Content uploaded by Natalia Gagarina
Author content
All content in this area was uploaded by Natalia Gagarina on Dec 26, 2020
Content may be subject to copyright.
is is a contribution from Linguistic Approaches to Bilingualism 1:3
© 2011. John Benjamins Publishing Company
is electronic file may not be altered in any way.
e author(s) of this article is/are permitted to use this PDF file to generate printed copies to
be used by way of offprints, for their personal use only.
Permission is granted by the publishers to post this file on a closed server which is accessible
to members (students and staff) only of the author’s/s’ institute, it is not permitted to post
this PDF on the open internet.
For any other use of this material prior written permission should be obtained from the
publishers or through the Copyright Clearance Center (for USA: www.copyright.com).
Please contact rights@benjamins.nl or consult our website: www.benjamins.com
Tables of Contents, abstracts and guidelines are available at www.benjamins.com
John Benjamins Publishing Company
Linguistic Approaches to Bilingualism : (), –. ./lab...arm
– / - – © John Benjamins Publishing Company
e impact of internal and external factors on
linguistic performance in the home language
and in L2 among Russian-Hebrew and
Russian-German preschool children
Sharon Armon-Lotem1, Joel Walters1 and Natalia Gagarina2
1Bar-Ilan University, 2Zentrum fuer Allgemeine Sprachwissenscha, Berlin
is paper evaluates the contribution of external background factors which
pertain to the child’s environment (e.g., parents’ education, parents’ occupa-
tion, family size, etc.), and internal ones which reect the child’s time related
experience with language (e.g., chronological age, age of L2 onset, etc.) to the
development of linguistic skills in the two languages of bilingual children. 65
Russian-German (Mean age: 66mo, Range: 47-86mo) and 78 Russian-Israeli
migrant children (Mean age: 70mo, Range: 58-81) with comparable mean length
of L2 exposure (M=37mo) and family size (1.88 children) but dierent Socio-
Economic Status (SES), were tested with a battery of language tasks and their
parents were interviewed. Overall, internal, temporal, factors showed a stronger
relationship to language measures than external, environmental, factors: age of
L2 onset and length of L2 exposure correlated with L2, while parents’ education/
occupation showed positive correlations with both L1 and L2 measures. In the
Russian-German cohort, which had a sub-group with relatively lower SES, SES
positively correlated with L1 success as well.
Keywords: Child bilingualism, migration, external sociolinguistic factor,
internal sociolinguistic factors, German, Hebrew, Russian
Introduction
Standardized tests are used for screening and placement of second language learn-
ers in education programs. Nonetheless, Tucker (1999) in a World Bank study
on the use of two languages in education, reports that it takes up to ve years
of exposure to achieve monolingual norms, while our own ndings (Abutbul-Oz
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
2009) show that by the time children enter school (aer 2-3 years of L2 exposure) a
third of typically developing (TD) bilingual children still score below monolingual
norms, and most of those who score within norms are still below the monolingual
mean. Moreover, in order to reach a level of language prociency adequate for aca-
demic performance, understanding the basic tasks in kindergarten or elementary
school – children need 5+ years of exposure to language (Tucker 1999). Following
Hakuta and Garcia (1989), for understanding bilingual language development be-
yond purely linguistic aspects, social background factors inuencing second lan-
guage (L2) acquisition and rst language (L1) maintenance should be considered
(Ho, Core, Place, Rumiche, Señor, & Parra, in press). De Houwer (2000) also
explored the range of linguistic environments of four-year-old bilingual children
and found that three main environmental constituents were crucial for language
acquisition: mother input, sibling and peer input and television programs.
e present paper investigates the impact of extralinguistic inuences, elic-
ited from sociolinguistic interviews with parents and evaluates the contribution
of external and internal factors to the development of linguistic skills in the two
languages of bilingual children. e term external environmental factors is used
in here for factors pertaining to the child’s environment, i.e., factors that the child
brings to the language learning eort (including parents’ education and occupa-
tion, birth order and family size). Internal temporal factors are those that reect
the child’s time related language experience (including chronological age, age of L2
onset and length of L2 exposure). e study tracks the impact of these factors in
tandem with the language performance of bilingual children. e linguistic perfor-
mance targeted in the two languages included lexical, morphological, and syntactic
skills and was elicited by means of a variety of experimental techniques. is variety
of linguistic skills and experimental techniques makes it possible to test which of
these skills are more sensitive to internal and external factors, and which are more
resilient to factors which inuence bilingual language acquisition, and thus might
be used in testing for language impairment among bilinguals in future studies.
Research on the background factors which inuence the language acquisition
process has been conducted mainly among bilingual children where one of the
languages is English (Ho, 2006; Gleitman, & Newport, 1995; Paradis 2010) and
the focus has been on the relationship between extralinguistic factors and lan-
guage development in English as the societal and target language. Pearson (2007)
is one exception to this trend with a focus on L1 Spanish. Pearson weighed the
role of ve factors -- input, language status, access to literacy, family language use,
and community support -- in learning a minority language. In extensive studies
of Spanish-speaking children in Miami, she found that L1 input at home played
an important role in children’s maintenance of the minority language (cf. Klassert
and Gagarina 2010 who nd the same trend for Russian as a home language).
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
e extent to which L1 input at home which leads to L1 maintenance is related
to Socio-Economic Status (SES) was investigated by Lambert and Taylor (1996).
ey found that mothers from low SES encouraged their junior high children to
learn the L2/English in order to succeed educationally, while mothers from higher
SES encouraged L1 maintenance and saw additive bilingualism as a goal. On the
other hand, Oller and Eilers (2002) found that children of professionals do bet-
ter in L2 than children from working class families, with hardly any dierence in
L1 maintenance. When there was such a dierence, children with working class
mothers did better in the L1. at is, high SES parents value L1 culture more than
low SES parents, but seem to provide less support for the L1 and more support
for the L2 at home, while low SES parents encourage L2 acquisition as the key to
academic success but do not support it at home. Notably, these studies focused on
school age children and tested both written and oral skills, while the present study
investigates spoken language in children at the onset of bilingualism in preschool
years.
e present study (a) takes a broad perspective, including both the home
language as well as the societal language and relating to issues of home language
maintenance in addition to L2 acquisition; (b) uses a wide variety of tasks, both
normed instruments and tasks developed explicitly for this study, covering a va-
riety of linguistic structures in order to identify structures and tasks which cor-
relate with bilingual variation as well as those which are more resilient to bilingual
variation and might have the potential as screening indicators of Specic Lan-
guage Impairment (SLI); (c) addresses language pairs which have so far received
very little research attention. e data presented here come from two immigrant
groups with the same L1 (Russian), but typologically dierent L2s (Hebrew/Ger-
man), where neither of the languages is English. is design will allow generaliza-
tion of ndings beyond studies which focus on L2 English.
Our predictions are that external environmental factors (e.g., parents’ SES,
family size) will play a greater role in L1 maintenance, which was found to be
strongly inuenced by the home environment, while internal temporal factors
(e.g., age of onset and length of L2 exposure), are expected to correlate with suc-
cess in L2 acquisition, since they reect the child’s experience with the L2. More
specically, parents’ educational level and occupation are expected to inuence
linguistic performance in both L1 and L2, where children of parents with higher
levels of education and skilled/academic jobs will show better L2 performance and
children of parents in less skilled jobs or unemployed will show more L1 main-
tenance. L2 exposure, on the other hand, is expected to correlate positively with
performance on standardized tests, lexical abilities, and morphological and gram-
matical knowledge in the target language, but not with such abilities in the L1.
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
Method
Participants
Russian-German (N=65) and Russian-Israeli (N=78) immigrant children with com-
parable length of L2 exposure (M=37mo) and family size (M=1.88 children) partici-
pated in 6 to 7 sessions of 30 to 45 minutes each (depending on the child’s age) in
both languages. Children who were suspected as language impaired were excluded
from this study prior to data collection. None of the children were previously sus-
pected as language delayed, nor did any children show low performance in both lan-
guages. e composition of the two cohorts was very similar for gender, birth order,
number of siblings, and length of exposure (LoE) to the target language in preschool,
but somewhat dierent for age and age of L2 onset, with a wider age range in the
Russian-German cohort, which included a larger group of four-years-old than the
Russian-Hebrew cohort (Table 1). In addition, sociolinguistic interviews with par-
ents revealed a somewhat higher SES for the Russian-Hebrew cohort, as indicated
by parents’ education and occupation, and dierences in terms of age of L2 onset.
