Content uploaded by Cesko Voeten
Author content
All content in this area was uploaded by Cesko Voeten on Aug 15, 2021
Content may be subject to copyright.
578
The adoption of sound change
Cesko Voeten
Cesko Voeten
The adoption
of sound change
Synchronic and diachronic processing
of regional variation in Dutch
This dissertation investigates how sound change is adopted by speakers and
listeners, based on a currently-ongoing cluster of changes in Dutch termed the
‘Polder shift’. The main aim of the dissertation is to form a bridge between five
key areas of linguistics: historical phonology, sociophonetics, psycholinguistics,
neurolinguistics, and quantitative linguistics. A unified account of these dierent
angles to the study of sound change is not trivial. This dissertation uses
psycholinguistic experiments combined with detailed quantitative analysis to
study the contributions of the dierent components to the adoption of sound
change in the medium and long term.
The population studied in this dissertation is sociolinguistic migrants: in this
case, Flemish speakers of Dutch who have migrated to the Netherlands, and
thereby migrated from a non-Polder-shift area to a Polder-shift area. The
methods adopted in this dissertation include a corpus study of regional variation,
longitudinal psycholinguistic experiments over nine months’ time, cross-sectional
psycholinguistic experiments spanning multiple decades of apparent time, and two
neurolinguistic studies using EEG. Results show that the sociolinguistic migrants
rapidly acquire allophonic variation at the phonological level (albeit not necessarily
the associated sociolinguistic knowledge), but that it takes a long time (more than
nine months, up to multiple decades) for this to carry forward to their behavioral
production and perception, and moreover is subject to significant individual
dierences. The contributions by this dissertation show how the fundamentally
sociolinguistic phenomenon of sound change can be studied empirically using
psycho- and neurolinguistics, and profit from recent innovations in statistics.
ISBN 978-94-6093-364-6
The adoption of sound change
Synchronic and diachronic processing of regional
variation in Dutch
1 Introduction 1
2 Regional variation in on-going sound change: the case of the Dutch
diphthongs 25
:::
EO
3 How long is “a long term” for sound change? The eect of duration
of immersion on the adoption of on-going sound change 53
4 Individual dierences in the adoption of sound change 91
5 Noticing the change: misrepresentation, not misperception, of allo-
phonic variants in sound change 131
6 ERP responses to regional accent reect two distinct processes of per-
ceptual compensation 151
7 Conclusion 181
A Prime–target list for Experiment 1 from Chapter 3 193
B Word list for Experiment 2 from Chapter 3 197
C Word list for Experiment 1 from Chapter 4 201
D Full BLUPs for Experiment 1 from Chapter 4 205
E Prime–target list for Experiment 2 from Chapter 4 207
陈轶亚
史濛辉 杨青
刘敏 毕一飞 王曼 李倩
邹婷
t
Γν+
√νπ σΓν
+(x−µ)
νσ −
ν+
Acknowledgments
permutes
胡瀚 郑婷婷
程航 吴疆
actuation problem
constraints problem
embedding problem
evaluation problem
transition problem
linguistic structure
l
ʁ
l→ʁ
l>*ɫ >w>gw>*ɣw>ʁ
Chapter 1
linguistic structure
social
structure
individual mem-
bers
adoption problem
Introduction
1.1 The Polder shift: an on-going vowel shift in
Dutch
eːøːoː
eiøyou
eː eːj
ɛiœyɔu
ɛi
ɛi
eːøːoː
ɛiœyɔu
1.1. The Polder shift: an on-going vowel shift in Dutch
6y
ouøyei
Au
•
A•
E•œyEi
•
ai
•
a: •
o:ø:e:
Ou
•
O•
Y
•
I•
uyi
Figure 1.1: Vowel diagram showing the changes constituting the on-going vowel
shift. The arrows indicate the diachronic changes; secondary arrows to
indicate upgliding diphthongization are not included to prevent clut-
tering the diagram.
eːøːoː
eːøːoː
ɛiœyɔu
ɛːœːɑː l
eː→eːj l
Introduction
ʋj
phonological
synchronic diachronic
1.2. Theories about the lifecycle of sound change
1.2 Theories about the lifecycle of sound change
1.2.1 Misperception as a source of sound change
misperception
realized
heard perceived
᷇
᷇
produced
not
Introduction
not
1.2. Theories about the lifecycle of sound change
1.2.2 Speaker-induced sound change and the role of the
representation
bidirectional
Introduction
prototype selection
and
1.2.3 Exemplar Theory
1.2. Theories about the lifecycle of sound change
reductive
td
Introduction
1.2.4 Which comes rst: perception or production?
u ʉ
u+t u
t u
ʉ
ʉt
u ʉ
phonemes f v phonemes p
b
1.2. Theories about the lifecycle of sound change
i ɪ
1.2.5 Types of change
implementation
Introduction
are
norm
r
ever
1.2.6 Summary
1.2. Theories about the lifecycle of sound change
produce
transmit
can
can
Introduction
1.3 The psycholinguistics of variation in perception
1.3.1 Perceptual learning as the antagonist of misperception
compensating
used
f
s ?
?
f f
s
s s
?
? s
1.3. The psycholinguistics of variation in perception
f∼s
?
f sr
? s fn
f f s
s f∼s
Introduction
ɛç
ɛx
1.3.2 Methodological innovations for psycholinguists
1.3. The psycholinguistics of variation in perception
lme4
lme4 rePCA
θ rePCA
θ
lme4
glmmTMB
Introduction
most important
buildmer
feasible
l
ɫ
eːɫ
a priori
l l
ɫ
mgcv
ɫ
ɫ
1.3. The psycholinguistics of variation in perception
trees
glmertree
p <
y
β data
p(y|β=)< .
parameter p(β=ˆ
β|y)
Introduction
aposteriori
p <
p > any
ei eː
ei→eː
eː→ei
a priori
p in favor of
is
a priori
1.4. This dissertation
permutes
1.4 This dissertation
what factors
inuence the adoption of sound change?
not
can
what is the synchronic diatopic diusion of the
sound changes involved in the Polder shift?
sociolinguistic migrants
Introduction
(how) do sociolinguistic migrants adopt the Polder
shift? which individuals, after how much time, are
more likely to adopt the Polder shift?
longitudinal group
cross-sectional
individual
(how) is the adoption of
the Polder shift reected in ERPs?
This chapter has been submitted.
Abstract
eːøːoːɛiœy l
l
l
2.1. Introduction
2.1 Introduction
nature
eːøːoː
eiøyou
ɛiœyɔu aiɒyɑu
l
l
l
l
l
Regional variation in on-going sound change: the case of the Dutch diphthongs
Table 2.1: Modes of implementation of historical sound changes, after Bermúdez-
Otero (2007).
Lexical dimension
Phonetic dimension Abrupt Gradual
Abrupt Change in underlying forms Lexical diffusion
(Janson 1983) (Wang 1969)
Gradual Neogrammarian change Change by exemplars
(Osthoff & Brugmann 1878) (Bybee 2002)
l
that
how
l
l
2.2. Method
l
l
l
l
l
ɫ
2.2 Method
2.2.1 Data and measurements
Regional variation in on-going sound change: the case of the Dutch diphthongs
l l
dialectal
teachers of Dutch
× × ×
eːøːoːɛiœyɔu
l feː bøːk
boːx boːtən doː poːk dɛi
mɛit dœyn lœys tœyn sɔus
l keːl veːl bøːl
ɣoːl sxoːl zoːɫ ɣɛil ɦɛil
rœyl œyl l
2.2. Method
Figure 2.1: Overview of the different cities and towns from which data were sam-
pled. Dots indicate major cities and towns according to Barbiers et al.
(2006). Open circles indicate the cities and towns which were sam-
pled for the teacher-corpus data. The overlaid colored circles indicate
the corresponding regions, which are summarized in the following
table (based on van der Harst 2011:55):
Region Color Cities/Towns
Netherlands-Randstad Alphen aan den Rijn, Gouda
Netherlands-Middle Tiel, Veenendaal, Ede, Culemborg, Elst
Netherlands-North Assen, Veendam, Windschoten
Netherlands-South Sittard, Geleen, Roermond
Flemish Brabant Lier, Heist-op-den-Berg
Flemish Limburg Ieper, Poperinge
Flanders-East Oudenaarde, Zottegem
Flanders-West Tongeren, Bilzen
Regional variation in on-going sound change: the case of the Dutch diphthongs
l
+l l
2.2.2 Data analysis
⟨⟩ bam
mgcv
l l l l ×
l l
t
ρ=. ɔu
t
2.2. Method
l
l
×××=,
l
l l
ranges
dierence
l l ∆
l l
∆
∆ ∆
∆ ∆
∆
Regional variation in on-going sound change: the case of the Dutch diphthongs
= −
∆ =l−l
∆ = ∆ −∆
∆
n−
n
χn−=∑n
i=(x −x i)
(x )
data.csv analysis.R
2.3 Results
ɔu
l l
l included
2.3. Results
/œy/
/o:/
/Ei/
/e:/
/ø:/
0 25 50 75 100
0 25 50 75 100
200
300
400
500
600
200
300
400
500
600
200
300
400
500
600
Time (%)
F1 (Hz)
Region
NR
NM
NN
NS
FB
FL
FE
FW
Figure 2.2: Overview of the vowel trajectories as smoothed curves, for the F1 data
(closed/open dimension) when followed by a nonapproximant con-
sonant. The following consonant itself is not included. The ribbons
around the curves indicate the 95% CI.
Regional variation in on-going sound change: the case of the Dutch diphthongs
/œy/
/o:/
/Ei/
/e:/
/ø:/
0 25 50 75 100
0 25 50 75 100
200
300
400
500
600
200
300
400
500
600
200
300
400
500
600
Time (%)
F1 (Hz)
Region
NR
NM
NN
NS
FB
FL
FE
FW
Figure 2.3: Overview of the vowel trajectories as smoothed curves, for the F1 data
(closed/open dimension) when followed by coda /l/. The curves in-
clude the coda /l/ in its entirety. The ribbons around the curves indi-
cate the 95% CI.
2.3. Results
/œy/
/o:/
/Ei/
/e:/
/ø:/
0 25 50 75 100
0 25 50 75 100
1,000
1,500
2,000
2,500
1,000
1,500
2,000
2,500
1,000
1,500
2,000
2,500
Time (%)
F2 (Hz)
Region
NR
NM
NN
NS
FB
FL
FE
FW
Figure 2.4: Overview of the vowel trajectories as smoothed curves, for the F2 data
(front/back dimension) when followed by coda /l/. The curves include
the coda /l/ in its entirety. The ribbons around the curves indicate the
95% CI.
Regional variation in on-going sound change: the case of the Dutch diphthongs
2.3.1 Change 1: diphthongization of /e:,ø:,o:/
l
eːøːoː
>
only
øː
eː
ɛi
œy
χ
eː øː l
oː œy
2.3. Results
/œy/
/o:/
/Ei/
/e:/
/ø:/
0 25 50 75 100
0 25 50 75 100
−100
−50
0
50
100
150
−100
−50
0
50
100
150
−100
−50
0
50
100
150
Time (%)
F1 (Hz)
Region
NR
NM
NN
NS
FB
FL
FE
FW
Figure 2.5: Differences in vowel diphthongization before nonapproximant conso-
nants by the separate regions, relative to the Netherlands-Randstad
region.
Regional variation in on-going sound change: the case of the Dutch diphthongs
Table 2.2: Regional differences in diphthongization before nonapproximant con-
sonants. The “Timespan” column reflects the start and end point of the
consecutive stretch of largest significant differences from the Randstad;
the column labeled “95% CI” gives the 95% Bayesian credible interval
of this largest difference. Only significant results are shown. The two
right-hand columns are the lexical-diffusion measure; the χ2measures
the amount by which individual words deviate from the peak differ-
ence in the middle column.
