Upload
karsen
View
35
Download
0
Embed Size (px)
DESCRIPTION
One-to-multiple vowel mapping in the perceptual development of Dutch learners of Spanish. Paola Escudero McGill University and University of Utrecht Paul Boersma University of Amsterdam Boston University Conference on Language Development 26 November 3, 2001. - PowerPoint PPT Presentation
Citation preview
One-to-multiple vowel mapping in the perceptual development of Dutch learners of Spanish
Paola EscuderoMcGill University and University of Utrecht
Paul BoersmaUniversity of Amsterdam
Boston University Conference on Language Development 26 November 3, 2001
Single-Category Assimilation (SCA)
Problem:– poor category differentiation– no lexical contrast no produced contrast
General solution: category split
L2English
L1Dutch
//
/æ///
L2S.English
L1Spanish
/i/
/I//i/
L2English
L1Japanese
/l/
/ /̈/R/
Two-Category Assimilation (TCA)
Very good category differentiation Problem:
boundary mismatch confusion General solution: boundary shift
L2English
L1Dutch
/ph/
/b8/ /p/
/b/
L2Spanish
L1Dutch
/u/
/o/
/u/
//
Multiple-Category Assimilation (MCA)
Category differentiation is “too good” Problems:
extra lexical contrasts, boundary mismatch, SCA Possible solutions?: category split/merger/loss,
boundary shift
L2Spanish
L1Dutch
/i/
/e/
/i//I///
L2English
L1Polish
/s/
/S/
/s//þ//§/
Preliminary hypothesis
1) MCA exists and it is problematic.
L2Spanish
L1Dutch
/i/
/e/
/i//I///
L2Spanish
L1Dutch
/u/
/o/
/u/
//
2) Therefore, Dutch learners will have poorer front than back vowel categorisation in Spanish.
Subjects Dutch learners of Spanish 38 (11+18+9) Bilingual Dutch-Spanish 3
Dutch-only 11
Spanish-only 44
Click the Spanish vowel that you heard
I
E
A
O
U
Task: “vowels from a Spanish text”125 Spanish target stimuli, 55 Spanish fillers
Beginning Dutch learners of Spanish
0 5 10 15 20 250
5
10
15
20
25
e i errors
1
1
1
11
1
1 1
1
11
Intermediate Dutch learners of Spanish
0 5 10 15 20 250
5
10
15
20
25
e i errors
2
2
22
2 22
2
2
2
2
22
22
22
2
Advanced Dutch learners of Spanish
0 5 10 15 20 250
5
10
15
20
25
e i errors
33
33
333
3
3
Dutch/Spanish bilinguals
0 5 10 15 20 250
5
10
15
20
25
e i errors
4
4
4
Spanish-only listeners
0 5 10 15 20 250
5
10
15
20
25
e i errors
5
5
5
5
55 5
5
5
5 5
5 5
5
5
5 5
5
5
5555
5
5 5
555
5
5
5
5
5
5
55
5
5
5555
5
All subjects + Dutch-only
0 5 10 15 20 250
5
10
15
20
25
e i errors
0
00
0
0
0
00
00
0
1
1
1
11
1
1 1
1
11
2
2
22
2 22
2
2
2
2
22
22
22
233
33
333
3
34
4
4
5
5
5
5
55 5
5
5
5 5
5 5
5
5
5 5
5
5
5555
5
5 5
555
5
5
5
5
5
5
55
5
5
5555
5
Developmental sequencefor the front/back error ratio
BEG-INT-ADV error ratio: r = -0.32 (p < 0.03)
number of ei errorsnumber of ou errors
2.0
01.0
0.5
DUT BEG
INTADV
BILSPA
We have observed:
Dutch learners of Spanish perform poorer on front than on back vowels. Our prediction is borne out.
Learners have better performance according to their experience.
This hints at the existence of MCA, its problematic nature, and its reduction during L2 development. But we haven’t yet directly shown the existence of MCA, let alone its problematic nature…
Basic hypothesis
Multiple Category Assimilation exists.
Click the Dutch vowel that you heard
h IE k h EE k h I k h E k
h UU k h EU k h U k h AA k
h OE k h OO k h O k h A k
Task 1: “vowels from a Dutch text”125 Spanish target stimuli, 55 Dutch decoys
Dutch labelling task (Dutch mode)
Therefore: MCA exists
stim :
/i/
/e/
DUTresp :
i
I
BEGresp :
i
I
INTresp :
i
I
ADVresp :
i
I
BILresp :
i
I
Main hypothesis
Problems with Spanish front vowel categorisation by Dutch learners are caused by MCA.