Despite the overall similarity between the two populations across a variety
of background measures, they diered in terms of SES, as measured by parents’
Table1. Background information for the two populations
Russian-German (N=65) Russian=Hebrew (N=78)
Age (in months)* M=65.78, range 47-86,
SD=10.5
M=70.04, range 54-81,
SD=5.73
Gender 33 male, 32 female 35 male, 43 female
Home language All 65 Russian-dominant 71/78 Russian-dominant
Age of L2 Onset(in
months)**
M=28.34, R=12-46,
SD=10.04
M=33.34, R=0-66,
SD=18.14
Length of L2 Exposure (in
months)
M=37, range 13-65,
SD=14.86
M=36.5mo, R=5-75mo,
SD=18.5
Birth order 35 rst-born, 30 later-born 39 rst-born, 39 later-born
Family size M=1.89, range 1-5,
SD=0.91
M=1.86, range 1-4,
SD=0.59
Parent education –Mother
(in years)**
M=12.61, range 10-17,
SD=1.97
M=14.35, range 9-21,
SD= 2.47
Parent education -Father (in
years) M=12.75, range 10-18 ,
SD=2.01
M=13.42, range 10-20,
SD=2.23
Parents occupation See next two gures
* p<0.01, **p<0.001
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
education and occupational status. Figure 1 presents the fathers’ occupational sta-
tus in the two cohorts (wherever this information was available) distinguishing
professionals (e.g., lawyer, doctor, teacher, engineer, programmer), skilled workers
(e.g., secretary, junior engineer, technician, paramedic), semi-skilled workers (e.g.,
cleaner, construction worker, driver, advertisement agent, caregiver) and those
unemployed (not included in Figure 1 since only 2 fathers in the Russian-German
cohort were unemployed).
As seen in Figure 1, fathers of the Russian-Hebrew children are mostly profes-
sionals or work in skilled jobs, while the majority of fathers of the Russian-Ger-
man children work in skilled and semiskilled jobs. at is, in terms of fathers’ oc-
cupation, the Israeli cohort has a higher SES, even though there are no signicant
dierences in years of education (see Table 1).
Figure 2 presents the same information for maternal occupation, showing an
even larger gap between the two national cohorts. As Figure 2 indicates, moth-
ers of the Russian-Hebrew children have a higher SES. While half of the mothers
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
Professional Skilled Semi-skilled
Father's Occupation
Russian-German
Father's Occupation
Russian-Hebrew
Figure1. Fathers’ occupational status
0%
10%
20%
30%
40%
50%
60%
Professional Skilled Semi-skilled Unemployed
Mother's Occupation
Russian-German
Mother's Occupation
Russian-Hebrew
Figure 2. Mothers’ occupational status
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
of the Russian-German children are unemployed, most mothers of the Russian-
Hebrew children, like the fathers, work in professional or skilled jobs. is could
reect the dierence in years of education (12.6 in Germany, 14.3 in Israel) or
availability of aordable daycare, thus positioning the Russian immigrant group
in Germany at a somewhat lower SES level.
Parent Interviews
Extralinguistic information was collected in individual parent interviews, at the
interviewee’s home, workplace or other location the parent chose. e interviews
were conducted in Russian (the mother tongue of the parents) by a Russian-na-
tive-speaking research assistant. Parents were asked questions about their child’s
age, place of birth, number and age of siblings, their own age, age on arrival in
Israel or Germany, marital status, country of origin, educational level, occupation,
languages spoken at work, the religious aliation, and contact with the country
of origin (OC). e interview continued with a discussion of the following topics:
information about the child’s language acquisition history and language learning
experiences (rst exposure to L1 and L2, transition from L1 to L2); his/her family
(siblings, grandparents); languages spoken and language policy at home; major
transitions (from home to kindergarten, from L1 kindergarten to L2 kindergarten,
from family to peers, from kindergarten to school); friends and peers; everyday
activities (weekday/weekend routine, favorite toys and games, TV preferences,
favorite songs, evening circles and aernoon activities, attendance at concerts/
plays); celebration of holidays; and reports on the child’s world knowledge in L1
and L2 (seasons/days/months). At the end of the conversation, the interviewer
and parent lled out summary charts with information about the child’s language
history and enrolment in dierent educational frameworks.
Language tasks
Linguistic data were collected from children individually in a quiet place in the
child’s preschool in monolingual sessions with native-Russian and native-He-
brew/German speaking research assistants.
Standardized tests. Standardized tests were used in the two target languages and a
non-word repetition measure was used in L1 Russian to assure that all participants
were not at risk for SLI. For German, we used the Sprachstandscreening (Grimm
2003) with two dierent versions depending on the child’s age. Children aged
3-3;11 years old were asked to repeat 13 nonsense words (aer one training item)
and construct plural forms from 10 singular nouns. Children aged 4-5;11 were
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
asked to repeat 18 nonsense words (one training item), 6 semantically meaningful
sentences and 9 anomalous sentences. For Hebrew, we used the Goralnik Diagnos-
tic Test (Goralnik, 1995) which includes measures of vocabulary (15 pictures to
name), comprehension (following directions), sentence repetition (5 sentences),
narrative (generating a story for a picture sequence) and pronunciation (of the 15
names). e standardized measures were supplemented by a battery of tasks eval-
uating linguistic performance in both languages. Data for the standardized tests
were scored using the respective test manuals and converted to z-scores, based on
monolingual norms in each language in order to be able to compare the results
across the two languages.
Sentence repetition (SR). SR tasks assessed verbal morphosyntax, prepositions,
complex syntax (wh-questions, passives, relative clauses), and case marking. Sen-
tence length varied from 4-7 words in Hebrew, 6-9 words in Russian, and 6-9
words in German, reecting the relative prevalence of function words and inec-
tions in the three languages. Syntactic complexity was equivalent across the three
languages. In Hebrew, 40 sentences tested verb inections (8 items for each of
ve categories), 30 sentences tested prepositions (5 items per category), and 20
sentences tested complex syntax (4 items per category). In Russian, 40 sentences
tested inections (5 items per inection category), 20 sentences tested complex
syntax (4 items per category), and 12 tested case markings (6 items per category).
Two dierent SR tasks were used to test prepositions due to dierences between
the two L2s. For the comparison with Hebrew, 35 Russian sentences tested prepo-
sitions (5 items per category) and for the comparison with German, 48 sentences
tested prepositions (6 items per category). In German, data for imitation of inec-
tions and complex syntax were not collected, since these features were included
in the German standardized SR test (Sprachstandscreening fuer das Vorschulalter
(Grimm 2003)). us, in German, 48 sentences tested prepositions (6 items per
category), and 12 sentences tested case (6 items per category).
Data from the SR tasks were analyzed in terms of correct vs. incorrect response
to target structures, since the aim was to evaluate specic linguistic knowledge
rather than the inuence of working memory and bilingual processing, where
ability to repeat the entire sentence would have been a more appropriate criterion.
is yielded a continuous scale for each imitation task, which does not reect
knowledge in a particular linguistic domain, but rather an estimation of overall
ability in the particular domain tested by the task.
Sentence completion (SC). Use of inections was also tested with a SC task (fol-
lowing Dromi et. al 1999). In this task, children were asked to complete sentences
with a target verb presented in the previous sentence in a less inected form. e
task included 42 items in Hebrew, testing 7 inectional morphemes (6 items per
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
category), 137 items in Russian, testing 9 inections (16 items in all categories
except innitives which had only 9 items per category), and 34 items in German,
testing 4 inections (4-13 items per category). Data from the SC task were ana-
lyzed in terms of correct target inection ignoring lexical substitutions.