Vowel Region Timespan (%) Peak diff. (Hz) 95% CI (Hz) χ2p
/e:/ NM 77 – 100 63.09 31.21 – 94.97
/e:/ NN 80 – 91 46.19 14.26 – 78.11
/e:/ NS 85 – 100 45.38 13.53 – 77.24
/e:/ FB 49 – 100 66.49 34.35 – 98.63
/e:/ FL 55 – 100 111.26 79.35 – 143.17
/e:/ FE 77 – 100 65.03 31.87 – 98.19
/e:/ FW 75 – 100 90.97 59.18 – 122.77
/ø:/ FB 89 – 100 39.35 12.77 – 65.92
/ø:/ FL 81 – 100 62.51 35.92 – 89.10
/ø:/ FW 88 – 100 49.51 22.94 – 76.09
/o:/ FB 85 – 100 47.69 23.17 – 72.20 11.19 .01
/o:/ FL 67 – 100 81.86 57.10 – 106.62 2.59 .46
/o:/ FE 87 – 100 47.68 23.29 – 72.06 50.94 <.001
/o:/ FW 74 – 100 86.24 61.86 – 110.62 22.73 <.001
/Ei/ NM 73 – 87 48.89 10.39 – 87.40 3.49 .06
/Ei/ FB 46 – 91 77.42 39.50 – 115.35 0.89 .35
/Ei/ FL 71 – 100 64.02 26.26 – 101.79 2.62 .11
/Ei/ FE 69 – 92 71.23 33.39 – 109.07 2.80 .09
/Ei/ FW 78 – 89 46.24 8.26 – 84.22 1.02 .31
/œy/ NN 53 – 76 −37.49 −74.12 – −0.86 14.07 <.001
/œy/ FB 62 – 100 73.40 37.06 – 109.73 3.06 .22
/œy/ FL 79 – 100 58.08 21.86 – 94.31 2.64 .27
/œy/ FE 88 – 100 45.43 9.03 – 81.83 6.37 .04
/œy/ FW 87 – 100 48.00 11.50 – 84.49 5.03 .08
2.3. Results
2.3.2 Change 2: lowering of /Ei,œy,Ou/
ɛiœyɔu aiɒyɑu
before
l
eː
øː oː ɛi
œy
øːoːœy
ɛi
χ oː ɛi
ɛi
oː
p.
ɛi
Regional variation in on-going sound change: the case of the Dutch diphthongs
Table 2.3: Regional differences in the lowering of /Ei,œy,Ou/. Only significant re-
sults are shown.
Vowel Region Timespan (%) Peak diff. (Hz) 95% CI (Hz) χ2p
/e:/ FE 19 – 29 −35.05 −66.97 – −3.13
/e:/ FW 13 – 34 −40.52 −72.26 – −8.79
/ø:/ NS 0 – 61 −45.22 −71.35 – −19.10
/ø:/ FB 9 – 49 −34.84 −60.47 – −9.22
/ø:/ FL 2 – 51 −40.42 −66.14 – −14.71
/ø:/ FE 11 – 65 −48.90 −74.78 – −23.02
/ø:/ FW 11 – 58 −45.07 −70.64 – −19.49
/o:/ NS 5 – 30 −39.13 −63.54 – −14.72 11.04 .01
/o:/ FB 7 – 52 −44.99 −69.30 – −20.68 32.89 <.001
/o:/ FL 7 – 26 −34.48 −58.99 – −9.98 9.38 .02
/o:/ FE 8 – 59 −46.02 −70.22 – −21.83 7.55 .06
/o:/ FW 12 – 38 −32.45 −56.44 – −8.46 17.37 <.001
/Ei/ NM 29 – 44 39.01 0.69 – 77.33 3.30 .07
/Ei/ NN 39 – 56 38.71 0.72 – 76.71 4.24 .04
/Ei/ FL 11 – 44 −76.94 −114.58 – −39.30 0.33 .57
/Ei/ FE 15 – 41 −65.47 −103.16 – −27.77 9.13 <.01
/Ei/ FW 17 – 45 −67.50 −105.31 – −29.68 1.28 .26
/œy/ NS 15 – 61 −73.24 −109.58 – −36.91 3.87 .14
/œy/ FL 10 – 60 −97.61 −133.71 – −61.51 1.01 .60
/œy/ FE 11 – 64 −86.28 −122.42 – −50.14 0.76 .68
/œy/ FW 15 – 65 −97.30 −133.56 – −61.04 3.15 .21
2.3. Results
2.3.3 Change 3: blocking of diphthongization before /l/
∆ l l
χ
∆
eː
œy
∆ eː
øː
oː
ɛi
l
l ∆
eː ∆
l l
χ
l
l combinations
χ
p
χ
Regional variation in on-going sound change: the case of the Dutch diphthongs
Table 2.4: Differences in the ranges of diphthongization before nonapproximant consonants versus before coda /l/, split out
by vowel and region in order to answer RQ 3. Only regions significantly different from the Netherlands-Randstad
are shown.
Vowel Region ΔTTP 95% CI ΔNR 95% CI (Hz) χ2p
/e:/ FB −26.61 −60.69 – 7.48 59.31 10.63 – 107.99 33.96 <.001
/e:/ FL −10.84 −45.13 – 23.44 75.07 26.25 – 123.90 47.30 <.001
/e:/ FW −10.82 −44.78 – 23.14 75.09 26.50 – 123.69 40.83 <.001
/e:/ NS 23.67 −8.68 – 56.02 109.59 62.10 – 157.07 76.38 <.001
/ø:/ FB 56.67 31.55 – 81.78 47.73 10.94 – 84.52 49.31 <.001
/ø:/ FE 50.52 22.29 – 78.75 41.59 2.61 – 80.57 24.40 <.001
/ø:/ FL 56.44 31.28 – 81.60 47.51 10.69 – 84.33 8.71 <.01
/ø:/ FW 55.85 31.50 – 80.19 46.91 10.65 – 83.18 3.90 .048
/o:/ FL 73.59 46.04 – 101.14 47.89 7.40 – 88.37 52.30 <.001
/Ei/ FL 61.07 8.63 – 113.50 82.57 6.91 – 158.23 16.50 <.001
/œy/ FB 72.76 21.53 – 124.00 107.27 35.92 – 178.62 106.03 <.001
/œy/ FE 54.07 2.56 – 105.59 88.58 17.02 – 160.13 193.81 <.001
/œy/ FL 37.58 −12.55 – 87.71 72.08 1.52 – 142.64 78.99 <.001
2.3. Results
2.3.4 Change 4: vocalization and retracting eect of coda /l/
l
l
eː oː
ɛi œy
l
oː
øː
œy
Regional variation in on-going sound change: the case of the Dutch diphthongs
/œy/
/o:/
/Ei/
/e:/
/ø:/
0 25 50 75 100
0 25 50 75 100
−500
0
500
−500
0
500
−500
0
500
Time (%)
F2 (Hz)
Region
NR
NM
NN
NS
FB
FL
FE
FW
Figure 2.6: Differences from the Randstad in the retraction of vowels including a
following coda /l/, averaged over gender.
2.3. Results
Table 2.5: Regional differences in the retracting effect of coda /l/.
Vowel Region Timespan (%) Peak diff. (Hz) 95% CI (Hz) χ2p
/e:/ NN 43 – 53 145.88 9.43 – 282.33 1.97 .16
/e:/ FB 1 – 99 458.69 322.60 – 594.79 1.30 .25
/e:/ FL 4 – 84 595.26 458.75 – 731.77 2.49 .11
/e:/ FE 2 – 100 689.18 553.29 – 825.07 0.06 .81
/e:/ FW 0 – 100 632.59 496.38 – 768.79 0.29 .59
/ø:/ FB 0 – 92 406.11 289.08 – 523.13 11.21 <.001
/ø:/ FL 37 – 89 260.18 143.59 – 376.78 3.27 .07
/ø:/ FE 0 – 98 694.84 579.86 – 809.82 0.44 .51
/ø:/ FW 35 – 95 528.36 413.00 – 643.71 5.34 .02
/o:/ NN 89 – 94 −110.78 −216.65 – −4.91 2.90 .09
/o:/ FB 76 – 100 262.49 157.64 – 367.35 9.29 <.01
/o:/ FL 80 – 100 219.23 114.69 – 323.76 12.40 <.001
/o:/ FE 51 – 99 432.83 328.59 – 537.07 25.19 <.001
/o:/ FW 66 – 100 295.56 190.87 – 400.25 5.59 .02
/Ei/ NS 66 – 91 209.11 65.89 – 352.33 0.11 .74
/Ei/ FB 0 – 100 649.40 506.24 – 792.56 1.59 .21
/Ei/ FL 0 – 97 597.89 455.55 – 740.23 0.97 .33
/Ei/ FE 0 – 99 699.22 557.11 – 841.33 2.95 .09
/Ei/ FW 0 – 99 637.34 494.76 – 779.92 1.97 .16
/œy/ NS 68 – 74 126.21 8.48 – 243.95 3.53 .06
/œy/ FB 28 – 99 552.74 434.87 – 670.62 1.81 .18
/œy/ FL 42 – 95 485.18 367.96 – 602.41 0.08 .78
/œy/ FE 8 – 99 792.47 676.23 – 908.71 7.75 <.01
/œy/ FW 30 – 97 702.30 585.62 – 818.98 0.81 .37
Regional variation in on-going sound change: the case of the Dutch diphthongs
2.4 Discussion
eːøːoː ɛiœyɔu
l
l
eːøːoː
eːøːoːɛiœy
eːøːoː ɛi œy
ɛiœy
ɛi
œy eːøːoː
l
l
2.4. Discussion
oː ɛi
l l
∆
l l
∆ more
l
l l l l
l
∆
l
∆
∆
l l
ɛiœy
eːøːoːɛiœy l eːøːoːɛːœː
ɛiœy
l l
ɛiœy
l
χ
Regional variation in on-going sound change: the case of the Dutch diphthongs
l
l
ɫ
l
oː
χ
oː χ
oː
oː
truly
continuum oː
oː ɔu
ɔu ɑu pɑul
pɑːɫ pɔːɫ
ɛiœy oː ɔu
oː
oː ɔu
χ
2.4. Discussion
ɔu
χ
χ
eːøːoː
ɛiœyɔu
Regional variation in on-going sound change: the case of the Dutch diphthongs
2.5 Conclusion
eːøːoː ɛiœyɔu
l l
Acknowledgments
Watching Dutch Change
This chapter has been submitted.
Abstract
3.1. Introduction
3.1 Introduction
3.1.1 Investigating the adoption of on-going sound change
how much
eːøːoː eiøyou
l
r
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
eiøyou ɛiœyɔu
ɛiœyɔu aiɒyɑu
ɹ
3.1.2 Sound change as second-dialect acquisition
sociolinguistic migrants
any
slightly
only
3.1. Introduction
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
3.1.3 The present study
l
l a priori
3.2. Experiment 1: rhyme decision
3.2 Experiment 1: rhyme decision
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
eː
ɛi
l
a priori
l
3.2. Experiment 1: rhyme decision
3.2.1 Method
Participants
Stimuli
eː∼ɛi oː∼ɑu ɛ∼ɛi aːʀ∼aːɹ eːɫ∼ɛiɫ
oːɫ∼ɑuɫ ɛɫ∼ɛiɫ
aːʀ∼aːɹ ʀ∼ɹ
ə(ɹ) ɹ
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
Table 3.1: Overview of the final population from which data were obtained.
“FDS” indicates a Flemish-Dutch speaker (i.e. a sociolinguistic migrant);
“NDS” indicates a Netherlandic-Dutch speaker (i.e. a control partici-
pant).
Session
Participant 123
FDS-0 333
FDS-1 3 3
FDS-2 3
FDS-3 333
FDS-4 333
FDS-5 333
FDS-6 333
FDS-7 3
FDS-8 3 3
FDS-9 333
NDS-0 333
NDS-1 333
NDS-2 333
NDS-3 333
NDS-4 333
NDS-5 333
NDS-6 333
NDS-7 3
NDS-8 333
NDS-9 333
3.2. Experiment 1: rhyme decision
Table 3.2: Schematic overview of the conditions in the rhyme-decision task. Be-
cause the vocalic conditions all appeared twice (with vs. without a
following coda /l/), the four rows in the table together make seven con-
ditions for the experiment. Each condition consisted of 34 items.
