were found for duration, steady-state formant frequencies, and spectral change of formant
frequencies. Variation patterns in the spectral characteristics of the long mid vowels /e o ø/ and the
diphthongal vowels / iœyÅu/ were in accordance with a recent theory of pronunciation change in
Standard Dutch. Finally, it was found that regional information was present in the steady-state
formant frequency measurements of vowels produced by professional language users. © 2007
Acoustical Society of America. 关DOI: 10.1121/1.2409492兴
PACS number共s兲: 43.70.Fq, 43.70.Kv, 43.72.Ar 关AL兴 Pages: 1130–1141
I. INTRODUCTION
This paper describes the vowel system of Standard
Dutch as spoken in the Netherlands and Flanders and exam-
ines regional patterns of variation in the acoustic character-
istics of the 15 vowels of Dutch /Ä e ( i Å u + yeoøÅu i
œy/. The reasons for describing these regional variation pat-
terns were twofold.
First, in recent years it has become generally accepted
that a language’s vowel system is better characterized when
its description includes regional varieties than when it in-
cludes only a single idealized set of acoustic-phonetic char-
acteristics 共Clopper et al., 2005; Hagiwara, 1997兲. Earlier
studies on the vowel system of Standard Dutch 共Adank, van
Hout, and Smits, 2004; Pols et al., 1973; Van Nierop et al.,
1973兲 are therefore limited in that they do not include re-
gional varieties. Pols et al. describe the acoustic characteris-
tics of vowel tokens produced by 50 male speakers from the
Netherlands, who spoke Standard Dutch, while Van Nierop
et al. provide a description of vowel tokens produced by 25
female Standard Dutch speakers from the Netherlands.
Adank et al. describe the acoustic characteristics 共duration,
f
0
, and formant frequencies F
1
through F
3
兲 of realizations of
the vowels of Standard Dutch for ten male and ten female
speakers from the Netherlands and ten male and ten female
speakers from Flanders. Although Adank et al.’s description
is an improvement over Pols et al.’s and Van Nierop et al.’s
in the sense that speakers from Flanders are included as well,
it is limited because it excludes regional varieties.
Second, it is at present not feasible to establish neither
how the pronunciation of the vowels of Dutch varies across
the Dutch language area, nor how this pronunciation of these
vowels evolves over time, as no previous acoustic descrip-
tions are available. This paper attempts to fill this gap by
providing a comprehensive overview of the extent to which
Dutch vowels vary in their acoustic characteristics across
regional varieties in the Netherlands and Flanders. In doing
so, this overview could serve as a point of reference for
further studies on the vowel system of Standard Dutch.
The present study builds on Adank et al., who describe
recordings of 40 professional users of Standard Dutch 共i.e.,
teachers of the Dutch language兲. These recordings were
made using a sociolinguistic interview in which vowels and
consonants were elicited through a wide variety of tasks.
Adank et al.’s vowel tokens were recorded through a formal
reading task, i.e., reading nonsense words in carrier sen-
tences from a computer screen. Of the 40 speakers, 20 were
from the socioeconomic core area 共the culturally and eco-
nomically dominant region兲 in the Netherlands and 20 were
from Flanders’ socioeconomic core area.
a兲
Portions of this work were presented as “Distinguishing Regional Varieties
of Dutch” at The International Conference on Language Variation in Eu-
rope 共ICLaVE3兲 2005 and as “Regional Variation Patterns in the Vowel
System of Standard Dutch” at the Workshop on Accent, Variation and
Change on March 3, 2006 at the UCL Centre for Human Communication.
b兲
Author to whom correspondence should be addressed; electronic mail:
patti.adank@fcdonders.ru.nl
1130 J. Acoust. Soc. Am. 121 共2兲, February 2007 © 2007 Acoustical Society of America0001-4966/2007/121共2兲/1130/12/$23.00
Downloaded 10 Oct 2012 to 131.174.248.42. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
Originally, 160 speakers were recorded through Adank
et al.’s sociolinguistic interview. The present study describes
the remaining 120 speakers, who were distributed across six
regional varieties of Standard Dutch, with ten female and ten
male speakers per variety. In the description it is first estab-
lished whether the variation patterns reported for the two
standard varieties described in Adank et al. are representative
for regional varieties of Standard Dutch and, second, the
variation patterns across the six varieties are identified.
