A LONGITUDINAL STUDY OF ADULT FOREIGN...
Transcript of A LONGITUDINAL STUDY OF ADULT FOREIGN...
“ha-ħa!”A LONGITUDINAL STUDY OF ADULT FOREIGN LANGUAGE LEARNING
Bias in Auditory PerceptionAarhus, Sept 19, 2014
Andreas Højlund Nielsen, PhD student, CFIN/Linguistics/IMC (AU)
NATIVE DANISH LANGUAGE OFFICER CADETSLEARNING EITHER ARABIC OR DARI
WHY
Foreign speech learning in the lab: 10-30 days(e.g. Bradlow et al., 1997; Chandrasekaran et al., 2010; Lively et al., 1994; Logan et al., 1991;
Perrachione et al., 2011; Wang et al., 1999; Wong & Perrachione, 2007)
Models of foreign language speech learning: perceptual adaption relatively early in the learning process
(Best, 1995; Best & Tyler, 2007; Flege, 1995)
But how about “IRL”?
WHY
Also, immersion studies: differences between1.5-7 yrs(Flege, Bohn, & Jang, 1997; Flege & Liu, 2001)
Can we show adaptation to foreign speech sounds during intensive foreign learning?
And if so, how does this unfold in time?
20 language officer cadets in total
learning either
Dari (Afghan Farsi): n = 12or
Arabic (al-misr/MSA): n = 8
PARTICIPANTS
Highly select group
Very intensive training: 8 hr school, 8 hr homework, 8 hr sleep
18 mos of training: high level of proficiency (“fluent” speakers)
PARTICIPANTS
STIMULI
DARI (fricative voicing)“near common”: [ʃa] (“sh”)
Dari-only: [ʒa] (“zh”) (11 and 3 levels)
ARABIC (emphasis - laryngeal fricative)“near common”: [ha] (“ha”)
Arabic-only: [ħa] (“hha”) (11 and 3 levels)
PARADIGMS
IDENTIFICATION TASK:2AFC (incl. short exposure)
DISCRIMINATION TASK:AX (same-different)
TIME PLAN
T0: Nov 12-15, 2012
T1 (3 wks): Dec 8-10, 2012
T2 (6 mos/24 wks): May 4-6, 2013
T3 (18 mos/81 wks): June 2-4 2014
HYPOTHESES
ID-slopes of Dari learners to Dari stimuli: STEEPENING
ID-slopes of Arabic learners to Arabic stimuli: STEEPENING
ID-slopes of all learners to non-target stimuli: no change
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as z
ha
dari−learners’ ID functions | shazha | T0−T1−T2−T3
T0T1T2T3
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as h
ha
dari−learners’ ID functions | hahha | T0−T1−T2−T3
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as z
ha
arab−learners’ ID functions | shazha | T0−T1−T2−T3
T0T1T2T3
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as h
ha
arab−learners’ ID functions | hahha | T0−T1−T2−T3
T0T1T2T3
T0 T1 T2 T3−0.5
0
0.5
1
1.5
Time of measurement
Slop
e of
ID fu
nctio
ns
arab (n = 8) & dari (n = 12) learners’ mean slopes || ha−hha & sha−zha || T0−T1−T2−T3
Arabic learners − Arabic stimuliDari learners − Arabic stimuliArabic learners − Dari stimuliDari learners − Dari stimuli
}} *
} *
Arabic learners (n = 8) Dari learners (n = 12)
* *
Arabic stimuli: /hæ/ vs. /ħæ/ Arabic stimuli: /hæ/ vs. /ħæ/
Dari stimuli: /ʃa/ vs. /ʒa/Dari stimuli: /ʃa/ vs. /ʒa/
Frication Frication
VoicingVoicing
% /ʒ
a/ id
entif
icat
ion
% /ʒ
a/ id
entif
icat
ion
% /ħ
æ/ i
dent
ifica
tion
% /ħ
æ/ i
dent
ifica
tion
Time
Slop
e of
ID fu
nctio
ns
*
0 wks 3 wks 24 wks 81 wks
*
RESULTS - IDENTIFICATION
RESULTS - ID
A native Arabic speakerArabic stimuli: /hæ/ vs. /ħæ/
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as h
ha
dari−learners’ ID functions | hahha | T0−T1−T2−T3
T0T1T2T3
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as h
ha
arab−learners’ ID functions | hahha | T0−T1−T2−T3
T0T1T2T3
Arabic learners (n = 8) Dari learners (n = 12)
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as /h
ha/
Identification, ha−hha (Arabic native, pilotbochra)
RESULTS - ID
A native Dari speakerDari stimuli: /ʃa/ vs. /ʒa/
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as z
ha
arab−learners’ ID functions | shazha | T0−T1−T2−T3
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as z
ha
dari−learners’ ID functions | shazha | T0−T1−T2−T3
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (D3)
Arabic learners (n = 8) Dari learners (n = 12)
RESULTS - ID
−2 −1.5 −1 −0.5 0 0.5 1 1.5 21
2
3
4
5
6
7
8
slope diff
spea
king
gra
de (r
anke
d)
arab learners T2 vs. T0
−2 −1.5 −1 −0.5 0 0.5 1 1.5 21
2
3
4
5
6
7
8
slope diff
spea
king
gra
de (r
anke
d)
arab learners T1 vs. T0
Arabic learners (n = 8)
T0 to T1
T0 to T2
−0.5 0 0.5 1 1.50
2
4
6
8
10
12
slope diff
spea
king
gra
de (r
anke
d)
dari learners T2 vs. T0
−0.5 0 0.5 1 1.50
2
4
6
8
10
12
slope diff
spea