Left endpoint Right endpoint Possible coda /l/
[e:] [Ei] Yes
[o:] [Au] Yes
[E] [Ei] Yes
[ö] [ô] No
d
ʀ∼ɹ
eː∼ɛi
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
Amp ( Pa)
–0.57
0.91
Freq ( Hz)
0
5,000
F1 (Hz)
800
100
Time ( ms)
224 453 6900 930
[eː] [eːʲ]
[e̞i]
[ɛi]
Figure 3.1: Example waveforms, spectrograms, and F1 trajectories for the [e:∼Ei]
contrast. The four tokens shown in this figure correspond to the stimuli
containing 20%, 40%, 60%, and 80% morphing, respectively. Note
the lowering of the nucleus and the increase in the diphthong’s slope
over the four figures.
3.2. Experiment 1: rhyme decision
Table 3.3: Example of four words yolked across four participants. The percentages
refer to the amount of upgliding diphthongization ([e:]→[Ei]) present in
the speech signal.
Participant
Word 1 2 3 4
[de:t@] 20% 40% 60% 80%
[ble:t@] 40% 60% 80% 20%
[Xöe:v@ô] 60% 80% 20% 40%
[tVe:d@ô] 80% 20% 40% 60%
× =
Procedure
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
Time course
Visual
Auditory
Participant
Measure
500 ms
+
(until button press)
deete
[de:t@]40%
Z/M
button press
Figure 3.2: Example trial for the rhyme-decision task. At a later trial, the same au-
ditory stimulus will be presented, but the visual target ⟨deete⟩(/de:t@/)
will be replaced by ⟨dijte⟩(/dEit@/).
3.2. Experiment 1: rhyme decision
×=
Data analysis
< >
ɛiɑuɛi ɹ
eːoːɛ ʀ
glmertree
l l
α=.
buildmertree
buildmer
3.2.2 Results
eː∼ɛi
l
l
l
l
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
Step
p<.001
≤2>2
Following
p<.001
non-/l/ /l/
0%
50%
100%
Group
p<.001
FDS NDS
Following
p<.001
non-/l/ /l/
Group
p<.001
FDS NDS
Step
p<.01
≤3>3
Group
p<.001
FDS NDS
Step
p<.001
≤3>3
Figure 3.3: Logistic mixed-effects regression tree for the [e:∼Ei] continuum (20 par-
ticipants, 68 items). The target variable is the probability of indicating
an [Ei] percept.
3.2. Experiment 1: rhyme decision
Step
p<.001
≤2>2
Following
p<.001
non-/l/ /l/
Group
p<.001
FDS NDS
0%
50%
100%
Group
p<.001
FDS NDS
Step
p<.01
≤1>1
Step
p<.001
≤3>3
Following
p<.001
non-/l/ /l/
Following
p<.001
non-/l/ /l/
Figure 3.4: Logistic mixed-effects regression tree for the [o:∼Au] continuum (20
participants, 68 items). The target variable is the probability of indi-
cating an [Au] percept.
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
Step
p<.001
≤2>2
Following
p<.001
non-/l/ /l/
0%
50%
100%
Group
p<.001
FDS NDS
Group
p<.001
FDS NDS
Step
p<.001
≤3>3
Step
p<.001
≤3>3
Figure 3.5: Logistic mixed-effects regression tree for the [E∼Ei] continuum (20 par-
ticipants, 68 items). The target variable is the probability of indicating
an [Ei] percept.
3.2. Experiment 1: rhyme decision
Group
p<.001
FDS NDS
Session
p<.001
≤2>2
Session
p<.001
≤1>1
0%
50%
100%
Step
p = .01
≤3>3
Figure 3.6: Logistic mixed-effects regression tree for the [ö∼ô] continuum (20 par-
ticipants, 34 items). The target variable is the probability of indicating
an [ô] percept.
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
l
oː∼ɑu
<
l
l
l
ɛ∼ɛi
l
ɛi
ʀ∼ɹ
ɹ
ɹ
ɹ
ɹ
3.2.3 Discussion
3.2. Experiment 1: rhyme decision
l
a priori
eː∼ɛi
eː ɛi
eː ai
ɛiœyɔu aiɒyɑu oː∼ɑu
full
ɛ∼ɛi
eː∼ɛi
eː∼ɛi
l
eː∼ɛi NDS
oː∼ɑu
FDS
l
l l
l
ɛ∼ɛi
eː∼ɛi
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
eː∼ɛi
ɛi eː
eː
oː
eː∼ɛi oː∼ɑu
l
eː∼ɛi
l l
oː∼ɑu
ɛ∼ɛi
ɛi
ɹ
is
ɹ
ʀ∼ɹ
3.3. Experiment 2: word production
was
ɹ
fewer ɹ
ɹ
3.3 Experiment 2: word production
production
×
A priori
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
3.3.1 Method
Participants
Stimuli
× iuaː
× eːøːøːɛiœyɑu
aːʀɛ × eːløːloːlɛil
œylɑuaːʀɛl l
l
aːʀl ɑul
l øː
eiøyouɛiœyɑu
eːøːoːœːɑː
ʀ
ɹ
eiøyou
ɛiœyɑu eːɫøɫoːɫɛːɫœːɫ r
ɛ
eːøːoː ɛːœːɑː
3.3. Experiment 2: word production
l
ʀ
ɹ
ɛ
ɛː ɛ
eːɛiaːɹeːlɛilaːʀ
ɛ
×( −) =
Procedure
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
Table 3.4: Overview of the allophone variants used in the experiment (618 trials).
For the point vowels /i,u,a:/, both allophone variants are the same.
Realization used in prime items
Before non-/l/ Before /l/
Phoneme NDS Non-NDS NDS Non-NDS
/e:/ [ei] [e:] [e:] [ei]
/ø:/ [øy] [ø:] [ø:] [øy]
/o:/ [ou] [o:] [o:] [ou]
/Ei/ [Ei] [E:] [E:] [Ei]
/œy/ [œy] [œ:] [œ:] [œy]
/Au/ [Au] [A:] [A:] [Au]
/a:ö/ [a:ô] [a:ö]
/E/ [E] [E:] [E] [E:]
/i/ [i] [i] [i] [i]
/u/ [u] [u] [u] [u]
/a:/ [a:] [a:] [a:] [a:]
Time course
Visual
Auditory
Participant
Measure
1 s
+
(duration of prime)
[a:öd@]
2 s
aarde
[a:ôd@]
formants
Figure 3.7: Example trial for the production task.
3.3. Experiment 2: word production
× iuaː
Data analysis
∆
∆
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
l l
buildmer
lmerTest
pbkrtest
lme4
Mclust mclust
3.3. Experiment 2: word production
3.3.2 Results
ˆ
β t p<
l×
eːoːoː
œy
l× ˆ
β t
p<
l
×
× ×
∆
Table 3.5: Averages of the raw F3 data in Hz (20 participants, 68 items). Note how
the NDS consistently have lower F3s than the FDS, and that the FDS
do not appear to be moving closer to the NDS over the three sessions.
Session
Group Prime realization 1 2 3
FDS ND allophone 2,944 2,963 2,993
FDS non-ND allophone 2,971 2,982 2,972
NDS ND allophone 2,336 2,350 2,227
NDS non-ND allophone 2,367 2,352 2,205
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
FDS, session 1, /l/
FDS, session 2, /l/
FDS, session 3, /l/
NDS, /l/
FDS, session 1, non-/l/
FDS, session 2, non-/l/
FDS, session 3, non-/l/
NDS, non-/l/
-101 -101 -101 -101
-1
0
1
-1
0
1
F2 (Lobanov)
F1 (Lobanov)
Timepoint
25%
75%
Vowel
/i/
/u/
/a:/
/e:/
/ø:/
/o:/
/Ei/
/œy/
/Au/
Figure 3.8: Vowel-space plot of the raw F1/F2 data (20 participants, 550 items). For reasons of space, the data for the NDS
controls have been collapsed over the three sessions, as has the effect of prime realization. Observe how the
NDS have upgliding realizations of the vowels under investigation in the non-/l/ condition, and have non-upgliding
(downgliding) realizations in the coda-/l/ condition. The FDS, by contrast, have non-upgliding realizations in both
conditions, across all sessions; in the /l/ condition, they exhibit the same downglide as the NDS.
3.3. Experiment 2: word production
Table 3.6: Results for the F1 analysis (20 participants, 550 items). Observe the significant effect for
“Following segment = non-/l/”, which shows that the NDS produce more upgliding diph-
thongization in non-coda-/l/ environments than the FDS do. There are significant per-vowel
adjustments to this effect, but they do not obviate this main result. There are differences
across the three sessions of the experiment, but they, too, are not of sufficient magnitude
to make a meaningful contribution to the bigger picture.
Factor Estimate (SE) tdf pSig.
Intercept 8.04 ( 5.84) 1.38 25.08 .18
Following segment = non-/l/ −71.03 ( 8.88) −8.00 23.59 <.001 ∗∗∗
Vowel = /e:/ 21.00 ( 5.85) 3.59 111.35 <.001 ∗∗∗
Vowel = /ø:/ 22.90 ( 5.87) 3.90 109.64 <.001 ∗∗∗
Vowel = /o:/ −6.69 ( 6.18) −1.08 90.06 .28
Vowel = /œy/ 34.87 ( 6.27) 5.56 85.99 <.001 ∗∗∗
Vowel = /Au/ −96.40 (12.22) −7.89 23.17 <.001 ∗∗∗
Group = NDS −5.86 ( 7.99) −0.73 22.09 .47
Session (Linear) −2.84 ( 1.76) −1.62 19.06 .12
Session (Quadratic) 1.78 ( 1.74) 1.02 20.31 .32
Following segment = non-/l/ ×/e:/ 39.18 ( 8.13) 4.82 113.73 <.001 ∗∗∗
Following segment = non-/l/ ×/ø:/ 59.41 ( 8.30) 7.16 105.04 <.001 ∗∗∗
Following segment = non-/l/ ×/o:/ 46.71 ( 9.17) 5.09 74.07 <.001 ∗∗∗
Following segment = non-/l/ ×/œy/ −95.99 ( 8.59) −11.18 92.35 <.001 ∗∗∗
Following segment = non-/l/ ×NDS −93.21 (12.10) −7.70 20.78 <.001 ∗∗∗
Vowel = /e:/×Session (Linear) −1.59 ( 1.97) −0.81 18,239.26 .42
Vowel = /e:/×Session (Quadratic) −2.34 ( 1.96) −1.19 23,420.98 .23
Vowel = /ø:/×Session (Linear) 8.17 ( 1.97) 4.15 18,566.60 <.001 ∗∗∗
Vowel = /ø:/×Session (Quadratic) −5.89 ( 1.96) −3.00 23,448.58 <.01 ∗∗
Vowel = /o:/×Session (Linear) −1.10 ( 1.98) −0.55 20,059.96 .58
Vowel = /o:/×Session (Quadratic) −3.37 ( 1.96) −1.71 23,591.59 .09
Vowel = /œy/×Session (Linear) 3.95 ( 1.99) 1.99 20,285.82 .047 ∗
Vowel = /œy/×Session (Quadratic) 1.63 ( 1.97) 0.82 23,611.86 .41
Vowel = /Au/×Session (Linear) −14.00 ( 2.15) −6.51 23,856.46 <.001 ∗∗∗
Vowel = /Au/×Session (Quadratic) 14.36 ( 2.14) 6.71 23,859.16 <.001 ∗∗∗
Following segment = non-/l/ ×Session (Linear) 1.42 ( 5.54) 0.26 18.78 .80
Following segment = non-/l/ ×Session (Quadratic) 9.84 ( 4.43) 2.22 19.25 .04 ∗
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
Table 3.7: Results for the F3 analysis (20 participants, 68 items). Observe that the NDS have a signif-
icantly lower F3 than the FDS do, and that there is no significant evidence that this gap
narrows over the three sessions of the experiment.
Factor Estimate (SE) tdf pSig.