Naming. e size and richness of the lexicon, both nouns and verbs, were test-
ed via a standardized picture naming task developed for German by Kauschke
(2007). For German, 36 pictures for verbs and 36 for nouns were used. For Rus-
sian, a reduced version consisting of pictures uniformly recognized by monolin-
gual Russian-speaking adults included 31 verbs and 31 nouns. For Hebrew 36
pictures for verbs and 35 for nouns were used (a picture of a crucix was deemed
culturally inappropriate). e data from the noun/verb picture naming task were
analyzed for the number of correct items per category, with a total score for both
categories together.
Analysis
Since each task had dierent numbers of items and dierent scales, for compari-
son purposes, all scores were converted to z-scores which were calculated for the
children in each cohort. For the correlations we used the continuous data from
external and internal factors (e.g., number of month of exposure). In addition to
Pearson correlations which were applied to all the data, partial correlations were
conducted for non-standardized tasks, where age is not already addressed by the
standardized norms. For ANOVAs, continuous data were converted to categorical
data (e.g., short, middle, and long LoE) to enable comparison within and between
groups along with post-hoc tests for pairwise comparisons. Age was controlled for
when applying ANOVAs to non-standardized tests.
Results
Of the data obtained from the parents, only factors which were signicant for at
least one task in one of the three languages are reported here. e ndings are
presented as follows: we begin with the Israeli cohort, moving from L1 to L2 per-
formance, and then present the data for the German cohort, again beginning with
L1 and moving to L2.
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
L1 Russian performance among Russian-Hebrew bilinguals.
Only family size and, to a lesser extent, birth order show signicant negative corre-
lations with L1 maintenance in the Russian-Hebrew group (see Table 2). Chrono-
logical age showed no correlation with L1 maintenance in the Russian-Hebrew co-
hort. Similarly, SES did not play any role in this cohort, reecting its homogeneity.
Table2. Correlations between background factors and L1 Russian tasks in the Russian-
Hebrew cohort.
Imitation:
preposi-
tions
Imitation:
inections
Imitation:
complex
syntax
Imitation:
case
Sentence
completion:
inections Naming
Family size
(number of
children)
-.208 -.190 -.217 -.281* -.344** -.330**
Birth order -.138 -.147 -.141 -.227 -.217 -.289*
* Correlation is signicant at the 0.05 level (2-tailed).
** Correlation is signicant at the 0.01 level (2-tailed).
A weak negative correlation was found between family size and imitation of cases
in L1 Russian and somewhat stronger correlations were found between family size
and the use of inections in SC and naming. Birth order showed only a weak nega-
tive correlation with naming.
Focusing on family size, Table 3 presents z-scores comparing families with one
child (singleton) and those with more than one child
Table3. Level of performance (Means (SD)) on Russian language tasks in Israel as a
function of family size
N
Imitation:
prepositions
Imitation:
inections
Imitation:
complex
syntax
Imitation:
case
Sentence
completion:
inections Naming
Singleton 19 .12 (1.13) -.08 (1.33) .17 (.98) .16 (1.01) .13 (.99) .42 (.86)
More than
one child
49 -.12 (1.05) -.03 (1.01) -.12 (1.01) -.04 (1.01) -.09 (1.01) -.09 (1.02)
Table 3 shows that for some phenomena, singletons perform better in Russian,
suggesting that family size shows some inuence on L1/Russian language mainte-
nance. However, these dierences do not reach statistical signicance, apart from
Noun-Verb naming which shows a trend toward signicance (F(1,76)=3.903,
p=0.052).
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
L2 Hebrew performance among Russian-Hebrew bilinguals.
Two closely related background variables showed signicant correlations with L2
performance: Age of L2 onset and length of L2 exposure, while chronological age
did not yield any signicant correlations (see Table 4).
Table4. Correlations between background factors and L2 Hebrew tasks in the Russian-
Hebrew cohort.
Stan-
dard-
ized
Hebrew
test
Stan-
dard-
ized
vocabu-
lary
Stan-
dard-
ized
sentence
imita-
tion
Imita-
tion:
preposi-
tions
Imita-
tion:
inec-
tions
Imita-
tion:
complex
syntax
Sen-
tence
comple-
tion:
inec-
tions Naming
Internal variables
AoO in
months
-.440** -0.521** -0.291* -.240* -.355** -.030 -.386** -.408**
LoE in
months
.412** 0.476** 0.258* .200 .300** .054 .411** .395**
External variables
Father’s
educa-
tion in
years
.102 0.405** 0.057 .104 .037 .146 .066 .258*
* Correlation is signicant at the 0.05 level (2-tailed).
** Correlation is signicant at the 0.01 level (2-tailed).
Signicant negative correlations emerged between AoO and level of performance
in the L2/Hebrew standardized test for the total score, for the vocabulary score,
and for the sentence imitation score. Further support for the contribution of age
of L2 onset to success in L2 acquisition comes from the data elicited in the tasks
designed to target specic structures. Similarly, signicant negative correlations
emerged between AoO in months and performance on the non-standardized lan-
guage tasks:,verb inection imitation, prepositions imitation, inection in SC and
noun-verb naming. ese correlations (except for prepositions imitation) remain
signicant when chronological age is controlled for.
Likewise, signicant positive correlations emerged between LoE in months and
level of performance in L2 standardized tests, for the total score, for vocabulary, and
for imitation. e tasks developed for this study were more sensitive to length of
L2 exposure than the standardized test. Similarly, signicant positive correlations
emerged between LoE in months and level of performance on the non-standardized
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
language tasks: inection imitation, inection in SC and noun-verb naming. ese
correlations remain signicant when chronological age is controlled for.
None of the external variables potentially related to SES (parents’ education
and occupation, family size and birth order) showed signicant correlations with
L2 performance apart from father’s education which shows a strong correlation
with the standardized vocabulary score and a weak correlation with the non-stan-
dardized naming. is pattern reects the relative homogeneity of the Russian-
Hebrew cohort (see Table 4).
Age of L2 onset. Table 5 presents the z-scores for the Hebrew standardized test and
two of its components: vocabulary and sentence imitation. e z-scores for the
standardized measures presented here reect deviations from monolingual means
using monolingual norms. Participants were divided into three groups in terms
of age of L2 onset: earlier AoO (<25 months), intermediate AoO (25-36 months),
later AoO (>36 months).
Table5. Level of performance on the L2/Hebrew standardized test and subtests as a func-
tion of age of L2 onset
Age of L2 Onset N Total Imitation Vocabulary
Earlier AoO (<25 months) 22 -.38 (.96) .51 (.43) -.75 (1.05)
Intermediate AoO (25-36 months) 21 -.98 (.80) .09 (.80) -1.60 (1.18)
Later AoO (>36 months) 29 -1.82 (1.68) -.19 (1.11) -2.57 (1.85)
Table 5 shows better performance on all three measures as a function of AoO. One-
way ANOVAs conducted for each of the three standardized scores yielded signi-
cant dierences for age of L2 onset, for the total score (F(2,69)=8.32, p=0.001), for
imitation (F (2,69)=4.16, p=0.02), and for vocabulary (F (2,69)=9.82, p<0.001). A
series of pairwise comparisons with post hoc Schees traced this signicance to
the dierence between the later AoO group and the earlier AoO group (p=0.001
for total scores, p<0.001 for vocabulary scores, and p=0.02 for imitation scores),
and a nearly signicant dierence between the later AoO group and the interme-
diate AoO group (p=0.074 for total score and p=0.078 for the vocabulary score),
with no signicant dierence between the earlier and intermediate onset groups.
Furthermore, those children who began acquiring Hebrew before the age of
two performed within the monolingual norms (in the range of +/- one SD) both in
terms of the total score on the standardized test and in terms of its subsets. On the
other hand, those who began L2 acquisition aer the age of three were more than
1.5 SDs below the monolingual norms for the total score and for the vocabulary
score (even though the mean LoE was 21 months), but not for the SR score of the
standardized test.
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
Table 6 presents z-scores for the two measures testing verb inections: a) SC
probing 3rd person inections; b) sentence imitation testing 1st and 2nd person
inections and for the noun-verb naming task. Z-scores here and throughout the
rest of the paper were calculated on the basis of each child’s scores in comparison
to the entire group of children for each particular task, unless indicated otherwise.
