How Much Exposure To English Is Necessary For A Bilingual Toddler To Perform

INT J LANG COMMUN DISORD, NOVEMBER–DECEMBER 2014,
VOL. 49, NO. 6, 649–671
Research Report
How much exposure to English is necessary for a bilingual toddler to perform
like a monolingual peer in language tests?
Allegra Cattani†, Kirsten Abbot-Smith‡, Rafalla Farag†, Andrea Krott§, Frédérique Arreckx¶, Ian Dennis†
and Caroline Floccia†
†School of Psychology, University of Plymouth, Plymouth, UK
‡School of Psychology, University of Kent, Canterbury, UK
§School of Psychology, University of Birmingham, Birmingham, UK
¶Speech and Language Therapist, Tours, France
(Received May 2013; accepted December 2013)
Abstract
Background: Bilingual children are under-referred due to an ostensible expectation that they lag behind their
monolingual peers in their English acquisition. The recommendations of the Royal College of Speech and
Language Therapists (RCSLT) state that bilingual children should be assessed in both the languages known by the
children. However, despite these recommendations, a majority of speech and language professionals report that
they assess bilingual children only in English as bilingual children come from a wide array of language backgrounds
and standardized language measures are not available for the majority of these. Moreover, even when such measures
do exist, they are not tailored for bilingual children.
Aims: It was asked whether a cut-off exists in the proportion of exposure to English at which one should expect
a bilingual toddler to perform as well as a monolingual on a test standardized for monolingual English-speaking
children.
Methods & Procedures: Thirty-five bilingual 2;6-year-olds exposed to British English plus an additional language
and 36 British monolingual toddlers were assessed on the auditory component of the Preschool Language Scale,
British Picture Vocabulary Scale and an object-naming measure. All parents completed the Oxford Communicative
Development Inventory (Oxford CDI) and an exposure questionnaire that assessed the proportion of English in
the language input. Where the CDI existed in the bilingual’s additional language, these data were also collected.
Outcomes & Results: Hierarchical regression analyses found the proportion of exposure to English to be the main
predictor of the performance of bilingual toddlers. Bilingual toddlers who received 60% exposure to English or
more performed like their monolingual peers on all measures. K-means cluster analyses and Levene variance tests
confirmed the estimated English exposure cut-off at 60% for all language measures. Finally, for one additional
language for which we had multiple participants, additional language CDI production scores were significantly
inversely related to the amount of exposure to English.
Conclusions & Implications: Typically developing 2;6-year-olds who are bilingual in English and an additional
language and who hear English 60% of the time or more, perform equivalently to their typically developing
monolingual peers.
Keywords: bilingualism, lexicon, language delay and assessment.
What this paper adds?

What is known about this subject?
Bilingual children have a smaller vocabulary in each of their two languages than monolingual children and also take a
little longer to reach the same levels as monolinguals on various grammatical tasks. The relative amount of exposure
to each language is strongly related to the children’s rate of development in those languages.
Address correspondence to: Allegra Cattani, Drake Circus, University of Plymouth, Plymouth PL4 8AA, UK; e-mail: [email protected]
International Journal of Language & Communication Disorders
ISSN 1368-2822 print/ISSN 1460-6984 online C 2014 Royal College of Speech and Language Therapists
DOI: 10.1111/1460-6984.12082
14606984, 2014, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/1460-6984.12082 by Washington State University, Wiley Online Library on [30/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
650 Allegra Cattani et al.
What this paper adds?

A questionnaire that measures the proportion of exposure to English in the child’s input for use by health professionals
involved in screening and referral for language assessment. With the use of the English exposure questionnaire,
professionals should be able to interpret the performance of a bilingual toddler on a standardized monolingual test
more accurately. Most importantly, if a bilingual child hears English 60% of the time or more, professionals should
expect this child to perform as well as monolingual children and should use the same criteria for referral as they
would for a monolingual child.
Introduction (health visitors and early years’ practitioners) are not

required to receive specific training in what to expect
The bilingual population is increasing worldwide. Ac- in language development from monolingual let alone
cording to the Office for National Statistics (ONS) bilingual children.
(2011) the number of births to non-UK born moth- Since 2013 all early years’ workers are legally re-
ers in England and Wales has seen a marked rise over quired to judge whether each 2–3-year-old in their care
the last decade. These births accounted for 25.1% of all is at an age-appropriate level for a number of develop-
live births in 2010 compared with 15.5% in 2000. This mental criteria including language. Health visitors are
proportion has increased every year since 1990, when it now legally required to assess 24-month-olds on general
was just under 12%. developmental questionnaires which include language
Given the variability of the bilingual population, subcomponents, such as the revised Ages and Stages
the focus of this paper is towards infant or simultaneous Questionnaires (ASQ) (Squires and Bricker 2009). Re-
bilingual children, i.e. children acquiring two languages grettably, none of the new measures takes bilingual chil-
from birth or very early in their life, coming from a wide dren into account. While early identification is desirable,
array of linguistic backgrounds. It is well-established that concern has been expressed that bilingual children with
simultaneous bilingual children may have smaller vocab- delays or disorders relating to or including language de-
ularies in each of their two languages when compared velopment may be at risk of under-referral in the UK
with monolinguals learning one of those languages in (e.g. Bedore and Peña 2008, Crutchley 2000, Stow and
isolation (Bialystok 2009). They also take a little longer Dodd 2003). Under-referrals can occur for a variety of
to reach the same level as monolinguals on various gram- developmental disorders including pervasive disorders
matical tasks (cf. Gathercole 2007, Nicholls et al. 2011). (ASD), sensory (hearing impairment) and also specific
However, they make up for this in terms of advanced language impairment (SLI), i.e. language impairment
meta-linguistic awareness (cf. Bialystok 2007) and ear- that cannot be accounted for in terms of general intel-
lier development of executive functions (cf. Bialystok lectual disorder, hearing loss or environmental depriva-
2009, Kovacs and Mehler 2009, Poulin-Dubois et al. tion (e.g. Bishop 2006). Crutchley (2000) found in par-
2011, but for a good discussion on positive versus neg- ent interviews that the proportion of bilingual parents
ative effects of bilingualism on cognition, see also Baker (45%) who stated that professionals had initially failed
2011: ch. 7). Nonetheless, amongst the general popu- to diagnose their child’s difficulties or take note of the
lation the impression remains that one should expect parents’ worries was far larger than for monolingual par-
bilingual children to be delayed—even quite dramat- ents (18%). Under-referral has potentially more serious
ically delayed—in early acquisition of language (Stow consequences than over-referral since most professionals
and Dodd 2003) and later development of grammar agree that early identification of at-risk children, even
(Nicholls et al. 2011). without an intervention (for a watch-and-see approach,
Assessing children for early language development is see Paul 1996), is a necessary step to prepare for an
particularly problematic in the UK as children do not intervention (see also Ellis Weismer 2000).
have regular access to a paediatrician. Rather each child
is seen once at 12 months and once at around 2;6 years
by a ‘health visitor’, who is a specialist community public Procedures available to SLPs for assessment of
health nurse within the Nursing and Midwifery Council bilingual children
trained to carry out and interpret a general assessment When bilingual pre-schoolers are referred, the official
of motor, social and language development. In addition, guidelines of both the Royal College of Speech and
children are referred to speech and language profession- Language Therapists (RCSLT) (2007) in the UK and
als (SLPs) if either their parents or their ‘nursery’ (i.e. the American Speech–Language–Hearing Association
kindergarten) teachers flag issues with language devel- (ASHA) (1999) in the United States are for them to
opment. Unfortunately in the UK, early years’ workers be assessed in both of their languages (see also Peña and
Language exposure and bilingual screening 651
Halle 2011). However, two real-world facts make this ized assessment tools in languages other than English is
principle untenable. Firstly, bilingual pre-schoolers, par- limited.