Intercept 2,952.15 ( 80.15) 36.83 21.43 <.001 ∗∗∗
Group = NDS −634.04 (111.91) −5.67 20.45 <.001 ∗∗∗
Session (Linear) 14.94 ( 39.89) 0.37 23.46 .71
Session (Quadratic) 80.00 ( 36.10) 0.22 28.26 .83
Prime realization = non-NDS 8.78 ( 20.95) 0.42 1,893.13 .68
Group = NDS ×Session (Linear) −77.70 ( 53.72) −1.45 23.15 .16
Group = NDS ×Session (Quadratic) −70.11 ( 48.89) −1.43 27.76 .16
Group = NDS ×non-NDS 0.55 ( 28.55) 0.02 1,891.52 .98
Session (Linear) ×non-NDS −30.06 ( 35.43) −0.85 1,889.10 .40
Session (Quadratic) ×non-NDS −4.97 ( 37.13) −0.13 1,897.87 .89
Group = NDS ×Session (Linear) ×non-NDS −7.76 ( 48.47) −0.16 1,889.12 .87
Group = NDS ×Session (Quadratic) ×non-NDS 0.97 ( 50.44) 0.02 1,895.08 .98
<
× ×
∆ ∆
×
∆
∆ ⟨.,.,.⟩ ⟨.,.,.⟩
3.3. Experiment 2: word production
ˆ
β t p<
×
∆
×
p
∆
∆ ⟨.,.,−⟩ ⟨.,.,−⟩
Mclust
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
F3: Non-NDS ×Session (Linear)
F3: Non-NDS ×Session (Quadratic)
∆F1: Non-NDS ×Session (Quadratic)
F3: Non-NDS
∆F1: Non-NDS ×/ø:/
∆F1: Non-NDS ×/o:/×non-/l/
-25
0
25
50
75
-25
0
25
50
75
-25
0
25
50
75
Participant
Measure (Hz)
Figure 3.9: The estimated by-participants random effects (also known as “BLUPs”,
for “best linear unbiased predictors”) for the random slopes involving
the factor “Prime realization” in the supplementary models. Random-
effect vectors that were estimated with zero variance (i.e. all by-
participants random-effect coefficients are zero) have been omitted.
Each panel represents a single random slope from either the ∆F1 or
the F3 model; each dot is a participant. Each panel has been sepa-
rated into two panes: the left pane corresponds to the ten Flemish
participants and the right pane corresponds to the ten Netherlandic
participants. Cluster analyses revealed no clusters in any of these six
random slopes.
3.3. Experiment 2: word production
3.3.3 Discussion
l l
r
l
l
l
l
l
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
×
3.4 General discussion and conclusion
how much
ʀ ɹ
3.4. General discussion and conclusion
do
did
eː∼ɛi
begun
How long is “a long term” for sound change? The eect of duration of immersion on
the adoption of on-going sound change
eː∼ɛi
Acknowledgments
Watching Dutch Change
This chapter has been accepted for publication as: Voeten, C. C. (to appear). Individual
dierences in the adoption of sound change. . https://doi.org/
./
Abstract
eːøːoː
ɛiœyɔu
4.1. Introduction
4.1 Introduction
4.1.1 The adoption of sound change
Individual dierences in the adoption of sound change
al-
most
4.1.2 Perception, production, and the individual
s
s ɹ
s s
ɹ s ʃ
s
s ʃ ʃ
time course
per se
vzɣ fsx
b p
4.1. Introduction
Individual dierences in the adoption of sound change
4.1.3 Phonological change vs. phonetic change
uː
uː ʉː ʉː
uː
uːʉː
ʊ
cudcould
4.1. Introduction
ʊ cud ʌ
uː
4.1.4 The present study
eːøːoː eiøyou
l
ɫ
rʋj
ɛiœyɔu
l
l
Individual dierences in the adoption of sound change
l
l
sociolinguistic migrants
l
use
eːøːoː
ɛiœyɑu
ɛiœyɑu
l
eːøːoː
ɛiœyɑu
eːøːoː
ɛiœyɑu
4.1. Introduction
l l
eːøːoː
ɛiœyɑu
eːøːoː
l
ɛiœyɑu
ɛiœy
aiɒy
l
l
Individual dierences in the adoption of sound change
eː∼ɛi oː∼ɑu
ɛ∼ɛi
eːoː ɛiɑu
ɛ∼ɛi
ɛ ɛi
4.2. Experiment 1: production
4.2 Experiment 1: production
4.2.1 Method
Participants
Individual dierences in the adoption of sound change
20
30
40
50
60
70
Ghent Migrant Migrant AOA Leiden
Age (years)
Figure 4.1: Dot and violin plots of the ages of the participants in the Ghent
(n= 43, 1 missing age value), migrant (n= 18, and Leiden (n= 45,
2 missing age values) groups. For the migrant group, the chronolog-
ical ages have been linked to the participants’ ages of arrival in the
Netherlands (“AOA”).
4.2. Experiment 1: production
Table 4.1: Regional backgrounds of the participants, defined as the province
in which they attended high school. Gelderland–Zeeland are Nether-
landic provinces, while the others are Flemish.
Group
Region Ghent Migrant Leiden
Gelderland 0 0 2
Netherlandic Limburg 0 0 1
North Holland 0 0 6
South Holland 0 0 28
Utrecht 0 0 5
Zeeland 0 0 2
Antwerp 5 1 0
Brussels 0 1 0
East and West Flanders 0 1 0
East Flanders 23 4 0
Flemish Brabant 6 4 0
Flemish Limburg 0 2 0
West Flanders 14 5 0
(missing) 0 0 1
Stimuli
l
l +l
øː ɑu
l
l
Individual dierences in the adoption of sound change
Table 4.2: Number of words in each cell of the stimuli design of the production ex-
periment, with the column “Analyzed” reflecting which of these were
included in the data analysis. In addition to the vowels mentioned here,
the practice part of the experiment elicited 30 tokens (5 per combina-
tion) of /i,u,a:/ before coda /l/ and before non-/l/.
Following segment +frequency
Vowel Analyzed coda /l/ non-app. +LF non-app. +HF
/e:/ Yes 20 20 20
/o:/ Yes 20 20 20
/Ei/ Yes 20 20 20
/œy/ Yes 20 20 20
/ø:/ No 11 20 0
/Au/ No 0 20 20
iuaː l
Procedure
4.2. Experiment 1: production
Data analysis
∆
eːøːɛiœy
∆
eːøːoːɛiœyɑu ɛi
l
l l l
either
⟨,⟩
Individual dierences in the adoption of sound change
t bam R mgcv
b
Mclust
mclust
production.csv production.R
4.2.2 Results
iuaː øːɑu
Analysis by groups
a priori
l
bam
4.2. Experiment 1: production
Leiden, /l/
Leiden, non-/l/
Leiden, non-/l/, HF
Migrant, /l/
Migrant, non-/l/
Migrant, non-/l/, HF
Ghent, /l/
Ghent, non-/l/
Ghent, non-/l/, HF
−101 −101 −101
−1
0
1
2
−1
0
1
2
−1
0
1
2
F2 (Lobanov)
F1 (Lobanov)
Timepoint
25%
75%
Vowel
/i/
/u/
/a:/
/e:/
/ø:/
/o:/
/Ei/
/œy/
/Au/
Figure 4.2: Vowel diagrams of the data collected in the study (106 participants,
361 items). The figure is divided into nine panels, to account for the
3 groups ×3 types of following segment in the design. The vowels of
experimental interest are /e:,o:,Ei,œy/; the other vowels are included
for context.
Individual dierences in the adoption of sound change
Table 4.3: Group-level results for the production data (106 participants, 235 items). Critical factors are “Following segment”
and its interaction with the group predictors. The key observations are that the average participant produces signifi-
cantly more diphthongization before non-/l/ than before coda /l/, and that this additionally varies between the three
groups. The migrant group produces significantly more diphthongization in this context than the Ghent group, and
the Leiden group produces even more diphthongization in this context than the average of the other two groups.
Factor Estimate (SE) t p Sig.
Intercept 0.20 (0.03) 6.86 <.001 ∗∗∗
Vowel = /e:/ 0.02 (0.04) 0.45 .65
Vowel = /o:/ −0.24 (0.05) −5.38 <.001 ∗∗∗
Vowel = /œy/ 0.11 (0.04) 2.43 .02 ∗
Following segment = non-/l/ −0.79 (0.04) −22.12 <.001 ∗∗∗
Group = Migrant–Ghent −0.06 (0.02) −2.92 <.01 ∗∗
Group = Leiden–Others 0.01 (0.01) 0.46 .64
Following segment = non-/l/×Frequency = HF −0.05 (0.02) −2.58 <.01 ∗∗
Vowel = /e:/×non-/l/ 0.31 (0.06) 5.59 <.001 ∗∗∗
Vowel = /o:/×non-/l/ 0.56 (0.06) 9.91 <.001 ∗∗∗
Vowel = /œy/×non-/l/ −0.53 (0.05) −9.68 <.001 ∗∗∗
Following segment = non-/l/×Migrant–Ghent −0.19 (0.03) −7.14 <.001 ∗∗∗
Following segment = non-/l/×Leiden–Others −0.20 (0.02) −12.13 <.001 ∗∗∗
Vowel = /e:/×non-/l/×Migrant–Ghent 0.12 (0.03) 4.45 <.001 ∗∗∗
Vowel = /o:/×non-/l/×Migrant–Ghent 0.11 (0.03) 3.91 <.001 ∗∗∗
Vowel = /œy/×non-/l/×Migrant–Ghent −0.15 (0.02) −6.20 <.001 ∗∗∗
Vowel = /e:/×non-/l/×Leiden–Others 0.07 (0.02) 4.21 <.001 ∗∗∗
Vowel = /o:/×non-/l/×Leiden–Others 0.03 (0.02) 1.81 .07
Vowel = /œy/×non-/l/×Leiden–Others −0.04 (0.02) −2.85 <.01 ∗∗
4.2. Experiment 1: production
ˆ
β t p<
l×
t p<
l×
t p<
l
Analysis by participants
l
a priori
Individual dierences in the adoption of sound change
−0.8
−0.4
0.0
0.4
0.8
Participant
∆F1 (Lobanov)
Cluster
Netherlandic-like
Flemish-like
Figure 4.3: BLUPs for the “Following segment = non-/l/” term in the by-individu-
als (n= 106) model. Each dot is a participant’s individual random-
effect coefficient; lines indicate the standard errors. The left pane
shows the participants from the Ghent group, the middle pane shows
the participants from the migrant group, and the right pane shows the
participants from the Leiden group.
4.2. Experiment 1: production
4.2.3 Discussion
l
l l
l
Individual dierences in the adoption of sound change
4.3 Experiment 2: rhyme decision
4.3.1 Method
Participants
Stimuli
eː∼ɛi oː∼ɑu ɛ∼ɛi
×
ɫtdəɹ
l r t d
ɹ
ɹ
l
eː∼ɛi
4.3. Experiment 2: rhyme decision
Amp ( Pa)
–0.57
0.91
Freq ( Hz)
0
5,000
F1 (Hz)
800
100
Time ( ms)
224 453 6900 930
[eː] [eːʲ]
[e̞i]
[ɛi]
Figure 4.4: Example waveforms, spectrograms, and F1 trajectories for the [e:∼Ei]
contrast. The four tokens shown in this figure correspond to the stimuli
containing 20%, 40%, 60%, and 80% morphing, respectively. Note
the increase in the diphthong’s slope and the lowering of its nucleus
over the four figures.
Individual dierences in the adoption of sound change
⟨⟩neːbənɛibə
beːbəbeːjbəb̞eibəbɛibə
⟨⟩ beːbə
⟨⟩ veːjdə
beːjbə v̞eidə
Procedure
4.3. Experiment 2: rhyme decision
Data analysis
< >
eː∼ɛi oː∼ɑu ɛ∼ɛi
eːoːɛ
ɛiɑuɛi
l
l
buildglmmTMB buildmer
Individual dierences in the adoption of sound change
rhyme.csv rhyme.R
4.3.2 Results
α
p ∗ ∗∗ ∗∗∗
eː∼ɛi
ˆ
β z p< ˆ
β
z p<
l
ˆ
β z p<
ˆ
β
z p<
ˆ
β
z p<
ˆ
β z
p<
l
l
oː∼ɑu
ˆ
β z p< ˆ
β
z p<
l ˆ
β z p<
4.3. Experiment 2: rhyme decision
Table 4.4: Results of the rhyme-decision task (106 participants, 1,536 items). Only signif-
icant results are shown; the reader is referred to Appendix E for the full result
set. The key results are (1) the significant linear trends of participants indicat-
ing more diphthong percepts at later morphing steps; (2) participants becoming
more likely to give diphthong responses to a following coda /l/, demonstrating
perceptual compensation in the non-/l/ words; (3) significant between-groups
differences in the effect of morphing step in the [e:∼Ei] and [E∼Ei] models.