As before, participants were divided into the same three groups in terms of age of
L2 onset.
Table6. Level of performance on the inections and noun-verb vocabulary in
L2/Hebrew as a function of age of L2 onset
Age of L2 Onset N
Imitation:
inection
Sentence completion:
inections Naming
Earlier AoO (<25 months) 22 .44 (.56) .48 (.71) .48 (.51)
Intermediate AoO (25-36 months) 21 .10 (.62) .04 (.97) .33 (.51)
Later AoO (>36 months) 29 -.32 (1.26) -.33 (1.09) -21 (.84)
Table 6 shows better performance on all three linguistic measures as a function
of AoO. One-way ANOVAs, controlling for chronological age, yielded signicant
dierences among the three AoO groups for imitation of inections (F(2,73)=3.16,
p=0.03), for inections in SC (F(2,73)=3.57, p=0.018) and for noun-verb naming
(F(2,73)=5.14, p=0.003). Pairwise comparisons traced this signicance to the dier-
ence between the group with the later AoO and the group with the earlier AoO for all
three tasks (Schee yielded: p=0.02 for imitation of inections, p=0.016 for inec-
tions in SC, and p=0.002 for noun-verb naming); the intermediate group showed no
signicant dierence from the two other groups on all tasks except for a signicant
dierence from the group with the later AoO on the naming task (p=0.019).
Length of L2 Exposure. Table 7 presents the z-scores for the L2/Hebrew standard-
ized test for the same three measures, this time as a function of exposure to L2.
Participants were divided into three groups in terms of LoE: short (<25 months),
middle (25-48 months), long (>48 months).
Table7. Level of performance on the the L2/Hebrew standardized test and its subtests as
a function of Length of Exposure
Length of Exposure N Total Imitation Vocabulary
Short (<25 months) 25 -2.03 (1.33) -.08 (1.06) -2.74 (2.00)
Middle (25-48 months) 35 -.99 (1.32) .02 (.90) -1.70 (1.21)
Long (>48 months) 17 -.41 (1.03) .47 (.48) -.68 (1.03)
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
Table 7 shows better performance on total and vocabulary scores as a func-
tion of LoE, but did not show this eect for imitation. One-way ANOVAs con-
ducted for each of the three standardized scores yielded signicant dierences for
the L2/Hebrew total standardized test score (F(2,70)=7.16, p<0.001) and for the
vocabulary score (F(2,70)=9.73, p<0.001), but not for imitation. Pairwise com-
parisons for the two signicant measures (tested via Schee) showed that chil-
dren with more than four years of exposure performed signicantly better than
those exposed to Hebrew for less than two years, and this held for both measures
(p<0.001). e middle exposure group also scored signicantly better than the
lowest group for total score (p=0.014) and for vocabulary (p=0.04), with a nearly
signicant dierence from the high exposure group for vocabulary (p=0.066).
In addition, children with more than four years of exposure performed within
monolingual norms (in the range of +/- one SD) both in terms of their total score on
the standardized test and in terms of its subsets (for details see Abutbul-Oz 2009).
On the other hand, those with fewer than two years of exposure fell more than
1.5SD below the monolingual norms for the total standardized test and the vocabu-
lary scores. Finally, while the middle exposure group was within the monolingual
norms for their total score and its imitation subtest, it fell below monolingual norms
in vocabulary, indicating a certain weakness of the bilingual group in this domain.
Table 8 presents the z-scores for the measures developed in the present study
for which correlations with LoE emerged, viz., the two measures testing verb in-
ections: a) SC probing 3rd person inections; b) sentence imitation testing 1st
and 2nd person inections as well as for the noun-verb naming task.
One-way ANOVAs controlling for chronological age show signicant dif-
ference among the three exposure groups for inection imitation (F(2,72)=3.56,
p=0.019), for inection in SC (F(2,72)=4.45, p=0.006), and for naming
(F(2,72)=5.36, p=0.002). Pairwise comparisons (Schee) indicated that children
with more than four years of exposure performed signicantly better than those
with less than two years of exposure for all measures (p=0.028 for imitation and
p=0.003 for the other two measures). e middle group also scored signicantly
Table8. Level of performance on verb inections and noun-verb vocabulary as a func-
tion of length of L2 exposure
Length of Exposure N
Imitation:
inections
Sentence
completion:
inections Naming
Short (<25 months) 25 -.32 (1.22) -.49 (.98) -.28 (.82)
Middle (25-48 months) 35 .05 (.87) .11 (1.01) .19 (.66)
Long (>48 months) 17 .49 (.48) .56 (.62) .50 (.56)
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
better than the lowest group on naming (p=0.042), and almost signicantly better
on SC (p=0.054), but did not dier from the highest exposure group on any task.
L1 Russian performance among Russian-German bilinguals.
Whereas in the Russian-Hebrew data most of the external variables examined in
this research did not show signicant correlations with L1 maintenance, the Rus-
sian-German data present a very dierent picture. SES shows strong correlations
with L1 maintenance and development in the Russian-German group (see Table 9).
As seen in Table 9, positive correlations are found between years of mother’s
education and L1 maintenance for all tasks and between mother’s and father’s oc-
cupation for some of the language tasks. While SES showed positive correlations
with L1 maintenance, family size as well as birth order showed a negative inu-
ence on L1 maintenance for imitation of prepositions, for case marking and for
inections in SC. Moreover, while months of chronological age do not show a
correlation with L1/Russian maintenance in the Russian-German cohort, L1/Rus-
sian performance correlates positively with age group (four year olds, ve year
olds and six year olds) for: imitation of prepositions (r(61)=0.35, p=0.006), imita-
tion of complex syntax (r(61)=0.29, p=0.023), verb inections in SC (r(61)=0.32,
p=0.013) and naming (r(61)=0.36, p=0.005).
Chronological age. Table 10 shows the ndings for imitation of prepositions and
complex syntax, the use of inections in SC and performance on the naming task,
Table9. Correlations between background factors and L1 Russian tasks in the Russian–
German cohort
Imitation:
prepositions
Imitation:
inections
Imitation:
complex
syntax
Imitation:
case
Sentence
completion:
inection: Naming
External variables
Mother’s educa-
tion in years .461** .289* .316* .437** .483** .426**
Mother’s occupa-
tion (groups) .296* .055 .242 .248 .359** .310*
Father’s occupa-
tion (groups) .288* .138 .204 .278* .384** .255*
Birth Order -.296* -.101 -.141 -.242 -.277* -.213
Family Size -.275* -.125 -.162 -.257* -.321* -.217
* Correlation is signicant at the 0.05 level (2-tailed).
** Correlation is signicant at the 0.01 level (2-tailed).
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
where signicant correlations were found. For each particular task, z-scores were
calculated on the basis of relative scores in comparison to the entire group of chil-
dren. Data are presented for fours, ves, and sixes.
One-way ANOVAs yielded signicant dierences among the three age groups
for: 1) preposition imitation, (F(2,58)=4.04, p=0.023), traceable to a signicant
dierence between the 4- and 6-year-olds; 2) inections in SC, (F(2,58)=4.25,
p=0.019), also due to a signicant dierence between the 4- and 6-year-olds; and
3) naming, (F(2,58)=6.02, p=0.004) traceable to a signicant dierence between
the 4- and 6-year-olds as well as between the 5- and 6-year-olds.
SES – Mothers occupation and education. Half of the mothers in the Russian-Ger-
man cohort were unemployed and stayed at home to take care of their children un-
til the latter entered the educational system. Our ndings show a trend for better
language scores among children of mothers who work in skilled and professional
occupations compared to those of children whose mothers are unemployed, with
signicant dierences for imitation of prepositions and inections in SC, as can be
seen in Table 11 (Means and SDs are based on z-scores).
Table 12 shows means and standard deviations for the relation between moth-
er’s education and performance in L1 Russian for the six tasks, showing signicant
dierences between mother’s educational level for all six tasks.