ticularly those of low socioeconomic status and/or first- Another option is to work with an interpreter. This
time parents will not be referred if early years’ workers in has its pitfalls as translation styles may vary from the
the UK are not provided with an easily applicable means general to the literal word for word (Stow and Dodd
for determining whether a particular bilingual child is in 2003) and result in data that are hard to interpret. Such
need of referral. Secondly, even when bilingual toddlers informal, ad-hoc procedures would be considered inap-
are referred to SLPs, the official guidelines are usually propriate as the sole or main source of information on
not followed because assessment in the child’s ‘other’ a child’s language level for a monolingual child (Thor-
language is challenging for a number of reasons. dardottir et al. 2006). For this reason, one recent stream
We briefly review the evidence that the assessment of thought puts a great deal of weight on asking the
of bilingual pre-schoolers in predominantly English- parents of bilingual children to assess how well they
speaking countries, not only in the UK, is problematic think their child is progressing. For instance, Paradis
even for SLPs. SLPs are frequently faced with children et al. (2010) developed a parental questionnaire for par-
for whom no standardized tests are available in the addi- ents of children aged 4;9–9;0, which is designed for use
tional language. Moreover, even when standardized tests with bilingual children from any language background
exist in the additional language, the monolingual norms along with guidelines for scoring in a clinical setting.
of either English or the additional language cannot be However, it was found that some parents were unable to
directly applied to bilingual children as they have been assess their children’s sentence comprehension and did
found to underperform monolinguals on these language not even understand the questions pertaining to this.
tests (e.g. Camilleri and Law 2007, Hemsley et al. 2010, One of the most promising methods is that of
Hoff et al. 2012, Restrepo and Silverman 2001). dynamic assessment (e.g., Camilleri and Law 2007,
One proposal is to use a composite scoring assess- Gutiérrez-Clellan and Peña 2001, Hasson et al. 2013,
ment (or conceptual vocabularies), i.e. the total number Hasson and Joffe 2007). The most common method is
of semantic concepts for which a child has a lexical test–teach–retest, where the difference between initial
form, regardless of which language it is in (Junker and score and later score is compared, after an intervention
Stockman 2002, Pearson et al. 1993). Modern online has been carried out, thus revealing a particular child’s
translation plus the availability of relevant software for potential to learn. If a child initially performs signifi-
calculating composite vocabularies can make this a quick cantly below the mean because of a reduced exposure to
option. However, research into composite methodol- a language, his or her intact learning capacity should be
ogy on bilingual children has produced mixed results revealed by a large difference in pre- and post-test scores
(Bedore et al. 2005, Hemsley et al. 2010, Pearson et al. compared with a child with a reduced language learning
1993, Thordardottir et al. 2006). Thordardottir et al. capacity. While this method appears suitable, its critical
(2006), for example, found that the conceptual vo- drawback is the time and thus the cost involved for the
cabulary of a bilingual group with balanced exposure SLPs.
to both languages was in fact significantly lower than Recently, with the intention of reducing time and
that of the monolingual English group. Pearson et al. cost, the Dynamic Assessment of Preschoolers’ Profi-
(1993), in contrast, found that the composite expressive ciency in Learning English (DAPPLE) was developed in
vocabulary of Spanish–English bilingual children was the UK to respond to the clinical need to distinguish be-
equal to that of a monolingual group. Recently, Hemsley tween disorder and bilingualism due to a child’s language
et al. (2010) carried out a longitudinal investigation of learning context (Hasson et al. 2013). The DAPPLE as-
language development in sequential Samoan–English sessment takes less than 60 min to administer and it
bilingual children during their first year at school. The examines the children’s ability to learn vocabulary, sen-
bilingual composite receptive vocabulary scores were tence structure and phonology. This battery of language
comparable with the monolingual scores but the bilin- skills assessments sounds promising as a pre-diagnostic
gual composite expressive scores were significantly below tool but is not designed for children younger than 42
the scores of monolingual peers. months.
In sum, the procedure of composite scores has great
potential for SLPs but with such mixed findings firm
conclusions cannot be drawn regarding a language disor- Reality of use of the official procedures for assessing
der particularly when the composite score of a bilingual
bilingual children
child is lower than that of monolingual peers. Crucially,
with the exception of the CDI parent reports (Commu- Probably in part because of the time and cost con-
nicative Developmental Inventories, e.g. Fenson et al. straints that SLPs face on the ground, the reality
2007, Hamilton et al. 2000), the availability of standard- of the procedures currently used to assess bilingual
children in English-speaking countries is that the of- bilingual child who is (1) dominant in English and (2)
ficial recommendations are not strictly followed by pro- performing significantly below the monolingual norms
fessionals. Caesar and Kohler (2007) surveyed 409 SLPs would be in need of a referral to SLPs. That would con-
in the United States, and found that 130 of these had at stitute a first step towards an early identification of bilin-
least one bilingual child on their caseload within an age guals at risk (excluding at this stage the case of those who
range from mid-preschool years up to secondary school. are non-English dominant—which will be addressed in
While 63% of the respondents mentioned Spanish as the general discussion). This practical approach would
one of the second languages, they also listed 33 other lan- enable the development of a systematic language mea-
guages. Only 48% mentioned using interpreters some sure easily used by early years’ workers, who can then, if
of the time. Some mentioned taking language samples, necessary, refer to SLPs. These would hopefully have re-
but these were predominately samples of the children’s duced caseloads as a result of not having to carry out the
spoken English and not of their additional language. initial screening process which would free up their time
The predominant method of assessing bilingual chil- to implement intervention methods such as dynamic
dren was through formal standardized procedures and assessment.
75% said that English was the language of the test or The first step was to develop a questionnaire which
procedure they used most frequently. The Peabody Pic- estimates the percentage of time a bilingual child hears
ture Vocabulary Test (PPVT) (Dunn and Dunn 1997) English. Bilingual children do not hear and use each of
and the Clinical Evaluation of Language Fundamentals their languages as frequently as monolingual peers who
3 (CELF) (Wiig et al. 1992) were the most frequently speak one language, and bilingual children have differ-
mentioned tests and very few reported using the Spanish ent experiences with their two languages that could lead
adaptation of the PPVT or any other test with norms to different outcomes. Indeed, it has been established
in the child’s additional language or with bilingual that the relative amount of exposure in each language
norms. is strongly related to the children’s rate of development
Stow and Dodd (2003) painted a similar picture in those languages (Gathercole and Thomas 2009, Hoff
of referral and therapy for bilingual children in the et al. 2012, Pearson et al. 1997, Scheele et al. 2010,
UK. In Australia, the situation is by no means differ- Thordardottir 2011). The variable of amount of expo-
ent (Williams and McLeod 2012). On a questionnaire sure has often been described in terms of the languages
returned by 128 SLPs, English was reported to be the spoken at home versus in the school. Recent attempts
primary language for assessment and intervention; three have been made to calculate the English input of 2-year-
quarters of them assessed the children using only En- old children by gathering more detail, for example using
glish standardized tests. Like the American SLPs study, a prospective language diary (De Houwer 2009, Hoff
many of them assessed children’s speech (77%) and lan- et al. 2012, Parra et al. 2011, Place and Hoff 2011). In
guage (34%) without assistance from an interpreter or this document, caregivers kept a log diary over a course
a family member. Although about half of the Australian of 7 days which measured the percentage of 30-min peri-
SLPs in the sample had at least minimal competence in ods when the child heard each of the two languages, the
a language other than English, the languages spoken by type of interaction context and the number of speakers.
these SLPs rarely matched the primary languages spoken Thordardottir (2011) chose to calculate relative amount
by the children on the SLPs’ caseload. of language exposure through a detailed questionnaire
completed by parents to assess the child’s exposure to
each language in situations with potential communica-
Aims
tive interactions, with data spanning a continuum of
Considering all the aforementioned obstacles, and given bilingual exposure. In the current study, we developed
that the waiting list in the UK to see an SLP, even after our own questionnaire which obtains precise calcula-
referral, is up to 12 months in some regions, a move tions of the percentage of language exposure of a child
towards basic screening by non-SLP professionals would (see appendix A).