Factor Estimate (SE) Odds ratio z p Sig.
Model = [e:∼
∼
∼Ei]
Intercept −1.46 (0.10) 1 : 4.30 −14.49 <.001 ∗∗∗
Step (Linear) 1.24 (0.13) 3.47 : 1 9.87 <.001 ∗∗∗
Step (Quadratic) 0.33 (0.08) 1.39 : 1 4.22 <.001 ∗∗∗
Following segment = /l/ 0.83 (0.11) 2.29 : 1 7.59 <.001 ∗∗∗
Group = Migrant–Ghent −0.51 (0.13) 1 : 1.66 −3.95 <.001 ∗∗∗
Step (Linear) ×/l/ 1.68 (0.21) 5.37 : 1 8.14 <.001 ∗∗∗
Step (Linear) ×Migrant–Ghent 0.63 (0.15) 1.87 : 1 4.06 <.001 ∗∗∗
Model = [o:∼
∼
∼Au]
Intercept 0.20 (0.07) 1.22 : 1 2.76 .01 ∗
Step (Linear) 1.73 (0.15) 5.65 : 1 11.48 <.001 ∗∗∗
Step (Quadratic) 0.45 (0.11) 1.57 : 1 4.17 <.001 ∗∗∗
Following segment = /l/ 0.95 (0.12) 2.59 : 1 7.83 <.001 ∗∗∗
Step (Linear) ×/l/ −0.66 (0.25) 1 : 1.93 −2.67 .01 ∗
Step (Quadratic) ×/l/−0.74 (0.22) 1 : 2.09 −3.32 <.001 ∗∗
Model = [E∼
∼
∼Ei]
Intercept −0.94 (0.07) 1 : 2.55 −13.12 <.001 ∗∗∗
Step (Linear) 2.49 (0.15) 12.06 : 1 16.44 <.001 ∗∗∗
Step (Quadratic) 0.27 (0.10) 1.31 : 1 2.59 .01 ∗
Step (Cubic) −0.74 (0.11) 1 : 2.09 −6.99 <.001 ∗∗∗
Following segment = /l/ 0.96 (0.12) 2.60 : 1 7.90 <.001 ∗∗∗
Group = Migrant–Ghent −0.21 (0.08) 1 : 1.23 −2.64 .01 ∗
Group = Leiden–Others 0.09 (0.04) 1.10 : 1 2.50 .01 ∗
Step (Linear) ×/l/ −1.38 (0.24) 1 : 3.98 −5.84 <.001 ∗∗∗
Step (Cubic) ×/l/ 0.87 (0.21) 2.38 : 1 4.13 <.001 ∗∗∗
Step (Linear) ×Migrant–Ghent 0.43 (0.17) 1.53 : 1 2.52 .01 ∗
Step (Linear) ×Leiden–Others −0.23 (0.08) 1 : 1.26 −2.91 <.01 ∗∗
Step (Linear) ×/l/×Leiden–Others −0.31 (0.11) 1 : 1.36 −2.95 <.01 ∗∗
Individual dierences in the adoption of sound change
ˆ
β
z p ˆ
β z p
<
l
oː∼ɑu
ɛ∼ɛi
ɛi
ɛ
ˆ
β
z p< ˆ
β z p
ˆ
β
z p<
l
ˆ
β
z p< oː∼ɑu
l
ˆ
β
z p< ˆ
β z p<
l l
ˆ
β
z p
ˆ
β
z p
ˆ
β z p
ˆ
β z p<
l
l
l l
l
l l
4.3. Experiment 2: rhyme decision
[e:∼Ei]
[o:∼Au]
[E∼Ei]
20 40 60 8020 40 60 8020 40 60 80
0
25
50
75
100
Step (%)
Diphthong percept (%)
Following segment
/l/
non-/l/
Group
Ghent
Migrant
Leiden
Figure 4.5: Averaged raw data from the rhyme-decision task (106 participants,
1,536 items). The general trends are that (1) participants become more
likely to indicate a diphthong percept at later morphing steps; (2) this
effect is larger before coda /l/ than before non-/l/, indicating that partic-
ipants are perceptually compensating in the latter but not the former
condition; (3) there are differences between the groups both at the
baseline and as a function of the morphing step.
ɛ∼ɛi
ɛi
eː∼ɛi ɛ∼ɛi
Individual dierences in the adoption of sound change
1 : 20.01
1 : 7.39
1 : 2.72
1:1
20 40 60 80
Step (%)
Odds ratio
Following segment
/l/
non-/l/
Figure 4.6: Partial-effect plot of the three-way interaction in Table 4.4. The plot
shows the difference between the Leiden group and the others, in their
interaction of “Step” (on the xaxis) and “Following segment” (as sepa-
rate lines), in the [E∼Ei] condition. The yaxis is on a logarithmic scale,
as this is the scale on which the partial effects in the logistic-regression
analysis are linear. Observe that the Leiden group has much lower
odds of reporting a diphthong percept than the other two groups in
the first three steps, but at the fourth morphing step this preference
reverses and the Leiden group has slightly higher odds of reporting
a diphthong percept than the two other groups. Finally, note that this
effect is much more pronounced, in having a steeper linear slope lead-
ing to larger group differences in the earlier steps, in the non-/l/ case
than in the /l/ case.
4.3. Experiment 2: rhyme decision
4.3.3 Discussion
l
eː∼ɛi l
l
eː
eː l
ɛi
eː
ɛi
l
eː∼ɛi entire
eː ɛi
ɛi oː∼ɑu
full
ɑu
oː∼ɑu
oː ɑu
eː∼ɛi
oː∼ɑu eː∼ɛi
Individual dierences in the adoption of sound change
l
eː∼ɛi oː∼ɑu
eːøːoː
ɛ∼ɛi
4.4 Link between production and perception
l
4.4.1 Method
×
×
l
4.4. Link between production and perception
z
between
correlation.R
4.4.2 Results
∆
∆
eː∼ɛi
r
l
r oː∼ɑu
less r ɛ∼ɛi
r
l r
Individual dierences in the adoption of sound change
Table 4.5: Correlations of the various random slopes for the rhyme-decision task
with the “Following segment = non-/l/” random slope from the pro-
duction task (n= 106). F24,82 = 3.23, p<.001, R2= .49, R2adj = .34.
The correlations are visualized in Figure 4.7.
Factor Estimate (SE) t p Sig.
Model = [e:∼Ei], Intercept .24 (.13) 1.80 .08
Model = [e:∼Ei], Step (Linear) −.27 (.12) −2.30 .02 ∗
Model = [e:∼Ei], Step (Quadratic) .02 (.09) 0.16 .87
Model = [e:∼Ei], Step (Cubic) −.12 (.09) −1.41 .16
Model = [e:∼Ei], /l/ −.23 (.09) −2.49 .01 ∗
Model = [e:∼Ei], Step (Linear) ×/l/ −.08 (.10) −0.77 .44
Model = [e:∼Ei], Step (Quadratic) ×/l/ .12 (.09) 1.40 .17
Model = [e:∼Ei], Step (Cubic) ×/l/ −.02 (.09) −0.22 .83
Model = [o:∼Au], Intercept .13 (.10) 1.33 .19
Model = [o:∼Au], Step (Linear) .01 (.15) 0.08 .94
Model = [o:∼Au], Step (Quadratic) −.17 (.11) −1.54 .13
Model = [o:∼Au], Step (Cubic) −.06 (.09) −0.60 .55
Model = [o:∼Au], /l/ .28 (.11) 2.53 .01 ∗
Model = [o:∼Au], Step (Linear) ×/l/ .16 (.12) 1.36 .18
Model = [o:∼Au], Step (Quadratic) ×/l/ .01 (.09) 0.12 .91
Model = [o:∼Au], Step (Cubic) ×/l/ −.09 (.10) −0.86 .39
Model = [E∼Ei], Intercept −.12 (.13) −0.91 .36
Model = [E∼Ei], Step (Linear) .43 (.13) 3.31 <.01 ∗∗
Model = [E∼Ei], Step (Quadratic) −.02 (.10) −0.20 .84
Model = [E∼Ei], Step (Cubic) .02 (.10) 0.23 .82
Model = [E∼Ei], /l/ −.09 (.11) −0.86 .39
Model = [E∼Ei], Step (Linear) ×/l/ .31 (.11) 2.87 <.01 ∗∗
Model = [E∼Ei], Step (Quadratic) ×/l/ .08 (.11) 0.73 .46
Model = [E∼Ei], Step (Cubic) ×/l/ .08 (.11) 0.77 .44
4.4. Link between production and perception
[E∼Ei]: Step (Linear)
[E∼Ei]: Step (Linear) ×/l/
[e:∼Ei]: Step (Linear)
[e:∼Ei]: /l/
[o:∼Au]: /l/
−2 −1 0 1 2 −2 −1 0 1 2
−2 −1 0 1 2
−0.5
0.0
0.5
−0.5
0.0
0.5
Perception effect (log odds)
∆F1 (Lobanov)
Figure 4.7: The partial correlations that reached significance in the analysis
(n= 106), backtransformed to the original linear-predictor scales. Par-
ticipants who produce more diphthongization are more likely to in-
dicate a diphthong percept in the [e:∼Ei] perception model at later
morphing steps as well as when a coda /l/ followed the vowel. In the
[o:∼Au] model, participants who produce more diphthongization are
less likely to indicate a diphthong percept when a coda /l/ follows.
Finally, in the control model [E∼Ei], participants who diphthongize
more in production are less likely to indicate a diphthong percept as
a function of the morphing step in both the non-/l/ and /l/ conditions.
Individual dierences in the adoption of sound change
4.4.3 Discussion
eː
ɛi
eː ei
l
l
l eː∼ɛi
oː∼ɑu
l l
l
eː∼ɛi
l
ɛ∼ɛi
more ɛi
ɛi
eː∼ɛi l
4.5. General discussion
4.5 General discussion
l
some
eː∼ɛi
ɛ∼ɛi
ɛi
Individual dierences in the adoption of sound change
eː∼ɛi
oː∼ɑu
oː∼ɑu
ɛi
4.6 Conclusion
4.6. Conclusion
n
Acknowledgments
Language & Speech
Individual dierences in the adoption of sound change
Funding
Watching Dutch Change
Declaration of Conicting Interest
This chapter has been submitted.
Abstract
eːøːoː eiøyou l
r ɹ
ei eː
r
5.1. Introduction
5.1 Introduction
processing
k>ʧ>s
caelum cielo ciel
perception
representation
encode
Noticing the change: misrepresentation, not misperception, of allophonic variants in
sound change
td
t d
5.1. Introduction
violations
per se
eːøːoː eiøyou
l
l
r ɹ
r
ɹ
ɹ
Noticing the change: misrepresentation, not misperception, of allophonic variants in
sound change
Table 5.1: The relevant allophonic rules involved in the on-going sound changes
and their regional differences.
Underlying form Netherlandic realization Flemish realization
/e:/ followed by coda /l/ [e:] [e:]
/e:/ elsewhere [ei] [e:]
/@r/ [@ô] [@r]
eːɫ∼eiɫ eː∼ei ər∼əɹ
encode
evaluate
5.1. Introduction
inter alia
Noticing the change: misrepresentation, not misperception, of allophonic variants in
sound change
5.2 Method
5.2.1 Participants
5.2.2 Stimuli
eː ei eiɫ eːɫ ər
əɹ
eːeiərəɹ
eːɫeiɫ
əɹ əɹ
ər
əɹ
r
5.2. Method
[ər] [ər] [ər] [ər] [ər] [əɹ]
Amp ( Pa)
0.98
Freq ( Hz)
0
5,000
F3 (Hz)
4,000
1,000
Time ( ms)
200 400 600 800 10000 1200
–0.99
Figure 5.1: Example waveforms, spectrograms, and F3 trajectories (the critical dif-
ference between the two types of rhotic realization) of all five tokens
of [@r] used as standards and one token of [@ô] used as deviant, which
together make up Table 5.2’s [@ô] condition.