Table10. Level of performance on the L1 Russian tasks as a function of chronological age
Chronological Age N
Imitation:
preposition
Imitation:
complex
syntax
Sentence
completion:
inections Naming
4-year-olds 18 -.40 (.83) -.45 (.98) -.27 (.81) -.29 (.77)
5-year-olds 26 -.05 (1.03) .12 (.93) -.18 (1.07) -.23 (.99)
6-year-olds 17 .51 (.95) .30 (1.01) .57 (.90) .66 (.96)
Table11. Mean scores (SDs) on L1 Russian tests as a function of mother’s occupation
N
Imita-
tion:
preposi-
tions*
Imita-
tion:
inec-
tions
Imita-
tion:
complex
syntax
Imita-
tion:
case
Sentence
comple-
tion: in-
ections* Naming
Professional 7 .67 (1.11) .02 (1.06) .47 (1.01) .48 (.91) .82 (.86) .72 (1.07)
Skilled 15 .04 (.87) .14 (.80) -.04 (.57) .18 (.83) .15 (.81) .078 (.77)
Semi-skilled 9 .40 (.89) .48 (.31) .55 (.61) .32 (.85) .42 (.69) -.08 (1.12)
Unemployed 31 -.30 (.99) -.22
(1.17)
-.25
(1.18)
-.30
(1.08)
-.39 (1.04) -.19 (1.01)
* Means are signicantly dierent at the 0.05 level (2-tailed).
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
Table 12 shows that the children whose mothers had more than 14 years of educa-
tion (N=18) outperformed those whose mothers had less than 14 years of educa-
tion on all tasks. One-way ANOVAs show signicant dierences for all linguistic
measures: imitation of inections (F(1,59)=6.54, p=0.013), imitation of com-
plex syntax (F(1,59)=4.76, p=0.033), imitation of prepositions (F(1,59)=14.01 ,
p<0.001), and imitation of case (F(1,59)=11.02, p=0.002); use of inections in SC
(F(1,59)=16.76, p<0.001), and noun-verb naming (F(1,59)=9.96, p=0.001).
Family size and birth order. Table 13 shows that singletons performed better in
L1 on a variety of L1 language tasks. Z-scores are presented for the imitation of
prepositions, inections, complex syntax, and case as well as the use of inections
in SC, all as a function of family size, comparing families with one child (single-
ton) and those with more than one child.
Table 13 shows that singletons performed better on all the Russian language
tasks. ese dierences were signicant only for SC with inections (F(1,59)=4.62,
p<0.05), and nearly signicant for imitation of case and for noun-verb naming. A
similar inverse relationship between birth order and L1 maintenance was found,
with rst born children scoring higher than later born children on a variety of L1
Russian tasks, but this nding did not reach signicance for any of the tasks.
Table13. Mean scores (SD) on L1 Russian tests as a function of family size
N
Imitation:
prepositions
Imitation:
inections
Imitation:
complex
syntax
Imitation:
case
Sentence
completion:
inections Naming
Singleton 23 .20 (1.09) .08 (1.00) .21 (.97) .30 (.98) .34 (1.01) .32 (1.02)
More than
one child
38 -.12 (.93) -.05 (1.00) -.13 (1.00) -.18 (1.02) -.21 (.94) -.19 (.94)
Table12. Mean scores (SD) on L1 Russian tests as a function of mother’s education
N
Imitation:
prepositions
Imitation:
inec-
tions
Imitation:
complex
syntax
Imitation:
case
Sentence
completion:
inections
Noun-Verb
Naming
Less than
14 years
43 -.28 (.98) -.20 (1.09) -.18 (1.06) -.25 (1.01) -.30 (.97) -.26 (.95)
14 years
and more
18 .67 (.66) .48 (.44) .42 (.69) .61 (.63) .72 (.66) .62 (.84)
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
L2 German performance among Russian-German bilinguals.
While age of L2 onset was the strongest predictor of L2 performance in the Rus-
sian-Hebrew cohort, it showed no correlation with linguistic performance in L2/
German in the Russian-German cohort. In contrast, chronological age correlates
with all of the structures tested, as shown in Table 14.
A positive correlation resulted between length of L2 exposure (in months) and
L2 performance for all tests
Another variable which showed signicant correlations with L2 German tasks
was parents’ occupation (external variable), which introduces the role of SES. Ta-
ble 14 shows a strong relationship between father’s occupation and performance
in L2 German for all ve measures. Similar, though somewhat weaker correlations
were found between mother’s occupation and success on the standardized task
and naming, and between mother’s education and imitation of prepositions.
Chronological age. Table 15 presents the group-based z-scores for the standardized
test as well as for imitation of prepositions and case, inections in SC, and perfor-
mance on noun-verb naming. e score for the standardized test is normed for
monolinguals, while the other scores are calculated for the bilingual group only.
For the standardized test, all three age groups scored close to monolingual
norms (within one standard deviation), but they are all nevertheless below the
Table14. Correlations between all background factors and L2 German in the Russian-
German cohort
Standardized
test (SR)
Imitation:
prepositions
Imitation:
case
Sentence
completion:
inections Naming
Internal (temporal) variables
Age in months .375** .659** .414** .482** .631**
Length of Exposure in
months .279* .427** .254* .385** .430**
External variables
Mother’s education in
years .238 .266* .179 .177 .174
Father’s occupation
(groups) .331** .372** .289* .289* .330**
Mother’s occupation
(groups) .325* .258 .131 .250 .263*
* Correlation is signicant at the 0.05 level (2-tailed).
** Correlation is signicant at the 0.01 level (2-tailed).
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
monolingual mean; even the older children barely reach the monolingual mean.
An ANOVA yielded a signicant dierence between the age groups (F(2,58)=4.89,
p=0.01), which is conrmed by a post-hoc Schee test as a signicant dierence
between the 4- and the 6-year-olds (p=0.011).
For the tasks developed for the present study, Table 15 shows a strong relation
between age and level of success for each task, with the 4-year-olds scoring be-
low the bilingual mean on all tasks and the 6-year-oldsscoring above the bilingual
mean on all tasks. One-way ANOVAs revealed signicant dierences between
the three age groups for all categories tested: for inection in SC (F(2,53)=8.47,
p=0.001), for preposition imitation (F(2,55)=26.6, p<0.001), for case imitation
(F(2,58)=6.03, p=0.004), and for the naming task (F (2,58)=17.32, p<0.001).
Pairwise comparisons using a Schee post-hoc test showed that the 4-year-olds
were outperformed signicantly by both the 5-year-olds (p<0.05) and the 6-year-
olds (p<0.001) on all tasks, with no signicant dierence between the 5-and the
6-year-olds on any of the tasks.
Length of exposure. Table 16 presents z-scores for the same tasks as a func-
tion of LoE. Participants were divided into three groups in terms of LoE: short
(<25 months), middle (25-48 months), long (>48 months). e z-scores for the
standardized test are normed for monolinguals, while the other z-scores were cal-
culated for the bilingual group only.
Table15. Level of performance on the dierent language tasks as a function of chrono-
logical age
Chronological Age N
Standardized
test (SR)
Imitation:
prepositions
Imitation:
case
Sentence
completion:
inections Naming
4-year-olds 18 -.97 -1.22 -.88 -.82 -1.04
5-year-olds 26 -.59 .28 .13 .07 .33
6-year-olds 17 -.11 .57 .48 .58 .51
Table16. Level of performance on L2/German tasks as a function of length of L2 exposure
Length of Exposure N
Standard-
ized test
(SR)
Imitation
preposition
Imitation
case
Sentence
completion:
inections Naming
Short (<25 months) 6 -1.01 (.76) -1.00 (.61) -.84 (1.48) -1.19 (1.03) -1.29 (.67)
Middle (25-48 months) 31 -.66 (.92) -.05 (.99) -.04 (1.00) -.09 (.90) -.03 (.95)
Long (>48 months) 22 -.31 (.76) .32 (.93) .22 (.80) .44 (.90) .51 (.89)
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
For the standardized test, Table 16 shows that while all three groups scored
close to monolingual norms (within one standard deviation), they all neverthe-
less fell below the monolingual mean. More specically, children with longer LoE
(over 4 years) barely reached the monolingual mean. No signicant dierences for
LoE were found on the standardized test.