seem a pragmatic move. Our starting idea was to develop Our second step was to include simultaneous bilin-
a measure with which an early years’ worker would be gual toddlers from a variety of language backgrounds, as
able, through discussion with the parent, to determine, long as one of their languages was English, as that is the
firstly, the percentage of time the child hears English situation which SLPs face in the UK, the United States,
and secondly, the child’s performance on the British Australia and Canada (e.g. Caesar and Kohler 2007,
English version of the MacArthur-Bates CDI (Fenson Paradis et al. 2010, Stow and Dodd 2003, Williams and
et al. 2007). McLeod 2012). We focused on simultaneous bilingual
Our working hypothesis was that simultaneous children who use two languages in either the receptive
English–dominant bilinguals might be expected to score and expressive modality. Finally, we restricted our focus
within monolingual norms on the CDI. Therefore a to children aged between 28 and 32 months, which is
the earliest age at which children are referred to speech 1.2). Most of the children were first (60%) or second
therapy in most areas of the UK. (30%) born, whilst the remaining children were third
The third step was to establish empirically what level (7%) or fourth (3%) born. Socioeconomic status of fam-
of English language development should be expected, ilies was calculated by educational achievement of each
given the proportion of time a given bilingual toddler parent as well as by parental occupation rated on the
hears English. We investigated whether the scores of UK nine-point scale proposed by Hollingshead (1975, cited
bilingual children on various production and compre- in Bornstein et al. 2003). The educational level of most
hension measures could be predicted by the proportion families was high (81%, i.e. degree and above) or mid-
of exposure to English (as compared with the additional dle (16%, i.e. completion of secondary school) status.1
language) experienced during a typical week. The cru- Information about occupational status appeared more
cial question was whether one could identify a minimal discriminating as the scores ranged from 3 to 9 with
cut-off point of English exposure above which the group a mean at 7.6. The latter did not differ between the
of bilingual children would perform similarly to mono- two groups. There was no difference between the bilin-
lingual peers, and over which at-risk children could be gual and monolingual households in the distribution of
easily identified. mothers or fathers across the levels of occupational sta-
We compared our sample of bilingual children to a tus and education (t(67) = 1.51, p = 0.14 and t(67) =
sample of monolingual children matched on age, gen- 1.20, p = 0.23, respectively).
der, parental occupational and educational level. Our The bilingual children were all born and raised in
rationale was firstly to replicate previous findings that England and were exposed to no more than two lan-
bilingual 2;6-year-olds as a group will indeed differ from guages. Among the bilingual children, 20 had both par-
their monolingual peers on language assessment tests, if ents speaking the same additional language to the child,
tested on one language only. Secondly we wished to pro- nine had only the father and six had only the mother
vide a direct benchmark for determining the language speaking the additional language. The additional lan-
exposure threshold above which bilinguals are compara- guages spoken were Arabic (13), French (3), Punjabi
ble with monolinguals. (2), Italian (2), Spanish (2), Catalan, German, Greek,
Irish Gaelic, Dutch, Finnish, Polish, Albanian, Czech,
Kurdish, Afrikaans, Swahili and Mandarin.
Methods
Participants English exposure questionnaire
Eighty-six children living in Plymouth, East Kent and We used a self-report English exposure questionnaire
the Birmingham area were recruited through the Uni- devised to obtain an objective estimate of the average
versity databases of the Plymouth Baby Lab, Kent Child proportion of the time a child hears English and the
Development Unit and Birmingham Infant and Child additional language during a typical week that did not
Laboratory. Forty-two were native British monolingual involve holiday periods such as Christmas break during
English-speaking and 44 were simultaneous bilingual the last year of the life of the child. In appendix A the
children born in the UK. There were two sets of ex- example questionnaire has been filled out by a French-
clusion criteria. The first set, which applied equally to speaking mother and an English father with a 30-month-
monolingual and bilingual children, resulted in the ex- old girl.
clusion of eight children: parents reported a diagnosis Section A identifies the number of language/s spo-
of hearing impairments or a speech delay (two chil- ken at home2 and accordingly directs the respondent
dren); premature birth (one child); children either did to subsections B or C or D, which are similar with
not cooperate or their parents intervened during the test the exception of the initial question. Section B as-
(two children); children did not complete three or more sesses bilingual children whose parents both speak the
assessment measures (three children). The second set same additional language at home (e.g. mother and fa-
of exclusion criteria applied only to bilingual children. ther both speak Russian). Section C is addressed to
These resulted in four children being excluded because the families in which one parent speaks English to
their parents did not fill in the English exposure ques- the child and the other parent speaks an additional
tionnaire and an additional three were excluded because language.
their parents rated them as having nearly 100% input These sections ask questions about the average num-
in one language. ber of hours per week a child spends in an English speak-
The final sample comprised 36 British monolingual ing childcare environment (nursery, day care, preschool,
(21 girls) and 35 bilingual (17 girls) children aged 28– child-minder, relative or friend) and the number of
32 months (for monolingual children: mean = 30.47, hours the child spends sleeping per 24 h. Other ques-
SD = 1.2; and for bilinguals: mean = 30.21, SD = tions ask how often the mother and the father talk to the
child in English as opposed to the additional language The auditory component of the Preschool Language Scale
when on their own with the child using a five-point ordi- 4 (PLS 4) (Zimmerman et al. 2009)
nal scale (e.g. always, usually, half the time); who speaks
more to the child when the two carers are together; and This test, which covers a range of English language
the number of hours per week that a child spends time skills normed for monolingual children between birth
with each parent individually. Parents are asked to report and 6 years, is designed to identify children with a
the number of hours they spend with their child rather language disorder. The PLS 4 consists of two compo-
than being asked to estimate the percentage of time their nents, an auditory and a production component. The
child hears English. Indeed it has been found that people auditory component contains sentence processing items
fail to keep detailed records of who is talking to the child along with some vocabulary items. The PLS 4 is fre-
in what language and in fact parents are rather poor esti- quently used by SLPs in the UK and USA with this age
mators of their own efforts and abilities (e.g. Kruger and group.
Dunning 1999).
Based on this information, calculations (see the
computations in appendix B) estimate the number of Object naming sub-task (adapted English SETK-2)
English-hearing hours per week. Scale responses (e.g.
whether the mother speaks English to the child always, The language test Sprachentwicklungstest—2 (SETK-
most of the time, half/half, rarely or never) are con- 2; Grimm 2000) was originally designed and standard-
verted into percentages (here 100%, 75%, 50%, 25% ized in German to measure receptive and expressive lan-
and 0%) which are used to recalculate the number guage skills in 24–36-month-old German children, and
of English hours. For example if the mother spends is divided into four subtests, of which only the ob-
10 h a week on her own with her child, and speaks ject naming sub-task (see appendix C for the English
to her mostly in English, then these 10 h will be- translation)3 was used. This object-naming task consists
come 7.5 h of English and 2.5 h of the additional of 30 items, of which the first six are actual objects. The
language. latter are matched to the original items in the German
The last section D was filled in by both monolin- version. The remaining 24 items are colour pictures,
gual and bilingual parents and provides details of the which were photocopied from the German test, except
date and place of birth of the child, the highest qual- the item ‘petrol station’ which has been replaced as it
ification of the mother and of the father along with did not look like an English petrol station. For each
their current occupations, the length of time living in item, the children were asked ‘What’s this?’ and were
the UK and the presence and number of younger or given a score of 1 if the child offered any of the English
older siblings. It also assesses any known developmental words given as options for that item. If the child gave
issues such as being born six or more weeks prematurely, a response which was not on the list (e.g. ‘egg’ for ball
hearing difficulties or more general developmental or ‘apple’ for ‘pear’) or in the other language, this was
delays. scored as 0.
