Transcript Inflectional Verb Errors in the Acquisition of Russian by Bilingual

Language Acquisition and Bilingualism:
Consequences for a Multilingual Society
Toronto, May 4-7, 2006
* This research was partially supported by the Israel Science
Foundation (Grant No. 4806)
Inflectional Verb Errors in the Acquisition of Russian
by Bilingual and Monolingual Children
Sharon Armon -Lotem, Bar Ilan University, Natalia Gagarina, ZAS, Berlin, Olga Gupol, Bar Ilan University, Israel
For more information, please write to: [email protected]
Method – bilinguals
 Research Objectives
 Method – monolinguals •
To compare the erroneous use of verbal inflections in the early verb
development of Russian, in Russian monolinguals and Russian-Hebrew
sequential bilinguals.
• Three L1 Russian-speaking children
Vanja (V.), Vitja (Vi.), Liza (L.)
• The mean length of recordings per month 2.5 hours
V. 1;5 – 4;5 ... [analyzed utterances ~32000]
Vi. 2;0 – 2;10 ... [analyzed utterances ~4500]
L. 1;2 – 3;0 [analyzed utterances ~5000]
• Excluded: frozen forms, immediate repetitions, (self-repetitions), citations, yesno sentences, exclamations...
 Research Questions
1. What are the types of erroneous use of verb inflection?
2. Is there a relation of erroneous use to the general development of verb
grammar in children measured in:
 MLU
 ratio of verb utterances over all utterances (VU%)
 the productivity of verb inflection
3. Do bilingual errors change over time?
4. Are the errors in the bilingual production evidence for second
V.
language influence on the acquisition process of the first language?
Vi.
 Predictions
L.
1. We expect to see the common monolingual errors in both groups
norms of linguistic monolingual behaviour.
3. Second language influence on the acquisition process of the first
language will manifest itself in contrastive structures.
4. Common monolingual errors will reduce over time, while unique
bilingual difficulties will increase.
Russian Verbal System
4,6
32,1
36,0
16
99
120
2;1
2;34
2;10
VU
(%)
10,0
30,7
47,3
10
52
120
1.764
1.974
3.400
1;8
1;11
2;10
as well as unique bilingual difficulties.
2. Bilingual verb development will show delay with respect to the
Age
MLU
•
onset of the V prod.
•
onset/development
of productivity
target-like use
•
1.332
2;1
2;3
2.153
2;10
2.758
1.017
1.199
3.054
11,1
22,5
41,4
Verb
lemmas
Child
Age
Galit
Eldar
• NumberSG, PL
• Gender Masc, Fem, Neut
• Person 1st, 2nd, and 3rd
• About 50 inflectional microclasses (i.e. the smallest subset of an
inflectional class above the paradigm, definable as the set of
paradigms which share exactly the same morphological
generalizations, but may differ via the application of
phonological processes (Dressler and Gagarina 1999)
• The 1st productive microclass:
two stems/bases: igra-t’ (OB-inf) – igraj-u (CB-1s)
s’’igrat’ –Pf
Present (Ipf) / Future (Pf)
SG
SG
1.8
35
10
4
3
3.4
167
57
Past-Gender
Past-Number
Lital
3,9
2
2.9
132
32
igraj-em
Fem.
igra-l-a
2 igraj-eš
igraj-ete
Masc.