To address this hypothesis, we have to look into:1) the accuracy of L2 front vowel categorisation
2) the reduction of MCA when perceiving the L2
Hypothesis for MCA reduction(Assumption: people have language-dependent
perception modes)
Hypothesis: learners show less MCA in their L2 perception mode than in their L1 perception mode.
A measure of MCA:the amount of use of // in an L1 labelling
task. A measure of MCA reduction:
reduction of // use betweena Dutch and a Spanish perception mode.
Click the Dutch vowel that you heard
h IE k h EE k h I k h E k
h UU k h EU k h U k h AA k
h OE k h OO k h O k h A k
Task 2: “vowels from a Spanish text”125 Spanish target stimuli, 55 Spanish fillers
Dutch-only listeners
MCA-reductionis small ornon-existent
stim :
/i/
/e/
"Dutch" moderesp :
i
I
stim :
/i/
/e/
"Spanish" moderesp :
i
I
Beginning Dutch learners of Spanish
MCA-reductionis small ornon-existent
stim :
/i/
/e/
"Dutch" moderesp :
i
I
stim :
/i/
/e/
"Spanish" moderesp :
i
I
Intermediate Dutch learners of Spanish
MCA-reductionis intermediate
stim :
/i/
/e/
"Dutch" moderesp :
i
I
stim :
/i/
/e/
"Spanish" moderesp :
i
I
Advanced Dutch learners of Spanish
MCA-reductionis large
stim :
/i/
/e/
"Dutch" moderesp :
i
I
stim :
/i/
/e/
"Spanish" moderesp :
i
I
Dutch/Spanish bilinguals
MCA-reductionis large
stim :
/i/
/e/
"Dutch" moderesp :
i
I
stim :
/i/
/e/
"Spanish" moderesp :
i
I
Does MCA reduction exist?
Yes . It correlates with experience level (BEG/INT/ADV): r = 0.39, (p < 0.01, 95% = 0.08...0.63)
Is it developmental?
Yes . Learners reduce the usage of // between perception modes.
Hypothesis for L2 categorisation accuracy
Learners have inaccurate front vowelcategorisation.
A measure of vowel categorisation accuracy:boundary location in an L2 labelling task.
Click the Spanish vowel that you heard
I
E
A
O
U
Task 3: “vowels from a Spanish text”125 Spanish target stimuli, 55 Spanish fillers
Spanish labelling task
stim :
/i/
/e/
BEGresp :
i
e
INTresp :
i
e
ADVresp :
i
e
BILresp :
i
e
SPAresp :
i
e
Is L2 labelling inaccurate?
Yes . It correlates with experience level (BEG/INT/ADV): r = 0.57, (p < 0.0002, 95% = 0.30...0.75)
Does this change developmentally?
Yes . Learners are inaccurate in L2 categorisation, i.e. the boundary between /e/ and // is inappropriate for Spanish.
Main hypothesis again
These problems with Spanish front vowelcategorisation by Dutch learners are caused byMCA.
MCA reduction correlates strongly with categorisation accuracy
r = 0.62
p < 0.00002
95% = 0.37...0.78
-5 0 5 100
5
10
15
20
L1 /I/-reduction (Exp1 minus Exp2)
1
1 1
11
1
1
1
1
11
2
2
2
22
2
22
2
2
22
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
Our interpretationof this high correlation:
It reflects a causal relationship:
low-level // losscauses
e-i accuracy in learners
Conclusion1) MCA exists: learners use 3 categories where the target
language has 2.
2) MCA is problematic: it leads to inaccurate L2 categorisation.
3) Language-dependent perception modes exist: people do different things depending on the language that they hear (or think they hear).
4) Learners develop: with experience, MCA decreases and categorisation accuracy increases.
5) Learners improve: we identified one strategy for solving MCA, namely loss of the extra category.
Dank u voor uw aandacht!Gracias por su atención!
Thank you for your attention!
Total errors are not a sign of MCA degree
//-reduction vs. /e/-// boundary: r = -0.60 (95% = -0.34 … -0.77)
//-reduction vs. total error count: r = +0.15 (95% = -0.17 ... +0.45)