II. MATERIALS
A. Database design, recordings, and acoustic
measurements
1. Speech communities, resigns, and towns
In each country, also referred to as “speech community,”
four regions were distinguished, a central region, and three
noncentral regions: an intermediate region and two periph-
eral regions. Twenty speakers were recorded per region. The
central region in each speech community was the socioeco-
nomic core area of that community. It was thought that the
speech from the speakers in the central region would reflect
the most prestigious variety of Northern Standard Dutch, or
NSD, 共spoken in the Netherlands兲 and Southern Standard
Dutch, or SSD, 共spoken in Flanders兲 in both communities. In
Adank et al., the vowel tokens of the two central regions are
described.
Regional pronunciation variation in Standard Dutch is
directly influenced by the dialects of the regions in question.
The more the regional dialects differ from Standard Dutch,
the stronger the accent of that region is present when speak-
ing the standard language 共e.g., Chambers, 2003; Labov,
1972兲. Covering the range of pronunciation variation implies
selecting peripheral areas. In both speech communities two
peripheral regions were selected that were maximally distant
geographically from each other and from the core area. The
intermediate region was a region geographically next to the
central region. The regional dialects of the intermediate re-
gion are closer to the standard language.
For NSD, the central region was the west, consisting of
the provinces Northern-Holland, Southern-Holland, and
Utrecht, also known as “the Randstad” and referred to as
“N-R” 共Netherlands-Randstad兲. The cities Amsterdam, Rot-
terdam, Utrecht, and The Hague are part of the Randstad.
The intermediate region for NSD enclosed the southern part
of the province Gelderland and part of the province Utrecht.
This region is referred to as “N-M” 共Netherlands-Middle兲.
The two peripheral regions for NSD were the province Lim-
burg, or “N-S” 共Netherlands-South兲, in the south of the Neth-
erlands, and the province Groningen, or “N-N” 共Netherlands-
North兲, in the north of the Netherlands.
In SSD, the central region was “Brabant,” denoted as
“F-B” 共Flanders-Brabant兲. Brabant enclosed the provinces
Antwerpen and Flemish-Brabant, with the cities of Antwer-
pen and Leuven, respectively. The intermediate region was
the province East-Flanders, referred to as “F-E” 共Flanders-
East兲. The two peripheral regions for SSD were the prov-
inces 共Flemish兲 Limburg, or “F-L” 共Flanders-Limburg兲, and
West-Flanders, or “F-W” 共Flanders-West兲.
Several towns were selected per region, following three
criteria. First, the selected towns in each region had a com-
parable socioeconomic profile. Second, they belonged to the
same dialect group. Third, the Dutch spoken in the towns
was regarded as characteristic of that region. No major cities
were selected, because it was expected that the Dutch spoken
in major cities is influenced by dialects 共or languages兲 other
than those spoken in the surrounding region, due to migra-
tion. Table I lists the selected towns per region and Fig. 1
shows the location of these towns in the Netherlands and
Flanders.
2. Speakers
All 160 speakers were Dutch teachers at secondary edu-
cation institutes at the time the interview was recorded.
Dutch teachers were selected because they are professional
language users who are expected to speak standard Dutch on
a daily basis. Furthermore, they are instructors of the stan-
dard language and may therefore be regarded as having a
normative role. A final reason for selecting Dutch teachers
was that it was assumed that their speech would show more
regional variation than broadcasters’ 共whose speech is gener-
ally used in pronunciation studies of the standard language,
cf. Bell, 1983兲.
The teachers who participated in the interview taught at
schools for secondary education in the selected towns. They
had to meet the following requirements. First, at the time of
the interview, they all lived in one of the selected towns, or
near that town in the dialectal region characteristic for that
region. Second, they were born in the region or moved there
before their eighth birthday. Third, they had lived in the
region for at least eight years prior to their 18th birthday.
Finally, the speakers were divided into two age groups, a
younger group and an older group. The speakers in the
TABLE I. The selected towns per speech community, for each of the eight selected regions.