igra-l
Michael
3,6
1,11
3.1
159
50
Zhenya
3,6
9 months
3,1
176
59
Tali
4
9 months
1,3
113
70
Leya
4,3
9 months
4,8
244
72
Patricia
5
9 months
2
140
48.6
33
Present- Number
G
t
ali
ld
E
ar
Lit
al
a
el
ny
ha
ic
he
Z
M
T
ali
a
y
Le
atr
P
ia
ic
•Gender in the past is the greatest source of errors
Bilinguals II
Past-Gender
Past-Number
Present -Person
Present-Number
June
March
June
March
Eldar
Eldar
Leya
Leya
•Gender in the past is still the greatest source of errors
•Person in the Present is a source of errors only for the early BL
•Number errors are evident for all children
•Person in the Present is a source of errors only for the early BL
34
•Number errors are evident for both children
35
Aspectual errors
 Bilingual subjects – sample II
Child
Age
Length of
exposure to L2
MLU
For the whole data
Verb
Utterances
VU%
Eldar I
Eldar II
4
4;9
3
3;9
3.4
2.01
167
193
57
79
Leya I
Leya II
4;3
5
9 months
1;6
4.8
5.9
244
210
72
87
Bilinguals
Child
Galit
Eldar
Lital
Michael
Zhenya
Tali
Leya
Patricia
Total
• Eldar is an early bi lingual and Leya i s a late bilingual
• There is a decrease in Eldar’ s ML U from ti me I to time II, but an
increase in the percentage of his verbal utterances out of all u tterances
(VU% )
19
Pf for Ipf
0
8
5
3
1
1
3
1
22
Ipf for Pf
1
10
9
2
10
7
10
9
58
•Monolingual children use perfective (Pf) for imperfective (Ipf) in the
analytical constructions
Comparability of data
Categories of analysis
For the general comparison of linguistic development, we computed MLU:
• one is comparable to onset of verb production
• two are comparable to onset of productivity
• five of the bilingual children can be compared to the monolingual
children at age 2;10
• In the second samples, the early bilingual is comparable to onset of
productivity, and the late bilingual can still be compared to the
monolingual children at age 2;10
•
•
In order to establish a comparable level of the morphological
development we computed the VU%:
• seven of the bilinguals are comparable to 2;10 and older. This also
holds for the later samples.
• only one child, who is a simultaneous bilingual, is comparable with
the onset of verb production
•Bilingual children: errors in both directions, with more imperfective
36
(Ipf) for perfective (Pf)
The number and percentage of
• verbal utterances of all utterances (VU%)
• morphological errors (for these utterances)
The erroneous uses of the verbs
1. Use of the wrong form in the context:
• Root infinitives
• Contextually infelicitous tense
• Luck of subject-verb agreement in person, number and gender
2. Wrong use of aspect
3. Use of the wrong pattern for the stem shift
Bilinguals II
Child
Eldar I
Eldar II
Leya I
Leya II
Pf for ipf
8
19
3
5
Ipf for pf
10
27
10
20
Bilingual errors continue to be in both directions, with more imperfective
for perfective
 Summary of Errors
37
igraj-ut
Neut.
igra-l-o
General verb production
Monolinguals
Typologically different stem vs. root system
Different gender system:
No neutral gender in Hebrew
Different inflections in present/future tense
Different inflections in past tense
Different tense categorization
Presence/absence of aspect
Age
V. Tokens
V. errors: Tokens
V. errors:
Tokens %
Child
Age
V tokens
V errors
(Tokens)
V.
2;1
2;3
2;10
56
460
309
9
7
9
16,1
1,5
2,9
Galit
3,5
35
3
V errors %
With respect to
the correct verbal
uses
8,6
Eldar
4
237
82
34,6
Vi.
2;1
2;3
2;10
33
129
299
9
9
5
27,3
7,0
1,7
Lital
3,9
180
28
15,6
Michael
3,6
194
27
13,9
Zhenya
3,6
244
41
16,8
L.
1;8
1;11
2;10
13
69
361
3
6
12
23,1
8,7
3,3
Tali
4
120
54
45
Leya
4,3
365
106
29
Patricia
5
162
43
26,5
22
Tense
RI
Agreement
Aspect
Stem errors
These errors are more typical of younger monolinguals. They are evident in the
bilingual group after the age of 3;6. Error types 1-3 reduce over time.
Liza
Verb onset: L.
Productivity
Two kinds of errors are unique to the bilingual group, and increase over time:
1. Wrong aspect (Imperfective for Perfective)
2. Wrong tense
Adult VU%
Vitja
23
Bilinguals II
Verb onset: L.
Productivity
Adult VU%
•At verb onset, the majority of errors are Root Infinitives (RIs).
Child
Age
V
tokens
V errors
(Tokens)
V errors %
With respect to the
correct verbal uses
•When morphology becomes productive, agreement errors become more
frequent.
Eldar I
Eldar II
4
4,9
237
219
82
98
34.6
44.7
•When adultlike VU% is found, RI disappear while stem errors gradually
become the main source of errors.