For the tasks which were developed for the present study, one-way ANOVAs
yielded signicant dierences among the three exposure groups for preposition
imitation (F(2,55)=4.48, p=0.012), for inection in SC (F(2,53)=6.07, p=0.004),
and for naming (F(2,58)=8.07, p=0.001), with no signicant dierence for case
imitation. Pairwise comparisons (Schee) indicated that children with more than
four years of exposure performed signicantly better than those exposed to Ger-
man with fewer than two years of exposure, and this held for all measures where
the ANOVA was signicant (p=0.005 for inections in SC, p=0.01 for imitation
of prepositions, and p<0.001 for naming). e middle group scored signicantly
better than the lowest group only on naming (p=0.029), but did not dier from the
highest exposure group.
Since younger children had a shorter LoE, the data were reanalyzed with a
two-way ANOVA with chronological age and LoE as independent variables.
is AGE X LoE ANOVA yielded a signicant dierence only for case imita-
tion, (F(2,47)=3.38, p<0.02) with near signicance on SC (p=0.05), and naming
(p=0.09), the latter due to a signicant dierence for the verb items on the naming
task, (F(2,47)=3.22, p=0.031).
Since the 5- and 6-year-olds did not dier signicantly as a function of chron-
ological age, a second analysis in terms of length of L2 exposure was conducted for
these two age groups alone, excluding the four year olds.1 One-way ANOVAs with
chronological age as a covariate showed signicant group dierences for: inec-
tions in SC, (F(2,39)=7.84, p=0.001), for case imitation, (F(2,39)=4.19, p=0.023);
and for naming, (F(2,39)=4.125, p=0.024). All of these eects were due to the dif-
ference between children with more than 48 months of exposure and those with
fewer than 48 months of exposure. With chronological age as a covariate, imita-
tion of prepositions did not show a signicant dierence as a function of LoE.
SES – Parents’ occupation and education. SES was a prominent factor in the perfor-
mance of the Russian-German group, which was more diverse in terms of parent
occupation (Figures 1 and 2). Table 17 presents the children’s performance on
imitation of prepositions and case marking, on the use of inections in the SC
task, and on the Noun-Verb naming task as a function of the father’s occupation.
Data for children of unemployed fathers were not presented since there were only
two such cases in the sample.
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
One-way ANOVAs comparing each of these measures across the three SES groups
revealed signicant dierences for all four measures: imitation of prepositions
(F(2,50)=6.33, p=0.004), imitation of case marking (F(2,50)=6.02, p=0.005),
inections in the SC task (F(2,50)=4.88, p=0.012), and noun-verb naming
(F(2,50)=5.04, p=0.01). A Schee posthoc test showed that these dierences can
be traced to a dierence between children whose fathers worked in a semi-skilled
job and those whose fathers worked in a professional job (0.004<p<0.009). A simi-
lar trend was found for children of professional mothers who scored better in L2
German than children of skilled and unemployed mothers on some of our tasks,
with dierences among the four groups being nearly signicant.
An analysis of the standardized SR task yielded similar results, shown in
Table18. One-way ANOVAs showed a signicant dierence for both father’s oc-
cupation, (F(2,57)=5.16, p=0.009), traced to the dierence between professional
and semiskilled fathers (p=0.008), as well as mother’s occupation (F(2,57)=4.52,
p=0.007), traced to signicant and near signicant dierences between profes-
sional mothers and other groups (p=0.032 skilled, p=0.15 semiskilled, p=0.008
unemployed). A two way ANOVA with father’s occupation and mother’s occu-
pation as independent variables showed even greater impact of SES when both
parents had a higher occupational status, (F(4,53)=4.29, p=0.007), than when only
one parent had a professional occupation. Father’s education had no eect on per-
formance for the normed tests, and mother’s education had a marginally signi-
cant inuence, where children of mothers with more than 14 years of education
scored higher than children of less educated mothers.
Table18. Level of performance on normed language tests as a function of fathers’ and
mothers’ occupation
Unemployed Semi-skilled Skilled Professional
N Mean N Mean N Mean N Mean
Father 0 25 –.90 (.71) 18 –.61 (.94) 16 –.06 (83)
Mother 30 −.77 (.75) 9 –.50 (.87) 15 –.68 (.85) 7 .44 (.82)
Table17. Performance on L2 tasks as a function of father’s occupation
N
Imitation:
prepositions
Imitation:
case
Sentence completion:
inections Naming
Professional 14 .57 (.60) .53 (.58) .55 (.82) .63 (.49)
Skilled 18 .10 (.95) .08 (.93) –.01 (.94) .03 (1.01)
Semi-skilled 21 –.56 (1.10) –.55 (1.09) –.46 (1.02) –.41 (1.10)
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
Discussion
Cross-national comparison. Overall, internal time-related factors correlated with
the children’s performance in their L2 in both cohorts, while external environ-
mental factors correlated with the children’s performance in the L1 measures in
both cohorts. External factors correlated with the L2 only in the Russian-German
cohort. A comparison of the two cohorts yielded the following observations: in
both cohorts LoE positively correlated with L2 performance, while family size
showed a negative correlation with L1 maintenance. Some dierences between
the two cohorts reect the dierent level of heterogeneity within each cohort. In
the Russian-Hebrew cohort, age of L2 onset had a strong impact on L2 acquisition
reecting the wide range of L2 onset - from birth (simultaneous bilinguals) to just
a year before data were collected (M=34.79 months, Range 0-66 months). In the
Russian-German cohort, by contrast, for around 60% of the children, the age of
L2 onset was between 12 and 30 months. In the Russian-German cohort, chrono-
logical age impacted both on L1 maintenance and L2 acquisition, a reection of
the wider age range in this cohort (47-86 months, with a large group of fours).
Moreover, the dierence in schooling systems, inter alia, the age when children
enter school, in the two countries might also have inuenced the level of L2 per-
formance. In particular, some of the older children in the German cohort were
already in rst the grade at the time of the data collection because in Germany
children enter elementary school at age ve, while all children in Israel were still
in the kindergarten.
Socioeconomic variables were also more inuential in the Russian-German
cohort, with parent’s occupation impacting on L2 acquisition and mother’s edu-
cation and occupation impacting on L1 maintenance. is, as mentioned earlier,
reects the heterogeneity of this cohort in terms of SES and the high level of un-
employment among mothers in the present study, as well as the fact that only 5
out of the 30 unemployed mothers had more than 14 years of education. Last, but
not least, family size had a negative impact on L1 language maintenance, which
could be attributed to a stricter “L1-only” policy at home with singleton children,
or perhaps a lack of transition to L2 at home, since the child is a minority in a Rus-
sian-speaking home. However, this social factor reached signicance only in the
Russian-German cohort, a reection of the negative correlation between mothers’
education and family size (r=-0.30, p=0.018) in that cohort. at is, mothers with
higher education have smaller families, and the children in these families show
stronger L1 maintenance
Linguistic measures: Imitation and Naming. Of the language measures used in the
present study, the standardized tests shed light on the role of both age of L2 onset
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
and length of L2 exposure. In Hebrew, performance on the standardized test and
on its vocabulary subtest negatively correlated with age of L2 onset and positively
correlated with length of L2 exposure. However, higher scores on the sentence
imitation component negatively correlated with age of L2 onset but showed no
correlation with LoE. Similarly, the German standardized test, that is a Sentence
Repetition (SR) task, showed no correlation with LoE. For both groups of L2 chil-
dren, children with more than 24 months of exposure scored within the mono-
lingual norms on sentence repetition. is suggests that two years of exposure are
more than enough to perform within the monolingual norms on SR. For this task,
children with two years of exposure and no diculty with working memory (as is
expected for children with typical language development), thus, gain support from
the model presented in the input sentence, while in other tasks, like naming, they
lagged behind, since they require pure initiation on the part of the child with no
previous model to rely on.
is nding has important implications for identifying SLI in bilingual
populations. Sentence repetition has been shown to be able to distinguish be-
tween typically-developing children and children with language impairment in
monolingual populations (Conti-Ramsden, Botting, & Faragher, 2001; Friedma-
nn & Lavi, 2006; See-Gabriel, Chiat, & Roy, 2008). Elicited imitation involves
listening, comprehension and production and taps into all levels of language (syn-
tax, morphology, phonology, semantics) and phonological memory (Rummer &
Engelkamp, 2001). e present ndings suggest that this measure, even when us-
ing monolingual norms, could be valuable for identifying SLI among bilinguals
as well, since it is less sensitive to bilingualism, showing less impact of LoE - see
Armon-Lotem, Danon and Walters (2008) for imitation of prepositions, Armon-
Lotem, Adam, Saiegh-Haddad and Walters (2008) for imitation of inections, and
Porat (2009) for imitation of complex syntax among English-Hebrew bilinguals
with typical language development and with SLI, but also Chiat et al. (under re-
view) for dierences based on SR tasks .