Oxford Communicative Development Inventory (Oxford

Language skill tests
CDI) (Hamilton et al. 2000)
The British Picture Vocabulary Scale III (BPVS III)
This measure uses parental report to assess compre-
(Dunn et al. 2009)
hension and production of 416 early English words.
BPVS III is a receptive vocabulary test for Standard Since our ultimate long-term goal is to be able to es-
British English between 2 years 6 months and 16 years tablish a level of English development using a basic lan-
11 months. It is the British version of the American guage measure usable by early years’ workers in consul-
equivalent the PPVT-III which has been found to be one tation with parents, an ideal measure would have been
of the most frequently used tools for diagnosis of bilin- a previously standardized British version of the Amer-
gual children in the United States (Caesar and Kohler ican MacArthur–Bates Communicative Development
2007). Each item consists of four colour illustrations on Inventory (Fenson et al. 2007) which not only asks par-
a plate and the task of the children is to select the picture ents about their child’s vocabulary but also about their
that illustrates the meaning of a word said by the test child’s expressive morphology and multi-clause utter-
administrator. The test ends at the ceiling set that is es- ances. Unfortunately, to date there is no standardized
tablished when a child made eight or more consecutive British version of this tool. We therefore chose the Ox-
errors within a subset. The scores are calculated as the ford CDI since results using this version have been pub-
number of correct responses. lished (Hamilton et al. 2000).
Additional language CDI group of children from the individual child score; this
difference was then divided by the group standard devi-
For the assessment of vocabulary size in the additional
ation. This means that z-scores for all the children are
language a version of the additional language CDI-tool,
on the same scale and hence that z-scores from bilingual
where available, was handed to the parent. The rationale
and monolingual children can be meaningfully com-
for this is to examine whether there is an (inverse) corre-
pared. Standardized normative scores were not used for
lation between vocabulary development in English and
the BPVS III and PLS 4 since the normative data for
in the additional language, and importantly whether the
BPVS III start from 36 months which is above the age of
amount of exposure to English predicts the vocabulary
our children. On the PLS four children fell right in be-
scores in both CDI versions.
tween the age bands for the standardized norms, which
The additional language CDIs included versions in
in this test are grouped separately for 24–29 and 30–35
Arabic (Safi, personal communication and in progress),4 months. Therefore, if normative scores had been used
French (Kern 2007), Italian (Caselli and Casadio the youngest children would have achieved higher scores
1995), Spanish (López Ornat et al. 2005), Catalan than our oldest children.
(Catalan, in progress and personal communication, Finally, for the CDIs Comprehension and Produc-
Serrat et al. 2010), German (Szagun et al. 2009), tion in English and the additional languages, percentage
Gaelic Irish (O’Toole and Fletcher 2010), Dutch (Zink scores were used, mainly because it was not possible
and Lejaegere 2002), Finnish (Lyytinen 1999), Polish to transform the scores on the CDIs in the additional
(Smoczynska, in progress, personal communication),4 language into z-scores due to the small sample in each
Slovak (Kapalková et al. 2010),5 and Mandarin (Tardif particular additional language and also because of the
et al. 2008) languages. For the two Punjabi bilingual variability in the total number of items for each lan-
children, the Punjabi–English bilingual tester translated guage (see below for a discussion).
the Oxford CDI into Punjabi.6
Procedure Comparison between monolingual and bilingual

Almost all children were seen twice by monolingual children
British English-speaking research assistants with no Before we discuss the main analyses which look at the
more than 10 days between the two visits. Prior to the English language development of bilinguals as a func-
first visit, all parents were sent the Oxford CDI (and the tion of the proportion of English in the input, we first
CDI in the additional language when available) to be present global comparisons between the monolingual
filled in at home and asked to observe their child for a control group and the bilingual children with all input
few days. During the first visit the children were assessed levels conflated. Independent two-sample t-tests were
on the PLS 4 Auditory Comprehension test. During the conducted to compare the scores of the English mono-
second visit, the parents returned the Oxford CDI and lingual children and the bilingual children on the En-
the self-report English exposure questionnaire (which glish receptive and expressive language assessments (with
was discussed with the researcher on the day of the test- equal variances not assumed; we shall come back to this
ing), whilst the children were assessed on the BPVS III later in the analyses).
and English SETK-2. Each testing session lasted 30 min, As can be seen in table 1, bilingual children’s re-
and if testing was not completed during this time, the ceptive vocabularies in English as assessed by the BPVS
parent and child returned for a third visit. III were on average significantly lower than those of
the monolingual children, t(56.4) = 3.94, p < 0.001;
mean diff. = 0.89, Cohen’s d = 0.98. Similarly, the
Results
expressive vocabulary scores as assessed by the English
Table 1 presents the mean values and standard devia- SETK-2 were significantly lower for the bilingual group
tions (SDs) of the bilingual and monolingual groups as a whole than for the monolingual controls, t(33.4) =
in terms of demographic data (age, gender, birth rank, 4.36, p < 0.001; mean diff. = 1.0, d = 1.12. Both
parental occupation and education) and language as- the Oxford CDI comprehension and production scores
sessment measures (BPVS III, PLS 4, English SETK-2, were also significantly lower for the bilingual children
Oxford CDI). Raw scores were obtained as measures than for the monolingual children, t(35.3) = 3.93,
for the BPVS III, PLS 4, and English SETK-2, and p < 0.001; mean diff. = 86.82, d = 0.95 and t(42.0) =
transformed into z-scores for all analyses, although for 5.36, p < 0.001; mean diff. = 130.1, d = 1.30, for word
clarity, some figures and tables are presented with raw comprehension and production respectively. Finally, for
scores. Z-score transformations were carried out by sub- the PLS 4 test, which assesses English comprehension,
tracting the mean value on a language test of the whole the bilingual children scored lower on average than the
Table 1. Mean values of demographic variables together with mean raw scores and z-scores
Monolingual Bilingual
Raw scores (SD) z-scores (SD) N Raw scores (SD) z-scores (SD) N
Age (months) 30.21 (1.16) 36 30.21 (1.16) 35
Proportion of girls 58.3% 36 48.6% 35
Birth rank 1.36 (0.59) 34 1.71 (0.87) 35
Parent’s occupation 7.86 (1.26) 34 7.35 (1.49) 35
Parents’ education 2.71 (0.46) 34 2.85 (0.50) 35
BPVS III 37.68 (9.88) 0.42 (0.78) 34 26.45 (12.76) −0.47 (1.03) 31
PLS 4 Auditory 41.19 (6.02) 0.15 (0.96) 31 38.90 (6.48) −0.21 (1.03) 21
English SETK-2 25.70 (2.48) 0.48 (0.35) 33 18.63 (8.54) −0.52 (1.21) 30
Oxford CDI Comprehension 399.29 (24.20) 35 312.47 (126.67) 34
Oxford CDI Production 372.71 (50.26) 35 242.62 (132.69) 34
Note: Mean values of demographic variables together with mean raw scores and z-scores, when appropriate, are shown for the BPVS III, PLS 4 Auditory, English SETK-2, Oxford
CDI word comprehension and word production scores of English monolingual and bilingual children (standard deviations are shown in parentheses). Parents’ occupation is provided
on a scale from 1 to 9 following Hollingshead (1975) (cited by Bornstein et al. 2003) and parental education on a three-point scale from 1 to 3. Missing values are due to unavailable
data (parents failing to provide a questionnaire or child refusing to participate).
monolingual children but this difference was not signif- exposure could be used to predict the vocabulary skills
icant, t(40.9) = 1.29, p = 0.2; mean diff. = 0.37). of bilingual children, after accounting for demographic
Our finding that as a group, bilingual children scored data.7 In the first block, predictor variables comprised
lower than monolinguals in terms of lexical knowledge, the age in months, birth rank in the family, gender of
when assessed on one of their languages, is not surpris- child, parent’s education and occupation scores. In the
ing and fits with the previous literature (e.g. Junker and second block, percentage of exposure to English was in-
Stockman 2002, Hoff et al. 2012, Gathercole 2002, troduced. Table 2 displays the percentage of variance
Thordardottir et al. 2006). Furthermore, on all mea- explained in each block for each language measure to-
sures, bilingual toddlers displayed higher variances than gether with the parameters of the regression equations.