king
gra
de (r
anke
d)
dari learners T1 vs. T0
Dari learners (n = 12)
T0 to T1
T0 to T2
rS(10)=0.60*
rS(10)=0.71*
DARI LEARNERS’SLOPE CHANGESWERE POSITIVELY
CORRELATEDWITH THEIR
SPEAKING GRADESp < 0.05
p < 0.05
RESULTS -
sha01−sha04 sha04−sha07 sha07−sha11−1
−0.5
0
0.5
1
1.5
2
Stimulus pair
d−pr
ime
(d’)
arab−learners’ dprimes (n = 8) || sha−zha || T0−T1−T2−T3
T0T1T2T3
ha00−h28 ha28−ha49 ha49−ha70−1
−0.5
0
0.5
1
1.5
2
Stimulus pair
d−pr
ime
(d’)
arab−learners’ dprimes (n = 8) || ha−hha || T0−T1−T2−T3
T0T1T2T3
Arabic learners (n = 8)
*
* }} }*
Stimulus pair
Stimulus pair
d-pr
ime
(d’)
d-pr
ime
(d’)
“Arabic” stimuli: /hæ/ vs. /ħæ/
“Dari” stimuli: /ʃa/ vs. /ʒa/
sha01−sha04 sha04−sha07 sha07−sha11−1
−0.5
0
0.5
1
1.5
2
Stimulus pair
d−pr
ime
(d’)
dari−learners’ dprimes (n = 12) || sha−zha || T0−T1−T2−T3
T0T1T2T3
ha00−h28 ha28−ha49 ha49−ha70−1
−0.5
0
0.5
1
1.5
2
Stimulus pair
d−pr
ime
(d’)
dari−learners’ dprimes (n = 12) || ha−hha || T0−T1−T2−T3
T0T1T2T3
Dari learners (n = 12)
* } * }
Stimulus pair
Stimulus pair
d-pr
ime
(d’)
d-pr
ime
(d’)
“Arabic” stimuli: /hæ/ vs. /ħæ/
“Dari” stimuli: /ʃa/ vs. /ʒa/
}*DISCRIM
DISCUSSION
Arabic contrast also includes RTR on subsequent high vowels
Voicing contrast in Arabic for the alveolar fricative (/s/ vs. /z/)
Effects from other L2s (e.g. English or French)
DISCUSSION
Large individual differences:learning biases in a very homogenous sample?
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as /h
ha/
Identification, ha−hha (arab1, T0−T1−T2−T3)
T0T1T2T3
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as /h
ha/
Identification, ha−hha (arab2, T0−T1−T2−T3)
T0T1T2T3
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as /h
ha/
Identification, ha−hha (arab3, T0−T1−T2−T3)
T0T1T2T3
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as /h
ha/
Identification, ha−hha (arab4, T0−T1−T2−T3)
T0T1T2T3
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as /h
ha/
Identification, ha−hha (arab5, T0−T1−T2−T3)
T0T1T2T3
ha00 ha07 ha14 ha21 ha28 ha35 ha42 ha49 ha56 ha63 ha700
10
20
30
40
50
60
70
80
90
100
Continuum from less to more aspirated (90 ms to 160 ms)
Perc
enta
ge id
entif
ied
as /h
ha/
Identification, ha−hha (arab6, T0−T1−T2−T3)
T0T1T2T3
Arab learners’ ID functions | ha-hha | T0 to T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (arab3, T0−T1−T2−T3)
T0T1T2T3
Arab learners’ ID functions | sha-zha | T0 to T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (arab1, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (arab2, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (arab3, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (arab4, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (arab5, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (arab6, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (arab3, T0−T1−T2−T3)
T0T1T2T3
Dari learners’ ID functions | sha-zha | T0 to T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (arab3, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (dari1, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (dari4, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (dari5, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (dari6, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (dari7, T0−T1−T2−T3)
T0T1T2T3
sha01 sha02 sha03 sha04 sha05 sha06 sha07 sha08 sha09 sha10 sha110
10
20
30
40
50
60
70
80
90
100
Continuum from voiceless to voiced fricative
Perc
enta
ge id
entif
ied
as /z
ha/
Identification, sha−zha (dari8, T0−T1−T2−T3)
T0T1T2T3
CONCLUSION
HYPOTHESES:
ID-slopes of Dari learners to Dari stimuli: STEEPENING
ID-slopes of Arabic learners to Arabic stimuli: STEEPENING
ID-slopes of all learners to non-target stimuli: no change
√
%
%
CONCLUSION
Early adaptation to the Dari voiced fricative (sh-zh) (for Dari learners)
No adaptations to the Arabic emphatic pharyngeal fricative (h-hh)
Large individual differences despite homogeneity
THANKS TOThe research team:Nynne Thorup HornStine Derdau SørensenSandra KostadinovaMikkel Wallentin
The language officers &The Institute for Languages at the
Royal Danish Defence College
As well as:Bill McGregorAziz & Mojhda RamikiGeorgetown University Press Chris BaileyMartin SnejbjergKit RigdenIvana KonvalinkaNicoline Munck Vinther
Seed funding support from the Interacting Minds Centre (IMC)