Noticing the change: misrepresentation, not misperception, of allophonic variants in
sound change
Table 5.2: Design of the six conditions used in the experiment.
Standard (68×5tokens) Deviant (60×1token)
[e:] [ei]
[ei] [e:]
[ei l
&] [e: l
&]
[e: l
&] [ei l
&]
[@r][@ô]
[@ô] [@r]
5.2.3 Procedure
5.2. Method
5.2.4 Data analysis
eegUtils
× ×
gam mgcv
Noticing the change: misrepresentation, not misperception, of allophonic variants in
sound change
×
t
p
present absent
=(−
( − ))
5.3 Results
× ×
5.3. Results
Condition
Factor [ei] [e:] [e: l
&] [ei l
&] [@ô] [@r]
Reference smooth 0.88 1.34 −0.01 2.75 0.34 31.28
Deviant 0.58 −0.26 −0.11 −0.10 34.70 24.46
Group 0.01 −0.08 2.94 7.09 −1.71 0.17
Session −0.07 15.61 3.42 7.41 3.72 −1.18
Deviant ×Group 0.52 2.05 −0.67 −1.42 20.73 −12.30
Deviant ×Session 11.18 −0.41 4.31 −0.05 5.94 −0.39
Group ×Session −0.04 0.00 −0.04 0.92 3.16 −12.29
Deviant ×Group ×Session 0.53 15.14 −0.42 0.43 −0.07 −3.56
Table 5.3: Results of the statistical analyses, reported as Bayes factors on the log10
scale.
×
ei ei eː
ei
eː ei
eː
eː
Noticing the change: misrepresentation, not misperception, of allophonic variants in
sound change
[ei]: Deviant [ei]: Deviant ×Group [ei]: Deviant ×Group ×Session
[e:]: Deviant ×Group [e:]: Deviant ×Group ×Session
[@ô]: Deviant [@ô]: Deviant ×Group [@r]: Deviant
+10−1−2−3
Difference (µV)
Figure 5.2: Topographical plots of the marginal effects of interest whose Bayes fac-
tors indicated at least substantial support for the alternative hypothesis.
The baseline is the sounds used as standards, heard by the Nether-
landic controls, in the first session of the experiment.
5.3. Results
[@ô]: Deviant (Netherlandic group) [@ô]: Deviant (Flemish group)
+10−1−2−3
Difference (µV)
Figure 5.3: Side-by-side comparison of the “[@ô]: Deviant” effect (left) and the sum
of this effect and the “[@ô]: Deviant ×Group” effect (i.e. the MMN dif-
ference between the baseline Netherlandic and the contrasting Flem-
ish group; right). The magnitudes of the MMNs are very similar be-
tween the groups, but the Flemish group has the effect shifted signifi-
cantly towards the midpoint of the scalp.
əɹ
əɹ
əɹ ər
ər
Noticing the change: misrepresentation, not misperception, of allophonic variants in
sound change
5.4 Discussion
eː ei
ei
ei
eː ei
ei
eː
eː
eː∼ei
ei
5.4. Discussion
eː
ei eː
eː
eːɫ eiɫ
×
eːɫ eiɫ
əɹ ər
əɹ ər
ɹ
r
Noticing the change: misrepresentation, not misperception, of allophonic variants in
sound change
əɹ ər əɹ
əɹ
ɹ
ɹ
ei eː əɹ
ei stronger
do
5.5. Conclusion
5.5 Conclusion
ei
eː ɹ
ei
ɹ
eː
grammar
ei
Noticing the change: misrepresentation, not misperception, of allophonic variants in
sound change
Acknowledgments
This chapter has been published as: Voeten, C. C., & Levelt, C. C. (2019). ERP
responses to regional accent reect two distinct processes of perceptual compensation.
,, 546. https://doi.org/./fnins..
Abstract
6.1. Introduction
6.1 Introduction
mis-
perception
unsuccessful
ERP responses to regional accent reect two distinct processes of perceptual
compensation
eːøːoː
ɛiœyɑu r
eːøːoːɛiœyɑu
eiøyouɛiœyɑu ɹ
eːøːoː eːøːoː
ɛiœyɑu r
ɛ ɛː
6.2. Accent processing
6.2 Accent processing
reduced
increasing
ERP responses to regional accent reect two distinct processes of perceptual
compensation
that does not occur normally œy ɔɪ
6.2. Accent processing
ERP responses to regional accent reect two distinct processes of perceptual
compensation
6.3 Materials and method
6.3.1 Participants
× =
6.3. Materials and method
Table 6.1: Overview of the final population from which data were obtained.
Session
Participant 123
FDS-0 3 3
FDS-1 3 3
FDS-2 3
FDS-3 333
FDS-4 333
FDS-5 333
FDS-6 333
FDS-7 3
FDS-8 3 3
FDS-9 333
NDS-0 333
NDS-1 333
NDS-2 333
NDS-3 333
NDS-4 333
NDS-5 333
NDS-6 333
NDS-7 3
NDS-8 333
NDS-9 333
ERP responses to regional accent reect two distinct processes of perceptual
compensation
6.3.2 Stimuli
eːøːoːɛiœyɑuaːʀɛ
l
aːʀ
l
ɑul
ɑu l ×
iuaː
l l
øː
øː
eːøːoːɛiœyɑu eiøyouɛiœyɑu
6.3. Materials and method
l
eːøːoːɛːœːɑː
ɛ ɛ
aːʀ aːɹ
eːøːoːɛiœyɑu eːøːoːɛːœːɑː
l
l eiøyouɛiœyɑu
l ɛ
ɛː
aːʀ aːʀ
ʀ ɹ
eːn ein
eːn
eːɛiaːʀ
eːɛiaːʀ
ERP responses to regional accent reect two distinct processes of perceptual
compensation
Table 6.2: Overview of the allophone variants used in the experimental items.
Realization (NDS versus non-NDS) used in prime items
Before non-/l/ (Target) Before /l/ (Filler)
Phoneme NDS Typical for Non-NDS Typical for NDS Typical for Non-NDS Typical for
/e:/ [ei] NDS [e:] FDS [e:] FDS [ei] neither
/ø:/ [øy] NDS [ø:] FDS [ø:] FDS [øy] neither
/o:/ [ou] NDS [o:] FDS [o:] FDS [ou] neither
/Ei/ [Ei] NDS [E:] FDS [E:] FDS [Ei] neither
/œy/ [œy] NDS [œ:] FDS [œ:] FDS [œy] neither
/Au/ [Au] NDS [A:] FDS [A:] FDS [Au] neither
/a:ö/ [a:ô] NDS [a:ö] FDS
/E/ [E] NDS+FDS [E:] neither [E] NDS+FDS [E:] neither
/i/ [i] NDS+FDS [i] NDS+FDS [i] NDS+FDS [i] NDS+FDS
/u/ [u] NDS+FDS [u] NDS+FDS [u] NDS+FDS [u] NDS+FDS
/a:/ [a:] NDS+FDS [a:] NDS+FDS [a:] NDS+FDS [a:] NDS+FDS
6.3. Materials and method
Amp ( Pa)
–0.92
0.80
Freq ( Hz)
0
5,000
F1 (Hz)
800
100
Time ( ms)
0 380
Figure 6.1: Example waveform, spectrogram, and F1 trajectory (the critical differ-
ence between diphthongal and monophthongal realizations) for the
NDS realization of /e:n/ as [ein]. Towards the end of the vowel, the F1
falls.
ERP responses to regional accent reect two distinct processes of perceptual
compensation
Amp ( Pa)
-0.93
0.63
Freq ( Hz)
0
5,000
F1 (Hz)
800
100
Time ( ms)
0 312
Figure 6.2: Example waveform, spectrogram, and F1 trajectory (the critical differ-
ence between diphthongal and monophthongal realizations) for the
non-NDS realization of /e:n/ as [e:n]. The F1 stays stable throughout
the vowel.
6.3. Materials and method
6.3.3 Procedure and data acquisition
ERP responses to regional accent reect two distinct processes of perceptual
compensation
Time course
Visual
Audio
Participant
Measure
1 s
+
(duration of prime)
[a:öd@]
1 s
aarde
[a:ôd@]
RTEEG signal
(1 s; part of following trial)
+
Figure 6.3: Example trial for the production task.
6.4. Results
6.4 Results
6.4.1 Reaction times
glmer
lme4
aːʀeːɛi
buildmer
ˆ
β
SE t p<
ˆ
β SE
ERP responses to regional accent reect two distinct processes of perceptual
compensation
Table 6.3: Fixed-effect coefficients of the reaction-times analysis. The model addi-
tionally includes random intercepts by participants and by words, and
random slopes for the factor “Session” by participants and by words.
Factor Estimate (SE) t p Sig.
(Intercept) 827.16 (2.22) 372.06 <.001 ∗∗∗
Session (Linear) 54.22 (1.86) 29.13 <.001 ∗∗∗
Session (Quadratic) −10.59 (2.90) −3.65 <.001 ∗∗∗
Group = FDS 41.32 (1.93) 21.37 <.001 ∗∗∗
Prime = Identity −46.64 (2.03) −23.02 <.001 ∗∗∗
Allophone = Non-NDS −0.15 (1.71) −0.09 .93
Group = FDS ×Identity 11.61 (2.28) 5.10 <.001 ∗∗∗
Prime = Identity ×Non-NDS −6.00 (1.67) −3.59 <.001 ∗∗∗
t p<
ˆ
β SE
t p< ˆ
β
SE t p<
ˆ
β SE t p<
ˆ
β SE t p<
ˆ
β SE t p<
6.4.2 ERP results
⟨⟩
permutes
6.4. Results
(A) FDS
(B) NDS
0 200 400 600 800 0 200 400 600 800
0.0
0.5
1.0
1.5
2.0
Time (ms)
Amplitude (µV)
Allophone
Netherlandic
Non-Netherlandic
Figure 6.4: Grand-average waveforms calculated over the full dataset, averaged
over all participants, electrodes, and the three sessions.
×
eːøːoːɛiœyɑu ɛaːʀ
ɛ aːʀ
ei∼eː øy∼øː
ou∼oː ɛi∼ɛː œy∼œː ɑu∼ɑː aːɹ∼aːʀ ɛ∼ɛː
×
ERP responses to regional accent reect two distinct processes of perceptual
compensation
AF3
AF4
C3
C4
CP1
CP2
CP5
CP6
Cz
F3
F4
F7
F8
FC1
FC2
FC5
FC6
Fp1
Fp2
Fz
O1
O2
Oz
P3
P4
P7
P8
PO3
PO4
Pz
T7
T8
0 200 400 600 800
Time (ms)
Electrode
20
40
60
F
Figure 6.5: Permutation tests performed on the whole dataset, showing the effect
of the factor “Group”. A significant difference can be observed, which
reaches permutation-based significance between 390–470 ms.
AF3
AF4
C3
C4
CP1
CP2
CP5
CP6
Cz
F3
F4
F7
F8
FC1
FC2
FC5
FC6
Fp1
Fp2
Fz
O1
O2
Oz
P3
P4
P7
P8
PO3
PO4
Pz
T7
T8
0 200 400 600 800
Time (ms)
Electrode
4
8
12
16
F
Figure 6.6: Permutation tests performed on the data for the [E∼E:] contrast, show-
ing the effect of the factor “Allophone ×Group”. A significant differ-
ence can be observed, which reaches permutation-based significance
between 560–660 ms.