monolingual ones. This would be expected from a par- For the regression analyses performed on the lan-
ticipant sample whose language development depends guage comprehension/auditory measures tests (BPVS
on additional characteristics as compared with monolin- III, PLS 4 and CDI Comprehension), the demographic
guals (including, as we will see, the amount of exposure measures did not account for a significant proportion
to each language). In addition, for some measures (Ox- of variance, but in the second block the proportion of
ford CDI comprehension and SETK-2), some mono- English exposure was a significant predictor (marginally
lingual toddlers might have reached the ceiling levels, for PLS 4 though) (table 2 and figure 1) and explained
as reflected not only in higher scores but also in lower a significant amount of variance (an additional 24.1%
associated variance as compared with bilinguals. for the BPVS III, 11.6% for the PLS 4, and 26.4% for
A more interesting question, however, is the degree the CDI Comprehension). The regression analyses for
to which their performance depends on the proportion the two tests of lexical production (the English SETK-2
of their English input. We examined how the amount and the Oxford CDI Production) both revealed a signif-
of English exposure would predict the scores on the icant impact of demographic variables in the first block
language assessment tests, once corrected for the effect (43.2% variance explained for the SETK-2 and 36%
of demographic variables such as gender, age, birth order for the CDI Production), mainly due to the age vari-
and SES. able (SETK-2: standardized β = 0.47, t(22) = 2.84,
p = 0.009; CDI Production: standardized β = 0.51,
t(26) = 3.13, p = 0.004). In the second block, propor-
Relationship between English vocabulary skills, tion of exposure to English significantly explained an
demographic variables and English exposure additional 15.9% (SETK-2) and 30.4% (CDI Produc-
tion) of the variance.
Bilingual children spanned the full range of the propor-
tion of English in the input which can be found amongst
bilingual toddlers in modern Britain: the proportion of Identifying the percentage amount of English
English in their input ranged between 5% and 98% in exposure needed to perform within a monolingual
a typical week during the last year of the life of the child
range
(M = 58.23; SD = 26. 67).
Hierarchical regression analyses were carried out for In summary, so far the proportion of English in the
each language measure to determine whether English language input to a bilingual 2;6-year-old predicts his or
Table 2. Summary of the hierarchical regression analyses for the BPVS III, PLS 4 Auditory, English SETK-2, Oxford CDI word comprehension and word production scores of
bilingual children
BPVS III PLS 4 SETK-2 CDI Comp CDI Prod

Language exposure and bilingual screening
Coefficient Coefficient Coefficient Coefficient Coefficient

(SE coefficient) p (SE coefficient) p (SE coefficient) p (SE coefficient) p (SE coefficient) p
Block 1 Age 0.34 (0.41) 0.04 0.27 (0.31) n.s. 0.47 (0.47) 0.009 29.23 (0.27) n.s. 56.86 (0.51) 0.004
Birth rank −0.18 (−0.15) n.s. −0.05 (−0.03) n.s. −0.44 (−0.31) 0.065 −17.85 (−0.12) n.s. −39.21 (−0.25) n.s.
Gender −0.17 (−0.08) n.s. 0.16 (0.08) n.s. 0.04 (0.02) n.s. 2.54 (0.01) n.s. −42.27 (−0.16) n.s.
Education −0.13 (−0.06) n.s. 0.53 (0.25) n.s. 0.32 (0.14) n.s. 46.81 (0.22) n.s. 14.32 (0.06) n.s.
Occupation 0.14 (0.19) n.s. 0.22 (0.41) n.s. 0.17 (0.22) n.s. −0.37 (−0.004) n.s. 13.95 (0.16) n.s.
R2 0.207 0.482 0.431 0.154 0.36
F F(5,30) = 1.31 n.s. F(5, 19) = 2.61 0.072 F(5, 28) = 3.49 0.017 F(5, 32) < 1 n.s. F(5, 32) = 3.04 0.026
Block 2 Age 0.25 (0.31) 0.07 0.29 (0.34) 0.09 0.37 (0.37) 0.019 18.56 (0.17) n.s. 45.09 (0.40) 0.003
Birth rank −0.05 (−0.04) n.s. 0.02 (0.01) n.s. −0.27 (−0.19) n.s. 9.01 (0.06) n.s. −9.49 (−0.06) n.s.
Gender −0.32 (−0.15) n.s. 0.19 (0.09) n.s. 0.03 (0.01) n.s. −9.47 (−0.04) n.s. −55.56 (−0.21) 0.082
Education −0.41 (−0.20) n.s. −0.15 (−0.07) n.s. −0.02 (−0.01) n.s. 18.15 (0.07) n.s. −28.46 (−0.11) n.s.
Occupation 0.01 (0.02) n.s. 0.27 (0.50) 0.09 0.11 (0.15) n.s. −6.71 (−0.08) n.s. 6.94 (0.08) n.s.
Exposure 0.02 (0.57) 0.003 0.02 (0.43) 0.07 0.02 (0.47) 0.008 2.79 (0.59) 0.002 3.09 (0.63) < 0.0001
R2 0.448 0.598 0.591 0.418 0.664
F F(6, 30) = 3.25 0.018 F(6, 19) = 3.23 0.036 F(6, 28) = 5.29 0.002 F(6, 32) = 3.11 0.02 F(6, 32) = 8.55 < 0.0001
2
Note: For each block of variables entered in the regression and for each measure the value of the β-coefficient is given with the p-value for the associated t-test. For each block the resulting R with its F-value and level of significance is also
provided. P-values more than 0.10 are reported as n.s. (non-significant) and values ࣘ 0.05 are shown in bold.
657
50
400
45
English Word Comprehension

350
40 A
BPVS III raw scores

35 300
30 250
25 200
20
150
15
100
10 D
5 50
0 0
0 20 40 60 80 100 0 20 40 60 80 100
% Exposure English % English Exposure
30 400
350
English Word ProducƟon

25
B 300
SETK 2 raw scores
20 250
15 200
150
10
100
5 50 E
0
0
0 20 40 60 80 100
0 20 40 60 80 100
% English Exposure
% Exposure English
50
45
C A: BPVS III raw scores
B: SETK 2 raw scores
PLS 4 raw scores
40
C: PLS 4 raw scores
35 D: Number of understood words
from the Oxford CDI (max = 416)
30 E: Number of produced words
from the Oxford CDI (max = 416)
25
20
0 20 40 60 80 100
% English Exposure
Figure 1. Relationship between the percentage of English exposure and the English language assessments: BPVS III (n = 31), English SETK-2
(n = 30), PLS 4 Auditory (n = 21), Oxford CDI Comprehension (n = 34) and Production (n = 34).
her performance in the five English assessment measures In order to determine this, a two-stage cluster anal-
used in this study (marginally in the case of the PLS ysis was carried out, as recommended by Milligan and
4), once demographic data including age, birth order, Sokol (1980) (see also Punj and Stewart 1983). A first
gender, and parents’ occupation and education scores approximation of the data grouping was obtained via
have been accounted for. This brings us to the crucial a hierarchical cluster analysis, which was then refined
question as to whether there is a cut-off point in terms of by a k-means analysis (for a review of k-means clus-
the proportion of a child’s English input above which an ter analyses methods, see Steinley 2006). For the initial
early years’ worker can treat that child as monolingual hierarchical cluster analysis Ward’s minimum variance
for purposes of assessment for referral; and if yes, what method was used (Ward 1963), together with squared
is the percentage of exposure to English above which Euclidian distance as the similarity measure for each lan-
bilingual children achieve similar scores to monolingual guage assessment test (BPVS III, English SETK-2, Ox-
children? ford CDI comprehension and Oxford CDI production).