Moreover, the variety of structures used in the imitation tasks developed for
the present study further supports this suggestion, showing that for children with
typical language development (bilinguals as well as monolinguals) sentence imita-
tion is easier than SC or naming. ese latter tasks require manipulation of mor-
phosyntactic and lexical knowledge, without presenting a model which can aid bi-
lingual children with typical language development whose phonological memory
is intact. Crucially, this is not expected to help bilingual children with SLI whose
phonological memory seems to be more limited. It is important to note that all
groups had scores within the monolingual norms on the standardized sentence
imitation subtest of both the Hebrew and German standardized tests, whereas
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
naming was sensitive to external factors. is suggests that imitation is more resil-
ient to external factors, while naming is more sensitive to them.
Lexical knowledge, as measured by picture naming in all three languages, was
found to be highly sensitive to variations for virtually all external and internal
variables and for both languages in this study. A survey of Tables 2-18 shows that
the naming task and the vocabulary subtest of the Hebrew standardized test gener-
ate the steepest slopes of all the language measures used. e Noun-Verb naming
test showed a similar tendency to that of the vocabulary subsection of the L2/
Hebrew standardized test.
is nding contributes to the literature on the role of social variables in lexi-
cal development for monolingual children (see Ho, 2006, for a review) and a now
growing literature for bilinguals (Ho et al., in press). ose studies cover a range
of linguistic milestones and tasks, but focus primarily on vocabulary size assessed
via the MacArthur-Bates CDC (Fenson, Dale, Reznick, Bates, al & Pethick 1994)
and standardized measures of receptive vocabulary such as the Peabody Picture
Vocabulary Test (Dunn & Dunn 1997). e present study examined lexical abilities
in a production task involving picture naming for both nouns and verbs in both
languages of bilingual preschool children. In the spirit of Ho’s (2006) recom-
mendation in connection with mixed ndings regarding SES eects on language
outcomes and Hart and Risley’s (1995) detailed investigation of early vocabulary
development, we suggest that future studies of vocabulary in bilinguals could be-
gin by examining the relative inuence of input and social variables on dierent
word classes and semantic categories (cf., Hart & Risley 1995). In this light, it is
expected that nouns will be most prone to inuence from both external and inter-
nal factors, with SES variables such as parents’ occupation and education exerting
the strongest impact on vocabulary acquisition, especially on acquisition of nouns.
Nouns have several properties which make them more prone to social inu-
ences. Structurally, nouns, especially nouns in subject NPs, are more independent
and less language-specic than verbs (Gentner 2006), which take arguments and
are constrained by other syntactic processes. Second, nouns represent the larg-
est word class, constituting as much as 50 percent of the adult lexicon, and the
noun bias in Western languages is well documented (e.g. Bloom 2000). Finally,
nouns constitute a more varied word class, which makes their acquisition more
dependent on experience than verbs, the latter being grounded in grammatical
constraints. Further studies should examine additional word classes, a variety of
semantic categories like those examined by Hart and Risley (1995) and multiple
tasks (e.g., receptive vocabulary, naming, translation).
Internal temporal variables - L2 acquisition. Of the internal variables investigated
here, age of L2 onset correlated with L2 acquisition only in the Russian-Hebrew
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
cohort which had a wider range of L2 onset (0-66 months in Israel compared with
12-46 months in Germany). at is, while most of the Russian-German speaking
children were exposed to German within the critical period (with age of L2 onset
not exceeding three years and ten months), a large number of Russian-Hebrew
speaking children were exposed to Hebrew aer the age of three (N=29) and up
to the age of ve years and four months, while others were exposed almost from
birth (N=12). us, the Russian-Hebrew data (but not the Russian-German data)
provide insight into the crucial inuence of age of L2 exposure on L2 acquisition.
Our ndings suggest that children who are exposed to L2 within the critical pe-
riod for language acquisition (as the majority of the Russian-German children are)
do not show an inuence of age of L2 onset, being early child L2 acquirers (Meisel,
2008; Schwartz, 2004). e wide variability in terms of age of L2 onset within the
critical period in the Russian-Hebrew cohort (12 at birth, 10 prior to 24mo, and
22 from 24-36mo) further support this nding. While no signicant dierence
between simultaneous and early sequential bilinguals was found in the Russian-
Hebrew cohort, the performance in the L2 of children whose rst exposure to the
L2 was aer the critical period (a large number of the Russian-Hebrew children)
was sensitive to age of L2 onset, showing a gap from those who began acquiring
the L2 within the critical period.
Despite this dierence in AoO, LoE correlated with the L2 performance in
both cohorts. For the Russian-Hebrew children, LoE correlated with age of L2
onset and both of these variables inuenced L2 acquisition and in fact were
inseparable. In contrast, the Russian-German cohort, with its narrower range for
age of L2 onset, strengthens the impact of LoE where more variation was found.
External environmental variables – L1 maintenance and L2 acquisition. Of the ex-
ternal variables studied, only family size played a role in both cohorts, with single-
ton children performing better in L1/Russian in both cases. Within the Russian-
Hebrew cohort, where SES was rather homogeneous, this nding can be explained
by the monolingual nature of the home. Many families of Russian origin try to
maintain a Russian-only language policy at home in order to support L1 main-
tenance. is is easier to implement with singletons but more dicult with more
siblings who communicate in the L2 at home. While this account could explain
the ndings in the Russian-Hebrew cohort, the Russian-German cohort requires a
dierent explanation. In this cohort, family size correlated with mother’s occupa-
tion and mother’s education, and both strongly correlated with L1 maintenance.
at is, the relative success of singletons in L1 maintenance in this cohort was
not necessarily due to home language policy, but rather due to mother’s SES, as
is commonly reported in monolingual acquisition (Hart & Risley, 1995; Locke &
Ginsberg, 2003; Ravid, 2009).
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
SES played no role in the Israeli cohort in L2 acquisition, but showed strong
correlations with the L2 in the German cohort. e German cohort presented a
wider range of SES, including unemployed mothers and parents with fewer years
of education, thus oering a window into the impact of SES on L2 success, as mea-
sured by parents’ occupation. Here again the children of mothers and fathers with
more professional occupations outperformed the children of unemployed moth-
ers and semi-skilled fathers. is well recorded inuence of SES could be traced
to quality of input in higher SES groups (Ravid, 2009) on the one hand, and to
parents’ prociency in the L2 on the other (Oller & Eilers, 2002).
To conclude, the present study contributes to our understanding of the role of
internal and external factors in L1 maintenance as well as L2 acquisition. It shows
that SES inuences both L1 and L2, while age of L2 onset and length of L2 expo-
sure inuence only L2 acquisition and have virtually no relation to L1 mainte-
nance (but see Gupol 2010). Methodologically, the study shows the importance of
using a variety of language tasks in dierent linguistic domains to evaluate dier-
ent structures. It showed that sentence repetition, a well documented measure for
identifying language impairment, seems not to show high sensitivity to internal
and external variables which inuence typical bilingual acquisition. In addition,
SC and naming abilities are highly sensitive to both internal and external factors
in typical bilingual development. A possible implication of this nding is that pic-
ture naming tasks might be less appropriate for identifying language impairment
among bilingual children, while SR might be a better way of achieving the same
goal. Further research is necessary however to support this latter claim.