Leya I
Leya II
4,3
5,0
365
335
106
111
29
33.1
Monolingual verb production
• Onset of verb production: error rate of 16-27%
• Onset of inflectional productivity: strong reduction of errors
(under 9%)
• By the age of 2;10: further reduction of errors (ca. 2-3%)
Bilingual verb production in Russian
• Error rate of all children is higher (14%-45%), even when:
• MLU matches monolinguals at 2;10
• VU% is higher than monolinguals at 2;10
• exposure to Hebrew less than a year
• Error rate increases over time
Bilinguals
24
• Errors found in both populations may indicate a delay (for the early bilinguals)
and attrition (for the late bilinguals)
• Errors which are unique for bilinguals may suggest L2 influence.
• The disappearance of the monolingual errors over time supports an analysis in
terms of L2 influence.
• Errors found in both populations may indicate a delay (for the early bilinguals)
and attrition (for the late bilinguals)
• Errors which are unique for bilinguals may suggest L2 influence.
• The disappearance of the monolingual errors over time supports an analysis in
terms of L2 influence.
Bilinguals II
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Tense
RI
Agreement
Aspect
Stem errors
•Root Infinitives (RIs) are found, but are not the major source of errors
•Tense and aspect errors form a prominent share of the errors. Tense
errors are unique to the bilingual children
•The error pattern is clearly different from that of the monolinguals
 Conclusions
27
Tense
RI
Agreement
Aspect
Stem mistakes
June
March
June
March
Eldar
Eldar
Leya
Leya
•Root Infinitives (RIs) are found, but their share is smaller
•Tense and aspect errors form over 70% of the errors.
28
•The error pattern is clearly different from that of the monolinguals
29
Poster Design by [email protected]
Contrastive analysis
Distribution of errors
Bilinguals
Child
8
Five derivational conjugations (binyanim)
4 root-based types that consist of 24 subclasses
3 tenses: past, present, future
Present tense: agreement in gender and number
Past & Future tenses: agreement in person, gender and
number
• No morphological manifestation of aspect
•
•
•
•
•
Four kinds of errors were found to be typical for both groups of children:
1. The production of infinitives instead of finite forms
2. Wrong person in present tense
3. Wrong gender in past tense
4. Stem errors
 Findings
igra-l-i
Hebrew verbal system
•
•
Adult VU%
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Present-Person
PL
productive:
igra-t’ – igraj-u
play
unproductive:
pisa-t’ – piš-u
write
plaka-t´ – plač-u cry
sypa-t´ – sypl´-u
poor
•
•
•
Productivity
•Number errors are marginal
3,5
Bilinguals
•All children are older than 3, at time of recording, when the basic of the
morphology of the target language is ‘normally’ established
18
• Productive use of verb inflection three/four months after the onset of verb
production (cf. , Kiebzak-Mandera et al. 1997, Poupynin 1998, Gagarina
2003, etc.)
• Ratio of the verb utterances (VU) over all utterances becomes near-target
number of the VU (comparing children’s output and input) five to seven
months after the emergence of verbs
• MLU reflects these developmental changes
Verb onset
A dul t V U%
•Person in the present errors occur only during the onset of productivity
VU%
•All children use Russian with Parents and Hebrew with siblings a nd peers
Monolingual acquisition of Russian
Inflectional classes: OB (inf) – CB (pres.1s)
•
•
P r oduct i vi t y
•Gender in the past is the major source of errors
Past
PL
1 igraj-u
3 igraj-et
play
to play
Present- Number
V er b onset
Verb
Utterances
G
ali
t
Eld
ar
Lit
M al
ic
ha
el
Zh
en
ya
Ta
li
Le
y
Pa a
tric
ia
igrat’ – Ipf
Past-Number
Present-Person
Present- Number
MLU
•All but Galit are sequential bilinguals. Eldar, Lital and Michael are early BL.
10
46
140
Liza
Past-Gender
Past-Number
Present-Person
Length of
exposure to
L2
3,5
• There is an increase in Le ya’s M L U and VU %.
• Three tenses and two aspects
Perfective (Pf)
Imperfective (Ipf)
Past
+
+
Present
+
Future
+
(analytical – to be +inf)
Monolinguals
Vitja
Past-Gender
 Bilingual subjects and data
 Subjects & Data - Monolingual
Child
Agreement errors
• Eight L1 Russian-L2 Hebrew bilingual children
• Second recording of two of the L1 Russian-L2 Hebrew bilingual children nine
months later
• The mean length of recordings 45 min
• Total analyzed utterances in both phases: 2,947
• Excluded: frozen forms, immediate repetitions, (self-repetitions), citations,
yes-no sentences, exclamations...