Table 3. Number of children within each exposure category assigned to each cluster arising from K-means cluster analysis on
children’s scores on the BPVS III, English SETK-2, CDI word comprehension and word production
Bilingual < 60% Bilingual > 60% Monolingual = 100% Distance between clusters
BPVS III Cluster 1 12 2 5 22.13
Cluster 2 3 14 29
English SETK-2 Cluster 1 9 1 0 17.3
Cluster 2 5 15 32
CDI Comprehension Cluster 1 9 0 0 266.7
Cluster 2 9 16 35
CDI Production Cluster 1 14 1 2 243.5
Cluster 2 4 15 33
Note: The last column shows the mean distance between the two clusters for each measure.
Each child’s score (z-score or raw score depending on differed on these tests. Then, the bilingual child with
the measure, see above) was entered without the infor- the lowest amount of exposure to English (5%) was
mation about his/her linguistic background (monolin- removed and the analysis was rerun. Again the monolin-
gual or bilingual) and without information about their gual and the bilingual scores were significantly different.
amount of exposure to English. This was followed by a The bilingual child with the lowest amount of exposure
K-means cluster analysis to optimize the results. to English was progressively removed until there was no
From the hierarchical cluster analyses, dendograms longer any significant difference between the bilingual
for each of the four measures show that a two-cluster so- and the monolingual groups. This stage was reached
lution divides the entire population in two groups. For when bilingual children with exposure to English above
example for the BPVS III, 19 participants were assigned 54% were compared with the monolinguals for the
to cluster 1 and 46 to cluster 2. Out of the 19 assigned BPVS III measure, above 58% for the English SETK-2,
to cluster 1, 14 were bilinguals and five monolinguals above 53% for the Oxford CDI Comprehension and
(with low scores on the BPVS III). Cluster 2 consisted above 62% for the Oxford CDI Production.10 In other
of 29 monolinguals and 17 bilinguals (with high scores words, a bilingual child with a percentage of exposure
on the BPVS III). Inspection of the amount of exposure to English at 60% (as a rounded cut-off point) or
of bilinguals (N = 31) assigned to cluster 1 or 2 shows above is very likely to score similarly to a monolingual
that most of those with less than 60% exposure to En- child on all four measures, whereas a child exposed
glish were found in cluster 1 (12 children out of 15) to English less than 60% of the time is likely to score
and most of those with more than 60% of English were less than a monolingual child on the four standardized
found in cluster 2 (14 out of 16). Then the k-means tests.
cluster analysis (see table 3) confirmed and refined this As a subsequent step, the data on table 4 are in-
result by assigning 12 bilinguals with exposure under cluded to provide an illustration of the ranked individ-
60%, two bilinguals with exposure above 60% and six ual bilingual and monolingual children on the differ-
monolinguals to cluster 1, while cluster 2 was made ent measures (percentage of exposure to English, PLS
of three bilinguals with exposure under 60%, 14 bilin- 4 Auditory, BPVS III, English SETK-2, Oxford CDI
guals with exposure above 60% and 28 monolinguals. Comprehension and Production). The purpose of this
Table 3 clearly illustrates that for each language assess- table is to highlight children whose performance may be
ment measure, the cluster solution divides children into considered outlying relative to the distribution seen in
two groups: most monolinguals and the bilinguals with monolingual children. To do this we use the mean and
more than 60% exposure to English in one group, and SD derived from the monolingual distribution. More
most bilinguals with less than 60% exposure to English specifically, we have highlighted in light grey all the val-
in the other group. ues that fall under 1 SD and in dark grey the values
Further analyses were run to corroborate this first that fall under 2 SD below the average of the monolin-
classification of the children’s scores into two broad gual children for that measure. As can be seen, outliers
categories. First, the bilingual children were ranked are found randomly across monolinguals and bilinguals
according to their amount of English exposure from with an exposure above 60%, but are much more com-
5% to 98%. Then we performed independent-samples mon for bilinguals with exposure under 60%.
t-tests with unequal variance assumed8 between the 35 An interesting additional point is that variability
bilingual children and the 36 monolingual children on was higher in the 18 bilingual children with less 60%
each of the following tests: BPVS III, English SETK- exposure to English than in the 17 children with 60%
2, Oxford CDI Comprehension and Oxford CDI or above (table 4 and figure 1), suggesting a higher
Production.9 The two groups of children systematically homogeneity of children’s performances with higher
Table 4. Individual raw scores of all children (bilingual and monolingual) assessed on the five English assessment measures, ranked as
a function of their amount of exposure to English (from 5% to 100%)
English PLS 4 English CDI CDI Other CDI

exposure (%) Auditory BPVS III SETK-2 Comprehension Production Production
1 5 9 1 133 44 56.7
2 7 28 38 14 317 214
3 14 36 413 210 25.4
4 21 3 0 0 69.7
5 27 21 65 31 45.4
6 31 18 301 297 9.6
7 33 18 9 181 80 37
8 34 4 12 63 45 8.2
9 37 25 19 416 252 32.9
10 37 38 23 264 212 21.4
11 40 4 10 58 1 5.8
12 44 35 17 7 217 65 64.7
13 48 19 5 162 161 18
14 53 16 12 414 247 44
15 54 46 32 21 401 371 73.6
16 57 32 17 415 140 3.5
17 58 24 2 242 228 72.8
18 58 20 25 289 157 27.4
19 62 37 31 20 403 293 64.2
20 67 41 28 416 403 10.6
21 69 36 30 24 381 361
22 70 35 23 27 366 342 45.8
23 74 42 44 23 368 363 9
24 75 40 31 21 386 327 1.4
25 78 46 24 400 400 1
26 78 4 6 389 112 26.7
27 82 48 36 27 381 316 34
28 83 35 32 26 382 349 22.2
29 85 45 33
30 85 38 35 26 411 394
31 89 39 39 25 393 355 23.9
32 94 39 37 26 409 344
33 94 48 45 28 411 398 44.7
34 97 47 42 27 395 395
35 98 41 35 24 382 342 2.8
36 Monolingual 40 36 28 383 370
37 Monolingual 38 36 28 399 359
38 Monolingual 34 36 25 359 305
39 Monolingual 42 32 24 403 401
40 Monolingual 47 41 24 405 384
41 Monolingual 31 35 23 416 394
42 Monolingual 46 37 29 416 384
43 Monolingual 39 33 27 380 349
44 Monolingual 43 37 26 415 399
45 Monolingual 44 22 25 407 367
46 Monolingual 44 28 26 416 414
47 Monolingual 32 52 25 408 394
48 Monolingual 38 29 28 382 351
49 Monolingual 43 43 24 410 398
50 Monolingual 48 46 28 409 399
51 Monolingual 42 26 28 409 363
52 Monolingual 45 30 27 412 412
53 Monolingual 33 22 18 403 342
54 Monolingual 41 42 29
55 Monolingual 49 31 24 402 394
56 Monolingual 51 43 24 416 369
57 Monolingual 48 55 28 416 405
58 Monolingual 50 57 402 398
Continued
Table 4. Continued
English PLS 4 English CDI CDI Other CDI

exposure (%) Auditory BPVS III SETK-2 Comprehension Production Production
59 Monolingual 41 53 408 370
60 Monolingual 37 38 408 408
61 Monolingual 24 21 395 276
62 Monolingual 39 42 26 392 369
63 Monolingual 24 295 212
64 Monolingual 41 34 28 406 404
65 Monolingual 44 46 28 416 416
66 Monolingual 40 40 24 418 414
67 Monolingual 43 52 26 407 407
68 Monolingual 17 23 345 215
69 Monolingual 43 29 413 411
70 Monolingual 24 26 397 392
71 Monolingual 43 25 407 400
Note: The score for the additional language CDI production is expressed as a percentage of the total amount of words for that particular CDI. Pale grey cells correspond to values
that are under 1 SD below the mean of the monolinguals for this measure. Dark grey cells are values under 2 SD from the monolingual values. For example, the average score of the
monolinguals for the BPVS III is 37.7, with an SD = 9.88. Therefore all outliers scores smaller than 27.8 are coloured in pale grey in this column, and those smaller than 17.9 are in
dark grey. The black row marks the limit of 60% of English exposure above which bilingual children are not distinguishable from monolingual children.