Speech community Region Name Selected towns
Netherlands Central N-R Randstad Alphen aan de Rijn, Gouda
Intermediate N-M South-Gelderland Tiel, Veenendaal, Ede, Culemborg, Elst
Peripheral 1 N-S Limburg Sittard, Geleen, Roermond
Peripheral 2 N-N Groningen Assen, Veendam, Winschoten
Flanders Central F-B Brabant Lier, Heist-op-den berg
Intermediate F-E East-Flanders Oudenaarde, Zottegem, Ronse, Brakel
Peripheral 1 F-L Limburg Tongeren, Bilzen
Peripheral 2 F-W West-Flanders Ieper, Poperinge
J. Acoust. Soc. Am., Vol. 121, No. 2, February 2007 Adank et al.: Dutch vowels: regional variation patterns 1131
Downloaded 10 Oct 2012 to 131.174.248.42. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
younger group were between 22 and 44 years old at the time
of the interview and speakers in the older group were be-
tween 45 and 50 years old. Each region in Table I was thus
represented by 20 speakers: five young men, five older men,
five younger women, and five older women. Note that speak-
ing a regional dialect or not was not a criterion for selection.
It was assumed that growing up in a specific region implies
that regional features of the standard language play a role in
the acquisition and socialization process.
3. Carrier sentences
Dutch vowels have traditionally been divided into pho-
nologically short vowels, /Ä ( Å +/, phonologically long
vowels /Ä e ( ø o u y/, and diphthongs, /( Åu œy/ 共Booij,
1995兲. All target vowels were produced in a carrier sentence.
The sentences had the following generic structure for the
short vowels 共“V” indicates the target vowel兲:
In sVs en in sVsse zit de V
/ ( n sVs ɘ n ( n sVs ɘ z ( tdɘ V/
关In sVs and in sVsse is the V兴
The sentences had the following structure for the long
vowels and the diphthongs:
In sVs en in sVze zit de V
/ ( n sVs ɘ n ( n sVz ɘ z ( tdɘ V/
关In sVs and in sVze is the V兴
Of the three different consonantal contexts 共CVC,
CVCV, or V兲, the CVC contexts were selected for further
processing. The CVC-structure /sVs/ can be regarded as a
neutral context for Dutch vowels.
4. Recording procedure
The vowels were elicited through the sentences that
were presented to the speaker on a computer screen, with a
3 s interval between sentences. When the speaker made a
mistake, the interviewer interrupted the computer program
and went back at least two sentences and asked the speaker
to repeat these sentences. This task was performed twice. A
total of 4800 vowel tokens were thus recorded: two tokens of
each of the 15 vowel categories of Dutch, produced by 160
speakers.
5. Acoustic measurements: Duration, F
1
and F
2
The start and end times for the duration of each token
were labeled manually in the digitized speech wave. Labels
were placed at zero crossings at the onset and offset of the
glottal vibrations of the vocalic portion of the /sVs/ syllable.
When labeling it was ensured that the surrounding speech
sounds were not audible in the remaining signal. The dura-
tion of each vowel segment was defined as the interval be-
tween the segment labels at the start and end of the vocalic
portion.
The frequencies of F
1
and F
2
were stored at nine points
of the vowel token’s duration, with the first point at the start
of the vocalic portion and the ninth point at the end of the
vocalic portion, and the remaining points spaced at equal-
sized intervals, relative to the absolute duration. The nine
monophthongal vowels /Ä a ( i Å u + y/ were represented at
one time point only, i.e., at 50%—the fifth of the nine time
points—as Adank et al. 共2004兲 report that these vowels can
be separated fairly well based on their steady-state character-
istics for their first two formants only. The diphthongal vow-
els /iœyÅu/ and the long mid vowels /e o ø/ were repre-
sented at two time points, i.e., 25% and 75%, or the third and
seventh time point, as Adank et al. report that these vowels
cannot be adequately separated unless information about
their dynamic characteristics is supplied. They suggest that
FIG. 1. Map of the Netherlands and Flanders, showing
all selected towns,‘兲’ shows the location of each town.
1132 J. Acoust. Soc. Am., Vol. 121, No. 2, February 2007 Adank et al.: Dutch vowels: regional variation patterns
Downloaded 10 Oct 2012 to 131.174.248.42. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
the three long mid vowels for Dutch should not be treated as
monophthongal vowels, but instead as semidiphthongal vow-
els, when describing Dutch vowels acoustically, especially
for NSD. The monophthongal vowels, semidiphthongal vow-
els, and full diphthongal vowels were analyzed separately.