Acknowledgements
e research reported here was supported by the BMBF funded Consortium “Migration and
Societal Integration”, Grant No. 01UW0702B. We would like to thank Carmit Altman (BIU),
Zhanna Burstein (BIU), Annegret Klassert (ZAS), Hadar Oz (BIU), and Nathalie Topaj (ZAS)
for their contribution to data collection and analysis.
Note
. Analyses of the external factors for the 5- and 6-year-olds did not yield results which diered
from those found for the whole group and were thus deemed redundant.
© 2011. John Benjamins Publishing Company
All rights reserved
Sharon Armon-Lotem, Joel Walters and Natalia Gagarina
References
Abutbul-Oz, H. (2009). Person perception experiments on language preferences of Russian-
Hebrew sequential bilinguals. Paper presented at ISB7, Utrecht.
Armon-Lotem, S, Danon, G. & J. Walters. (2008). e use of prepositions by bilingual SLI chil-
dren: e relative contribution of representation and processing. In A. Gavarro & M.J. Frei-
tas, (Eds.), Language acquisition and development: Proceedings of the Generative Approaches
to Language Acquisition (GALA) (pp. 41-46). UK:Cambridge Scholars Publishing.
Armon-Lotem, S., Adam, G., Saiegh-Haddad, E., & J. Walters. (2008). Verb inection as indica-
tors of Bilingual SLI: Task eects and Cross-Linguistic Dierences. In T. Marinis, A. Papan-
geli& V. Stojanovik (Eds.), Child Language Seminar Electronic Proceedings (pp. 26-37). UK:
University of Reading.
Bloom, P. (2000). How Children Learn the Meanings of Words. Cambridge, MA: MIT Press.
Chiat, S., Armon-Lotem, S., Marinis, T., Polisenska, K., Roy, P., & See-Gabriel, B. (under re-
view). e potential of sentence imitation tasks for assessment of language abilities in se-
quential bilingual children. In V. C. Mueller Gathercole (Ed.), Bilinguals and assessment:
State of the art guide to issues and solutions from around the world. Multilingual Matters.
Conti-Ramsden, G., Botting, N., & Faragher, B. (2001). Psycholinguistic markers for specic
language impairment. Journal of Child Psychology and Psychiatry, 6, 741-748.
De Houwer, A. (2000). Children’s linguistic environments: a rst impression. In M. Beers, B. van
den Bogaerde, G. Bol, J. de Jong & C. Rooijmans (Eds.), From Sound to Sentence. Studies
on First Language Acquisition (pp. 57-68). Groningen: Centre for Language and Cognition.
Dromi, E., Leonard, L., Adam, G. & Zadunaisky-Ehrlich, S. (1999). Verb Agreement Morphol-
ogy in Hebrew-Speaking Children with Specic Language Impairment. Journal of Speech,
Language and Hearing Research, 42, 1414-1431.
Dunn, L.M. & Dunn, L.M. (1997). Peabody Picture Vocabulary Test, ird Edition. Bloomington,
MN: Pearson Assessments.
Fenson, L., Dale, P. S., Reznick, J. S., Bates, E., al, D. J. & Pethick, S. J. (1994). Variability in
early communicative development. Monographs of the Society for Research in Child Devel-
opment, 59, Serial No. 242.
Friedmann, N., & Lavi, H. (2006). On the order of acquisition of A-movement, Wh-movement
and V-C movement. In A. Belletti, E. Bennati, C. Chesi, E. Di Domenico, & I. Ferrari (Eds.),
Language Acquisition and Development (pp. 211-217). Newcastle, UK: Cambridge Scholars
Press.
Gentner, D. (2006). Why verbs are hard to learn. In K. Hirsh-Pasek, & R. Golinko, (Eds.), Ac-
tion meets word: How children learn verbs (pp. 544-564). Oxford University Press.
Gleitman, L. R., & Newport, E. L. (1995). e invention of language by children: Environmental
and biological inuences on the acquisition of language. In D. N. Osherson (Ed.), An Invi-
tation to Cognitive Science: Vol 1. Language (2nd ed., pp. 1-24). Cambridge, MA: MIT Press.
Goralnik, E. (1995). Goralnik Diagnostic Test. Even Yehuda, Israel: Matan (in Hebrew).
Grimm, H. (2003). Sprachstandscreening fuer das Vorschulalter. Göttingen : Hofgrefe.
Hakuta, K. & E.E. Garcia. (1989). Bilingualism and education. America Psychologist, 44 (2), 374-
379.
Hart, B., & Risley, R. T. (1995). Meaningful Dierences in the Everyday Experience of Young
American Children. Baltimore: Paul H. Brookes.
© 2011. John Benjamins Publishing Company
All rights reserved
e impact of internal and external factors on linguistic performance
Ho, E. (2006). How social contexts support and shape language development. Developmental
Review, 26, 55-88.
Ho, E., Core, C., Place, S., Rumiche, R., Señor, M. & Parra, M. (in press). Dual Language Expo-
sure and Early Bilingual Development. Journal of Child Language.
Kauschke, C. (2007). Erwerb und Verarbeitung von Nomen und Verben. Tübingen: Niemeyer
(Reihe Linguistische Arbeiten).
Klassert A. & Gagarina N. (2010). Der Einuss des elterlichen Inputs auf die Sprachentwicklung
bilingualer Kinder: Evidenz aus russischsprachigen Migrantenfamilien in Berlin. Diskurs
Kindheits- und Jugendforschung, 4, 413-425.
Lambert, W. E., & Taylor, D. M. (1996). Language in the lives of ethnic minorities: Cuban–
American families in Miami. Applied Linguistics, 17, 477–500.
Locke, A. & Ginsberg, J. (2003). Spoken language in the early years: the cognitive and linguistic
development of three- to ve-year-old children from socio-economically deprived back-
grounds. Educational and Child Psychology, 20 (4), 68-79.
Meisel, J. M. (2008). Child second language acquisition or successive rst language acquisition?
In B. Haznedar, & E. Gavruseva (Eds.), Current Trends in Child Second Language Acquisi-
tion: a Generative Perspective (pp. 55–80). John Benjamins.
Oller, D. K., & Eilers, R. E. (Eds.). (2002). Language and Literacy in Bilingual Children. Clevedon:
Multilingual Matters.
Paradis, J. (2009). Oral language development in French and English and the role of home input
factors. Report for the Conseil Scolarie Centre-Nord, Edmonton, Alberta.
Pearson, B. Z. (2007). Social factors in childhood bilingualism in the United States. Applied
Linguistics, 28, 399-410.
Porat, S. (2009). e use of structures involving syntactic movement by English-Hebrew Bilin-
gual children with SLI. Poster presented at the Research Conference on Bilingualism and
SLI, Jerusalem, Israel.
Ravid, D. (2009). Morphological knowledge in children and adolescents with SLI: e impact
of development and SES. Paper presented at the Research Conference on Bilingualism and
SLI, Jerusalem, Israel.
Rummer, R., & Engelkamp, J. (2001). Phonological information contributes to short-term recall
of auditory presented sentences. Journal of Memory and Language, 45, 451-467.
Schwartz, B. D. (2004). Why child L2 acquisition? In J. Van Kampen & S. Baauw (Eds.), Proceed-
ings of Generative Approaches to Language Acquisition (GALA) (pp. 47-66). Utrecht: LOT.
See-Gabriel, B., Chiat, S., & Roy, P. (2008). Early Repetition Battery. London: Pearson Educa-
tion, Inc.
Tucker, G.R. (1999). A Global Perspective on Bilingualism and Bilingual Education. Washington,
DC: Center for Applied Linguistics.
Authors’ addresses
Sharon Armon-Lotem and Joel Walters
Department of English and the Gonda Multi-
disciplinary Brain research Center
Bar-Ilan University
Ramat Gan 59200
Israel.
E-mail: armonls@mail.biu.ac.il
Natalia Gagarina
Zentrum fuer Allgemeine Sprachwissenscha
(ZAS)
Berlin.