English exposure. Levene tests confirmed that variance Although a bivariate Pearson correlation analysis car-
in children’s performances with 60% exposure or above ried out on the percentage of the total number of words
was lower than in bilinguals with less exposure to En- produced in each additional language CDI of bilingual
glish for the CDI Comprehension (F(1, 32) = 33.55, p children did show a trend towards a linear negative re-
< 0.001), and the CDI Production (F(1, 32) = 6.86, lationship with the English exposure, r = –0.27, p =
p = 0.01), and marginally lower for the English SETK-2 0.15 (figure 2), this was not significant (a hierarchi-
(F(1, 28) = 3.87, p = 0.059). Thus, it would appear cal regression analysis as above shows that the amount
that beyond a critical amount of input in a particular of exposure explains an additional amount of variance
language, the vast majority of children have the ability when introduced after demographical data, but not sig-
to acquire that language in a relatively similar way. nificantly so). One reason for this might be the unfor-
Finally, it was also verified that the distribution of tunate diversity across CDI adaptations in terms of the
scores in the bilingual children with exposure to English number of words included, meaning that some CDIs
above 60% was similar to that of the monolingual chil- tested a broader range of vocabulary including less fre-
dren in terms of variance by using Levene tests to com- quent words than others, leading to variation in exposure
pare the two groups for each language measure. Levene percentages and making the cross-linguistic comparison
tests were non-significant for all measures, with F-values difficult. To illustrate, whereas the Mandarin version
below 1 for the BPVS III, the CDI Comprehension and contains a broad sample of vocabulary of 767 words,
Production and the PLS 4, and F-value slightly higher the Arabic version contains only 416, i.e. almost half
for the English SETK-2 (F(1, 47) = 2.37, p = 0.13). the words.
This confirms that the variance was similar in mono- We then focused on the 13 Arabic–English bilin-
linguals and bilinguals with exposure to English above gual participants, a population which is more highly
60% for all measures. represented in our sample, and looked at the correlation
between the amount of English exposure and their score
The Additional language CDI-tool on the Arabic CDI. Although the norms for mono-
lingual Arabic speakers do not yet exist for the Arabic
Up to this point we have only examined the English abil- CDI and although we are aware that there exists a va-
ities of the bilingual children. However, the parents of riety of Arabic dialects, at least all our Arabic–English
29 of the 35 bilingual children also completed a version participants were assessed with the same tool, so any
of the CDI in the additional language which had been trend towards one direction or the other is meaningful.
adapted, or was in the process of being adapted or, in the Once corrected for demographic data i.e. age, birth or-
case of Punjabi, translated for the purposes of this study. der, gender, parental education and occupation scores,
As the proportion of exposure to English was the main the amount of exposure explained an additional 40.7%
predictor variable for the total expressive vocabulary of of the variance (F(1, 6) = 6.62, p = 0.042) and the
the Oxford CDI, we investigated if the inverse pattern Arabic CDI scores were significantly inversely related to
would be found for the additional language CDI (see the amount of exposure to English (standardized β =
also Pearson et al. 1997). –0.87).
100.0
90.0
80.0
Percentage of produced words

70.0
60.0
50.0
CDI AddiƟonal Language
40.0
CDI English
30.0
20.0
10.0
0.0
0 20 40 60 80 100
Percentage of exposure to English
Figure 2. Relationship between the percentage of English exposure and the proportion of words that bilingual children (n = 30) produce in
each CDI (in their additional language: diamonds; in English: squares).
Discussion was the strongest (and for comprehension the only) pre-
dictor of the performance of bilingual toddlers on the
The current study investigated whether the scores of a
CDI for both lexical comprehension and production
sample of UK simultaneous bilingual toddlers on vari-
and accounted for 11–26% of the variance. Moreover,
ous production and comprehension measures could be
two-stage cluster analyses suggested a 60% exposure to
predicted by the proportion of exposure to English as
English cut-off point above which the bilingual tod-
compared with the additional language, as measured by
dlers performed like their matched monolingual control
our English exposure questionnaire. The crucial ques-
group. Thus, for CDI comprehension all monolinguals
tion was whether one could identify a cut-off point above
and all the bilinguals for whom English composed 60%
which simultaneous bilingual toddlers as a group would
or more of their exposure fell into the same cluster,
perform similarly to monolinguals, and over which at-
whereas the bilinguals for whom English comprised less
risk toddlers could be easily identified. Our long-term
than 60% of their exposure mostly fell into another
goal underlying this study was to develop an assessment
cluster.
method for bilingual 2-year-olds which could be carried
These data are further supported by the fact that
out by early years’ workers in consultation with par-
not only did the proportion of exposure to English pre-
ents. The ideal measure for this would be a standardized
dict performance of the same toddlers on the BPVS
British version of the CDI in conjunction with an objec-
III and the English SETK-2, but also that the cut-off
tive measure of the proportion of the toddler’s English
point in terms of proportion of English in the input for
input.
these other two tests was highly similar to that for the
Since the MacArthur–Bates CDI is not standardized
CDI. That said, on the PLS 4 Auditory component, the
in the UK, we firstly assessed 35 simultaneous bilingual
only test which assessed the comprehension of mixed
28–32-month-olds on the Oxford CDI and 36 mono-
processing skills of vocabulary and sentences, bilingual
linguals matched for age and socioeconomic status. We
toddlers as a group could not be distinguished signifi-
also assessed both groups on two measures standardized
cantly from their monolingual peers, although marginal
for monolinguals (BPVS III and PLS 4) as well as an
significance was found for regression analysis of the rela-
object-naming test (English SETK-2). When possible,
tionship between proportion of exposure to English and
the bilingual toddlers were also assessed on the addi-
performance on the PLS 4. However, the majority of
tional language CDI.
the bilingual toddlers did not complete the PLS 4 and
those who did predominantly heard English more than
Cut-off point of exposure to English 60% of the time.
Another caveat is that whilst the cluster analyses
We indeed found in hierarchical regression analyses that, identified two distinct groupings at an English input
after accounting for demographic variables including proportion of 60% for all four measures, the iterative
age, birth order, gender, parent’s occupation and ed- t-test identified an English input proportion cut-off
ucation scores, the proportion of exposure to English of 53–54% for the two comprehension measures but
58–64% for the two production measures. Nonetheless to complete the CDI in the additional language, where
all these thresholds are similar and indeed tie in with this CDI was available, overall there was no significant
the study of Thordardottir (2011) in which the critical relationship between exposure to English and the scores
input level was proposed to be 40–60% exposure for in the additional language. This was mainly due to the
receptive vocabulary and 60% for expressive vocabulary variability created by the use of CDIs from different lan-
for bilingual children to score similarly to monolingual guages which had a large degree of variability in the total
children. Clearly, if an alternative estimate of the propor- word scores. However, when we only look at a sample of
tion of exposure to English were used, then one would Arabic–English toddlers who were all assessed with the
probably not end up with exactly 60% as some kind same CDI tool in Arabic, a significant negative relation-
of magic number. However, the conclusion would still ship did indeed emerge. Thus one tentative suggestion
hold that bilingual toddlers who are dominant in En- emerging from our study is to promote the use—where
glish can be assessed in English only and thus need to be possible—of the CDIs which have been standardized for
referred if their English test scores are 1.5–2 SD below the additional languages. In fact, in personal commu-
the monolingual means. nication, we have heard that SLPs in various regions of
the UK are asking interpreters to translate the American
MacArthur CDIs.