Finally, Adank et al. provided a description of the mea-
surements of the fundamental frequency for the two central
regions. However, as they found no differences between
these two regions, it was decided to exclude the analysis of
the fundamental frequency in the present paper. For further
specifics of the acoustic measurements, see Adank, van
Hout, and Smits 共2004兲.
III. RESULTS
A. Duration
1. Duration variation within speech communities
Figure 2 shows the average duration measurements for
all vowels across the four regions in the both speech com-
munities and Figure 3 shows the average durations per
region for both genders pooled across both speech commu-
nities.
A repeated-measures analysis of covariance 共ANOVA兲
was run on the duration measurements for each vowel token,
with vowel category as the within-subject factor and with the
speaker’s regional background 共region兲 and gender as
between-subjects factors. The analysis was carried out per
speech community.
The analysis for NSD showed a significant main effect
of the within-subjects factor vowel 共F关6.228,104兴 = 1253.32,
p⬍ 0.05, Huynh-Feldt corrected兲. Furthermore, effects were
found for between-subjects factors region 共F关3,152兴= 8.91,
p⬍ 0.05兲 and gender 共F关1,152兴 =8.45, p ⬍ 0.05兲, whereas the
region ⫻ gender interaction was not significant. This sug-
gests that the duration of some vowels varied across the four
NSD regions. Second, the effect for gender indicates that the
female speakers produced longer vowels than male speakers,
as can be observed in Fig. 3. A post-hoc analysis was carried
out on region to further investigate the differences between
NSD regions. The p value was set to 0.001 to correct for the
large number of analyses. The results showed that the vowels
of the central region N-R were overall shorter than for N-M
and N-N 共cf. Fig. 2兲. The results for SSD revealed an effect
of the within-subjects factor vowel 共F关5.118,104兴 = 1256.81,
p⬍ 0.05, Huynh-Feldt corrected兲 and a significant main ef-
fect of the between-subjects factor gender 共F关1,152兴= 20.45,
p⬍ 0.05兲, while region and the region ⫻ gender interaction
were not significant. Again, the female speakers showed
longer durations than the male speakers 共cf. Fig. 3兲.
2. Duration variation between speech communities
Figure 4 shows the average duration per speech commu-
nity, pooled across the four regions in each community. To
establish which vowels varied in their duration measure-
ments across both communities, a univariate ANOVA was
carried out for each vowel separately. The duration measure-
ments per vowel token served as the dependent variable, and
community served as the independent variable. Two univari-
ate ANOVAs were run: one for the two central regions and
one with the three noncentral regions nested under commu-
nity, for NSD and SSD separately. Because of the large num-
ber of analyses, p was set to 0.001.
The analysis for the two central regions indicated that
the durations for /œy/ 共F关1,78兴 =14.46, p艋0.001兲 /y/
共F关1,78兴= 30.52, p艋 0.001兲 and /i/ 共F关1,78兴 =16.62, p
艋0.001兲 were different for both communities. The analysis
for the six noncentral regions showed that the duration of /y/
共F关1,78兴= 84.84, p 艋 0.001兲 was shorter for NSD and that
the durations of /Ä/ 共F关1,78兴 =30.76, p艋 0.001兲,//
共F关1,78兴= 30.55, p艋 0.001兲,/Å/ 共F关1,78兴= 65.75,
p艋 0.001兲,/+/ 共F关1,78兴= 50.76, p
艋0.001兲, and /(/
共F关1,78兴= 44.17, p艋0.001兲 were all longer for NSD. There-
FIG. 2. Error bars 共bars represent one standard deviation兲 of average dura-
tion in ms per vowel, for the 80 NSD speakers 共top panel兲 and the SSD
speakers 共bottom panel兲. 共“ou”⫽/Åu/, “ui”⫽/y/, “eu”⫽/ø/, “A”⫽/Ä/, “O”⫽/
Å/, “Y”⫽/+/, “I”⫽/(/; o⫽central region, 䊐⫽intermediate region,
x⫽peripheral II, and ⌬⫽peripheral II兲.
J. Acoust. Soc. Am., Vol. 121, No. 2, February 2007 Adank et al.: Dutch vowels: regional variation patterns 1133
Downloaded 10 Oct 2012 to 131.174.248.42. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