6.4. Results
(A) FDS
(B) NDS
0 200 400 600 800 0 200 400 600 800
0.0
0.5
1.0
1.5
2.0
Time (ms)
Amplitude (µV)
Figure 6.7: Grand-average waveforms calculated over the full dataset, averaged
over the three sessions and the two allophone conditions. A differ-
ence in amplitude can be observed between the FDS and the controls,
which reaches permutation-based significance in the 390–470 ms win-
dow.
(A) FDS
(B) NDS
0 200 400 600 800 0 200 400 600 800
0
1
2
Time (ms)
Amplitude (µV)
Allophone
[E]
[E:]
Figure 6.8: Grand-average waveforms calculated over the [E∼E:] condition only,
averaged over all participants, electrodes, and the three sessions. A
difference in amplitude can be observed between the two allophone
conditions, in the NDS group only, which reaches permutation-based
significance in the 560–660 ms window.
ERP responses to regional accent reect two distinct processes of perceptual
compensation
ˆ
β
t,. p ˆ
β
t,. p<
ˆ
β
t,. p
ˆ
β t,. p
ˆ
β t,. p<
ˆ
β
t,. p<
6.4. Results
Table 6.4: Fixed-effect coefficients for the N400 effect.
Factor Estimate (SE) tdf pSig.
Intercept 1.43 (0.74) 1.94 17.78 .07
Group = FDS −0.89 (1.04) −0.85 17.97 .41
Session (Linear) 0.45 (0.18) 2.55 12,650.91 .01 ∗
Session (Quadratic) 0.79 (0.17) 4.55 12,672.79 <.001 ∗∗∗
Group = FDS ×Session (Linear) −1.17 (0.29) −3.97 12,250.75 <.001 ∗∗∗
Group = FDS ×Session (Quadratic) 0.21 (0.27) 0.78 12,665.72 .44
Table 6.5: Fixed-effect coefficients for the P600 effect.
Factor Estimate (SE) tdf pSig.
Intercept 0.81 (0.51) 1.60 24.60 .12
Session (Linear) −0.76 (0.31) −2.47 1,179.68 .01 ∗
Session (Quadratic) −0.17 (0.29) −0.59 1,400.81 .56
Allophone = Non-NDS 1.70 (0.44) 3.86 1,394.70 <.001 ∗∗∗
Group = FDS 0.49 (0.74) 0.66 27.79 .51
Allophone = Non-NDS ×FDS −2.51 (0.67) −3.76 1,395.54 <.001 ∗∗∗
ERP responses to regional accent reect two distinct processes of perceptual
compensation
6.4.3 Topographical distribution
ɛ ɛː
ɛː ɛ
6.4. Results
0.00
0.50
1.00
1.50
2.00
Amplitude (µV)
0.00
1.00
2.00
3.00
4.00
Amplitude (µV)
−2.00
−1.00
0.00
Amplitude (µV)
Figure 6.9: Topographical distribution of the N400 effect. The left head shows the FDS, the middle head shows the NDS, and
the right head shows the difference between the two.
ERP responses to regional accent reect two distinct processes of perceptual
compensation
1.00
2.00
3.00
Amplitude (µV)
0.50
1.00
1.50
2.00
Amplitude (µV)
−0.60
−0.40
−0.20
0.00
Amplitude (µV)
1.00
2.00
Amplitude (µV)
1.00
1.50
2.00
2.50
3.00
Amplitude (µV)
0.20
0.30
0.40
0.50
Amplitude (µV)
Figure 6.10: Topographical distribution of the P600 effect. The top three heads show the FDS, and the bottom three heads show
the NDS. From left to right, both rows display, respectively: the NDS-allophone words; the non-NDS-allophone
words; the difference between the two.
6.5. Discussion
6.5 Discussion
ɛ ɛː
ɛ∼ɛː
ɛː phonologically
any
ɛ∼ɛː
ERP responses to regional accent reect two distinct processes of perceptual
compensation
ɛː
ɛ∼ɛː
ɛː
ɛi l
6.6. Conclusion
×
6.6 Conclusion
pace
ɛ
ɛː
pace
ERP responses to regional accent reect two distinct processes of perceptual
compensation
Data availability
Ethics statement
Author contributions
Funding
Watching Dutch Change
Acknowledgments
Supplementary Material
7.1 The Polder shift and its adoption
what is the synchronic diatopic diusion of the sound changes in-
volved in the Polder shift?
eːøːoː ɛiœyɔu
l
l
(how) do sociolinguistic migrants adopt
the Polder shift?
7.1. The Polder shift and its adoption
r ɹ
which individuals, after how much time, are more
likely to adopt the Polder shift?
type
(how) is the adoption of the Polder shift reected in ERPs?
Conclusion
eː∼ei
ɹ
ɛ ɛː
ɛ
allophones
7.2 From compensation to adoption
7.2. From compensation to adoption
eː∼ei
ɹ
Conclusion
Table 7.1: Log10 Bayes factors for the hypothesis that the P600 difference between
the groups equals or exceeds that found in the [E∼E:] condition from
Chapter 6. Negative values indicate evidence against this hypothesis.
Condition Bayes factor (log )
[ei∼e:] −1.93
[øy∼ø:] −1.00
[ou∼o:] −2.18
[Ei∼E:] −2.03
[œy∼œ:] −1.57
[Au∼A:] −3.10
[a:ô∼a:ö] −2.62
[E∼E:] 0.00
ɛ ɛː
did not
ɛ∼ɛː brms
ɛ∼ɛː
µ ɛ∼ɛː
no
ɛ∼ɛː
7.3. Salience
ɛ∼ɛː
7.3 Salience
salience
eː∼ei
ɛ ɛː
l
sociolinguistic
əɹ
r
phonological
Conclusion
eː∼ei
salience
r ʀ
ɛ ɛː
ɛ ɛː
for them
7.4 Methodology
7.4. Methodology
new
l
l
a priori
β
a posteri-
ori
b
together
degree
Conclusion
× × ×
×
× × ×
against
p(β|y) p p(y|β)
a priori
7.5. Conclusions
7.5 Conclusions
l
l
does
Conclusion
minimally
Appendix A
Prime–target list for Experiment 1 from Chapter 3
Appendix A
Appendix B
Word list for Experiment 2 from Chapter 3
Appendix C
Word list for Experiment 1 from Chapter 4
Appendix C
Appendix D
/œy/×non-/l/
/e:/×non-/l/×HF
/o:/×non-/l/×HF
/œy/×non-/l/×HF
non-/l/
non-/l/×HF
/e:/×non-/l/
/o:/×non-/l/
Intercept
/e:/
/o:/
/œy/
−0.8
−0.4
0.0
0.4
0.8
−0.8
−0.4
0.0
0.4
0.8
−0.8
−0.4
0.0
0.4
0.8
Participant
∆F1 (Lobanov)
Cluster
1
2
Figure D.1: Separate BLUPs for the each of the random effects present in the by-individuals (n= 106) model for the production
data, with separate cluster analyses for each panel. As in Figure 4.3, each panel is separated into three panes; the
left pane shows the participants from the Ghent group, the middle pane shows the participants from the migrant
group, and the right pane shows the participants from the Leiden group.
Appendix E
Prime–target list for Experiment 2 from Chapter 4
Appendix E
Appendix F
Factor Estimate (SE) Odds ratio z p Sig.
Model = [e:∼
∼
∼Ei]
Intercept −1.46 (0.10) 1 : 4.30 −14.49 <.001 ∗∗∗
Step (Linear) 1.24 (0.13) 3.47 :1 9.87 <.001 ∗∗∗
Step (Quadratic) 0.33 (0.08) 1.39 :1 4.22 <.001 ∗∗∗
Step (Cubic) 0.03 (0.08) 1.03 :1 0.34 .74
Following segment = /l/ 0.83 (0.11) 2.29 : 1 7.59 <.001 ∗∗∗
Group = Migrant–Ghent −0.51 (0.13) 1 :1.66 −3.95 <.001 ∗∗∗
Group = Leiden–Others 0.07 (0.06) 1.07 :1 1.06 .29
Step (Linear) ×/l/ 1.68 (0.21) 5.37 :1 8.14 <.001 ∗∗∗
Step (Quadratic) ×/l/ −0.15 (0.14) 1 : 1.16 −1.07 .29
Step (Cubic) ×/l/ −0.13 (0.14) 1 :1.14 −0.92 .36
Step (Linear) ×Migrant–Ghent 0.63 (0.15) 1.87 :1 4.06 <.001 ∗∗∗
Step (Quadratic) ×Migrant–Ghent −0.04 (0.08) 1 :1.04 −0.47 .64
Step (Cubic) ×Migrant–Ghent 0.05 (0.07) 1.05 :1 0.66 .51
Step (Linear) ×Leiden–Others 0.01 (0.07) 1.01 :1 0.17 .87
Step (Quadratic) ×Leiden–Others 0.04 (0.04) 1.04 :1 1.11 .27
Step (Cubic) ×Leiden–Others −0.05 (0.03) 1 :1.05 −1.49 .14
Following segment = /l/×Migrant–Ghent 0.26 (0.13) 1.30 :1 2.05 .04
Following segment = /l/×Leiden–Others 0.08 (0.06) 1.08 : 1 1.31 .19
Model = [o:∼
∼
∼Au]
Intercept 0.20 (0.07) 1.22 :1 2.76 .01 ∗
Step (Linear) 1.73 (0.15) 5.65 :1 11.48 <.001 ∗∗∗
Step (Quadratic) 0.45 (0.11) 1.57 :1 4.17 <.001 ∗∗∗
Step (Cubic) −0.03 (0.11) 1 :1.03 −0.28 .78
Full results of Experiment 2 from Chapter 4
Factor Estimate (SE) Odds ratio z p Sig.
Following segment = /l/ 0.95 (0.12) 2.59 : 1 7.83 <.001 ∗∗∗
Step (Linear) ×/l/ −0.66 (0.25) 1 :1.93 −2.67 .01 ∗
Step (Quadratic) ×/l/ −0.74 (0.22) 1 : 2.09 −3.32 <.001 ∗∗
Step (Cubic) ×/l/ −0.19 (0.22) 1 :1.21 −0.88 .38
Model = [E∼
∼
∼Ei]
Intercept −0.94 (0.07) 1 : 2.55 −13.12 <.001 ∗∗∗
Step (Linear) 2.49 (0.15) 12.06 :1 16.44 <.001 ∗∗∗
Step (Quadratic) 0.27 (0.10) 1.31 :1 2.59 .01 ∗
Step (Cubic) −0.74 (0.11) 1 :2.09 −6.99 <.001 ∗∗∗
Following segment = /l/ 0.96 (0.12) 2.60 : 1 7.90 <.001 ∗∗∗
Group = Migrant–Ghent −0.21 (0.08) 1 :1.23 −2.64 .01 ∗
Group = Leiden–Others 0.09 (0.04) 1.10 :1 2.50 .01 ∗
Step (Linear) ×/l/ −1.38 (0.24) 1 :3.98 −5.84 <.001 ∗∗∗
Step (Quadratic) ×/l/ 0.25 (0.21) 1.29 :1 1.22 .22
Step (Cubic) ×/l/ 0.87 (0.21) 2.38 : 1 4.13 <.001 ∗∗∗
Step (Linear) ×Migrant–Ghent 0.43 (0.17) 1.53 :1 2.52 .01 ∗
Step (Quadratic) ×Migrant–Ghent 0.09 (0.09) 1.09 : 1 0.99 .32
Step (Cubic) ×Migrant–Ghent −0.20 (0.09) 1 :1.22 −2.25 .02
Step (Linear) ×Leiden–Others −0.23 (0.08) 1 :1.26 −2.91 <.01 ∗
Step (Quadratic) ×Leiden–Others −0.03 (0.04) 1 :1.03 −0.67 .50
Step (Cubic) ×Leiden–Others 0.04 (0.04) 1.04 :1 1.08 .28
Following segment = /l/×Migrant–Ghent 0.11 (0.12) 1.12 :1 0.92 .36
Following segment = /l/×Leiden–Others 0.01 (0.06) 1.01 : 1 0.15 .88
Step (Linear) ×/l/×Migrant–Ghent −0.24 (0.23) 1 :1.27 −1.04 .30
Appendix F
Factor Estimate (SE) Odds ratio z p Sig.
Step (Quadratic) ×/l/×Migrant–Ghent 0.28 (0.18) 1.32 :1 1.57 .12
Step (Cubic) ×/l/×Migrant–Ghent 0.39 (0.18) 1.47 :1 2.15 .03
Step (Linear) ×/l/×Leiden–Others −0.31 (0.11) 1 :1.36 −2.95 <.01 ∗∗
Step (Quadratic) ×/l/×Leiden–Others −0.18 (0.08) 1 :1.19 −2.26 .02
Step (Cubic) ×/l/×Leiden–Others −0.09 (0.08) 1 :1.10 −1.14 .25
Vowel normalization: A perceptual-acoustic study of Dutch
vowels
The Jour-
nal of the Acoustical Society of America116
The Journal of the Acoustical Society of America121
The Journal of the
Acoustical Society of America116
2nd International
Symposium on Information Theory, Tsahkadsor, Armenia, USSR, Septem-
ber 2–8, 1971
Perception & Psychophysics62
Proceedings of the 17th International Congress of Phonetic
Sciences (ICPhS), Hong Kong, 17–21 August
Jour-
Bibliography
nal of Sociolinguistics2
Dialect Change: Convergence and Di-
vergence in European Languages
Journal of Memory and Language94
Journal of Memory
and Language59
Linguistic Data Consortium
Proceedings of the
workshop on sound change actuation, University of Chicago
Lan-
guage Variation and Change23
International Journal of Psychophysiology43
Dynamische Syntactische Atlas van de Nederlandse Di-
alecten (DynaSAND)
Journal of
Memory and Language68
arXiv preprint arXiv:1506.04967
Journal of Statistical Software67
Journal of Phonetics13
Bibliography
Phonology28
Language
Proceedings of the 18th International Congress of Phonetic Sciences (ICPhS
XVIII)
Lan-
guage94
The
Cambridge handbook of phonology
The
Oxford handbook of historical phonology
Linguistics in the
Netherlands29
The acquisition of syntactic knowledge
NRC Handelsblad,
Evolutionary phonology: The emergence of sound patterns
Bidirectional Optimality Theory180
Phonology25
Praat: Doing phonetics by
computer
The phonology of Dutch
18th International Congress of Phonetic Sciences
(ICPhS 2015)
Bibliography
Laboratory Phonology3
The eect of geographic mobility on the retention of a local dialect
The R Journal9
Journal of Statistical Software80
The R Journal10
Language Variation and Change
14
Papers from the Fifth International Conference on Historical Linguis-
tics
University of Pennsylvania Working
Papers in Linguistics13
Language Spread and Language Policy: Issues, Impli-
cations, and Case Studies
Language68
Journal of Verbal Learning and Verbal
Behavior12
Journal of Phonetics66
The phonetics of English and Dutch
Bibliography
Brain Research1394
eegUtils: A collection of utilities for EEG analysis
Native listening
Interspeech 2008
Brain Research1084
University of Pennsylvania Working Papers in Lin-
guistics12
Fonologische atlas van de Neder-
landse dialecten. Deel IV: De consonanten
Language
(de)standardisation in Late Modern Europe: Experimental studies
University of Pennsylvania Working Papers in Lin-
guistics17
Journal of Psychophysiology21
Language and Speech52
Language Variation and Change24
Bibliography
Discourse Processes
The Journal of the Acous-
tical Society of America121
Journal of the
Royal Statistical Society: Series B (Methodological)17
Statistical methods and scientic inference
Journal
of Phonetics15
Journal
of Second Language Studies2
Journal of Psycholinguistic
Research38
Journal of Experimental Psychology:
Human Perception and Performance32
Behavior Research Methods50
Cognitive Brain Research1
Journal of Experimental Psychology: Human Perception and Performance
6
Cognition89
Dutch studies 4: Studies in Dutch
Bibliography
phonology
Psychophysiology27
The Journal of Socio-Economics33
Contexts of accommodation: Developments in applied so-
ciolinguistics
Language and social psy-
chology
Language in
Society2
Fonologische atlas van de Ned-
erlandse dialecten. Deel I: Het korte vocalisme
Australian Journal of Linguistics21
Brain and Language122
Journal of Germanic Linguistics23
Dutch Journal of Applied
Linguistics1
The initiation of sound change: Percep-
tion, production, and social factors
Bibliography
Phonetica55
Journal of Lin-
guistics29
Handbook of the International Phonetic Association: A guide to the use of the
International Phonetic Alphabet
Journal of the
International Phonetic Association29
The pronunciation of English: A course
for Dutch learners
Sociolinguistic variation in Cairene Arabic: Palatalization and the
Qaf in the speech of men and women
Journal of Statistical Software59
Phonology in perception
Journal of Phonetics34
Laboratory Phonology9
The Journal of the Acoustical Society
of America123
Journal of the International Phonetic As-
sociation30
A gure
of speech: A festschrift for John Laver
Bibliography
The vowel space paradox: A sociophonetic study on Dutch
Language Variation and
Change26
Journal of the American Statistical Association82
Linguistics48
Annals of Mathemat-
ical Statistics21
Brain and Language152
Dialects across borders
Research in Language13
Artikelen van de derde sociolinguïstische conferentie
Linguistic studies
oered to Joseph Greenberg on the occasion of his sixtieth birthday
Origins of sound change: Approaches to phonologization
On variation and change in diphthongs and long vowels of spoken
Dutch
Preattentive phonotactic processing
Bibliography
Language
Theory of probability
University of Pennsylvania Work-
ing Papers in Linguistics13
Annual Review of Psychology68
Minima of functions of several variables with inequalities as side
constraints
IEEE International Conference
on Acoustics, Speech and Signal Processing
Proceedings of the National Academy of
Sciences103
Biometrics
Behavior Research
Methods42
The Journal of the Acoustical Society of
America97
Language and Speech55
Frontiers in Psychology1
WIREs Cognitive Science1
Bibliography
Trends in Cognitive Sciences14
Psychonomic Bulletin & Review25
Proceedings of the National Academy of Sciences100
Proceedings of the second Berkeley symposium on mathematical statis-
tics and probability
Biological Psychology74
Neuropsychologia53
Journal of Statistical Software
82
Principles of linguistic change. Volume I: Internal factors
Principles of linguistic change. Volume II: Social factors
Language83
Language
A quantitative study of sound change
in progress
Journal of Phonetics38
Bibliography
Frontiers in
Psychology7
Cognition176
Behavioral and Brain Sciences22
Rivista di linguistica7
Experimental Brain
Research212
Frontiers in
Psychology6
Frontiers in Neuroscience11
Language, Cognition
and Neuroscience
Perception & Psychophysics28
Journal of Neuroscience Methods164
Nature244
Journal of Memory and
Language94
Cognitive Science32
Mem-
ory & Cognition21
Cognitive Science30
Bibliography
Journal of the International Phonetic Association
13
Historical Linguis-
tics: Problems and Perspectives215
Perception & Psy-
chophysics65
Behavioral and Brain Sciences13
Audiology and Neurotology2
Clinical Neurophysiology118
Language Variation and Change16
First course in probability and statistics
Revue de l’Institut International de Statistique
Cognitive Psychology47
Second dialect acquisition: Implications for theories of phonological
representation
Lan-
guage and Linguistics Compass9
Journal of Pragmatics52
Psychological Science5
Bibliography
Proceedings of the Chicago Lin-
guistics Society 17: Papers from the parasession on language and behaviour
Language change: Contributions to
the study of its causes
The initiation of sound change: Perception, pro-
duction, and social factors
Phonological projection: A theory of feature content
and prosodic structure
Onze Taal2
Journal of Memory and Language31
Morphologische Untersuchungen auf dem
Gebiete der indogermanischen Sprachen
Journal of
Experimental Psychology: Learning, Memory, and Cognition19
The Journal of the Acoustical Society of America119
Language and Linguistics Compass
6
Journal of Phonetics40
Etude sur les changements phonétiques et leurs caractères généraux
Event-related potential evidence of abstract phonological learning in the
laboratory
Bibliography
Diphthong dynamics: A cross-linguistic perceptual analysis
of temporal patterns in Dutch, English, and German
Journal of
Neuroscience Methods162
Psychophysiol-
ogy37
Frequency and the emergence of
linguistic structure
The actuation of sound change
Language and Speech
International Journal of the Sociology of Language
E-Prime
R: A language and environment for statistical computing
Salience in sociolinguistics: A quantitative approach
Language and Linguistics Compass9(1)
Taal en Tongval52
Neuroimage12
Bibliography
Psychological science29
Tapping into semantic recovery: An event-related potential
study on the processing of gapping and stripping
Sociolinguistic variation: Critical reections
Language83
→ʀ
‘r-atics: Sociolinguistic, phonetic and phonological characteristics of
/r/
Language and Speech
53
NeuroImage128
The
handbook of historical phonology
The Annals of Statis-
tics6
The R Journal8
The sociophonetics and phonology of Dutch /r/
Proceedings of the 18th International Congress of Phon-
etic Sciences (ICPhS XVIII)
Second dialect acquisition
Bibliography
Mechanisms of phonological change
Language93
International Journal of Psychophysiology77
Journal of Cogni-
tive Neuroscience22
Psy-
chophysiology48
Loquens1
Onze Taal6
Poldernederlands. Waardoor het ABN verdwijnt
Brain Topography27
Lan-
guage Variation and Change24
Language variation — European perspectives VIII: Selected papers from the
tenth International Conference on Language Variation in Europe (ICLaVE
10), Leeuwarden, June 2019
Linguistics Vanguard2
Ken mèn ’t rotte hoe of jenut spelt!! Een vergelijking van de
orthograsche diepte van het Standaardnederlands en het Haagse stadsdialect
Bibliography
TESOL Quarterly46
Dialects in contact
English World-wide9
Urban voices: Accent studies in the British Isles
Variatie en verandering in het gesproken Standaard-
Nederlands
Multilingua3-4
Journal of Sociolinguistics1
Ar-
tikelen van de vierde sociolinguïstische conferentie
Samen beleid maken
in het GO! Onderwijs van de Vlaamse Gemeenschap. Beleidscahier3
Journal of the International
Phonetic Association35
The interaction between the segmental and the prosodic phono-
logy in the midst of an on-going sound change: Resolving a contradiction in
the synchronic phonology of Dutch
permutes: Permutation tests for time series data
buildmer: Stepwise elimination and term reordering for mixed-
eects regression
buildmer: Stepwise elimination and term reordering for mixed-
eects regression
Bibliography
permutes: Permutation tests for time series data
buildmer: Stepwise elimination and term reordering for mixed-
eects regression
Language and Speech
Frontiers in Neuro-
science13
Regionale variatie in het taalgebruik van notabelen. Een soci-
olinguïstisch onderzoek in Middelburg, Roermond en Zutphen
p Psychonomic Bulletin & Review14
Linguistic change and stabilization in the transition from ado-
lescence to adulthood
Language and Lin-
guistics Compass6
Crossing oceans with voices and ears: Second dialect acquisition
and topic-based shifting in production and perception
Laboratory Phonology: Journal of the Association for Laboratory Phono-
logy10
Language
Journal of Speech and Hearing Research4
Lan-
guage Variation and Change12
Di-
rections for historical linguistics
Bibliography
Journal of Memory and Language34
Language and Speech58
Psychonomic bulletin & review21
Generalized additive models: An introduction with R
Phonological structure and phonetic form: Papers in laboratory phonology3
Ms.
The HTK book
PloS One5
Origins of sound
change: Approaches to phonologization
Language, Cognition and Neuroscience
32
The relationship between perception and production in adult
acquisition of a new dialect’s phonetic system
Leerboek der Phonetiek. Inzon-
derheid met betrekking tot het Standaard-Nederlands
eːøːoː ɛiœyɔu
l
l
Samenvatting in het Nederlands
exemplars
l
sociolinguïstische migranten
Samenvatting in het Nederlands
perceptie representatie
r
r
ʀ
ɹ
r
interpretatie perceptie
on-line
Samenvatting in het Nederlands
kleinere
ɛ ɛː
fonemische