Amount of exposure to English predicting early
language development: the view of language
The omission of productive syntax measure
dominance
The original intention was to pilot a measure which
Simultaneous bilingual toddlers have fewer opportuni-
early years’ workers could easily be trained to carry out
ties relative to their monolingual peers to hear multiple
and we thought this would be unlikely for measures
tokens of the same word in one given language, which
such as the analysis of language samples or carrying
would allow them to build up a robust lexical entry. It is
out the expressive component of the PLS 4. On reflec-
therefore not surprising that the amount of exposure to
tion, for a larger scale follow-up we would prefer to use
the two languages has a direct influence on their recep-
the Lincoln-UK version of the Toddler CDI (Meints
tive and expressive vocabularies. However, the amount
and Woodford 2011), although unstandardized, as it is
of exposure seems to have a catastrophic rather than a
closer to the original American MacArthur–Bates CDI
linear effect on the process of language acquisition, the
in also having measures of morphology and syntax. That
term catastrophic here being taken in its mathematical
said, almost all primary-school-aged SLI children score
sense. That is, when the Additional language is the mi-
at least 1 SD below the mean on vocabulary measures
nority language of the community and when exposure to
(Conti-Ramsden et al. 1997), no matter which ‘sub-
English is somewhat greater than that to the additional
group’ of disorder they fall under. And, further, almost
language, English seems to be optimally acquired.
all children with SLI have a history of initial language
The cut-off of 60% suggested by our data, and above
delay, which manifests itself in delayed and protracted
which bilingual children seem to acquire English very
receptive and expressive lexical acquisition. Although
similarly to their monolingual counterparts, could cor-
not all toddlers with delayed vocabulary development
respond to the value which determines their dominant
later receive a diagnosis of SLI, Dale et al. (2003) found
language. In addition to providing some quantification
that typically developing 2-year-olds who scored below
of the amount of exposure necessary for one language
the 10th percentile on the MacArthur–Bates Short Form
to become dominant, our data also suggest a strong
CDI in production were significantly more likely to fall
view of dominance, in which performance in the dom-
into the category of language-impairment in the 3–4-
inant language becomes indistinguishable from that of
year-old age range. Only a tiny proportion of the 6500
a monolingual child. That is, a child who is exposed to
or more children who scored above the 10th percentile
English above 60% of the time is able to acquire a lexical
for vocabulary production at 2 years went on to have
competence in English equal to that of a monolingual
difficulties in grammatical development at 3 and 4 years
child, alongside a lexical competence in the additional
of age.
language which is more unpredictable.
Grammatical development has been found to be
highly intercorrelated with lexical development not only
Bilingual children with less than 60% exposure to in normally developing monolingual children but also
in normally developing bilingual children, possibly as
English
a result of underlying general maturational processes.
Our current results do not provide a direct solution for That is, a Spanish–English bilingual child’s grammati-
those bilingual toddlers for whom English composes less cal development in Spanish is predicted by his or her
than 60% of their input. While we did ask the parents lexical development in Spanish (and not by his or her
lexical development in English) and vice versa (Con- least our instrument indicates a potential way forward
boy and Thal 2006, Marchman et al. 2004). Thus, we for assessing bilingual toddlers living in Britain and in
do not consider measurement of lexical comprehension the cognate English speaking countries (United States,
and production to be at all irrelevant in an initial assess- Canada, Australia, New Zealand and South Africa).
ment of 2-year-old bilinguals who may have a language,
hearing or other developmental disorder.
Clinical implications
Validity of English exposure questionnaire
Since 2013 early years’ workers in the UK are required
There are a number of critiques that one could make to check all 2-year-olds in their care. In addition, health
about our questionnaire. One might argue that detailed visitors are required to screen children using the Ages
diary data collected by the parent on when and how and Stages Questionnaire (ASQ) which consists of 30
much a particular language is spoken (e.g. Hoff et al. questions, of which six pertain to the communica-
2012, Place and Hoff 2011, but see also Paradis 2010 tion/language area (Squires and Bricker 2009). However,
with older children) might lead to a more accurate esti- in a systematic review of paediatric language screening
mate of the proportion of English in the input although in the United States, Nelson et al. (2006) noted that
there is suggestion that humans are poor at indicating the language component of the ASQ had not been in-
the frequency with which they carry out certain actions dependently validated. Moreover, they pointed out that
(e.g. Schwarz et al. 1991). there has been no study to date which has systematically
One might also quibble with the fact that the English compared the relative validity of two or more language
exposure questionnaire gives more weight to maternal screening instruments even for monolingual children.
than to paternal input; the quality of exposure to English Most importantly, the ASQ does not include sugges-
was evaluated by assigning more weight to the mother tions for bilingual children.
than to the father (two-thirds versus one-third) given The results of our study indicate that in principle
the same amount of time spent alone with their child. early identification is possible if bilingual children who
However, fathers generally produce less verbal output hear English 60% of the time or more are assessed using
than mothers to their child. They spend a greater pro- only English tools which are standardized for mono-
portion of their time interacting with their children in lingual children. These indications should have poten-
play activities, and their play is more physical than that tial positive implications since in principle early years’
of mothers, therefore directly impacting on the amount workers could screen in English approximately half of
of exposure to English and the additional language (e.g. the bilingual children, i.e. those who are above 60% on
Pancsofar and Vernon-Feagans 2006, Place and Hoff English exposure. This would also constitute a step for-
2011). ward for the issue of under-referral (Crutchley 2000) as
Furthermore, it could be argued that the English that the waiting list for SLPs would consist primarily of
exposure questionnaire did not evaluate the quality of children who have a language delay.
any exposure to English when the non-English parent On the other hand, the children with less than 60%
spoke to the child, the type of language spoken between exposure to English would also require an assessment
the parents in presence of the child, how much the par- in the additional language which, in turn, would be
ents talk to English friends or the amount of time spent their dominant language. It would be necessary to check
watching television in English versus the additional lan- whether the language scores match with the monolin-
guage. Also, the potential co-occurring intra-sentential gual scores on the additional language CDI. In a large
mixing or mixed language input within the same speaker number of cases, the CDI in the additional language
should ideally be explicitly measured to disentangle its could be available on-line as an assessment tool. Once
frequency and its effects from those of co-occurring dual a child at-risk for language delay has been referred to
language input. an SLP, she/he may find it helpful, particularly for the
Notwithstanding these potential problems, the data expressive tasks, to obtain a composite vocabulary score
from our questionnaire accounted for a significant of the CDIs to assess if these scores are equal or lower
amount of variance in all the language tests we used. The to monolingual scores as some literature on bilingual
fact that the proportion of exposure to English was a sig- children has suggested (for a discussion, cf. Bedore and
nificant predictor for all the regression analyses (albeit Peña 2008, Hemsley et al. 2010). Alternatively, more in-
marginally so for the PLS) indicates that our instrument depth follow-up assessments may be carried out such as
was at least roughly estimating the proportion of expo- dynamic assessment to determine the existence and na-
sure to English. Therefore, independent of whether our ture of any disorder. The suggestion that these children
English exposure questionnaire or a future alternative is could be assessed by exploiting the recent development
used as an instrument for early years’ workers, at the very of a relatively fast and non-costly dynamic assessment
procedure (Hasson et al. 2013) is intriguing, and ideally study showed that children’s achievement in English was
this should be adapted for younger children. not influenced by general factors such as gender, birth
order or educational and occupational scores. Rather,
the extent of English language mastery was strongly pre-
Research involving a diversity of languages in the dicted by the amount of exposure to English after the de-
bilingual group mographic variables were accounted for. Although there
is a general consensus that it is usually invalid to com-
The heterogeneity of additional languages included in pare bilingual children with monolingual norms, this
the study could be seen as a weakness in the field of devel- study clearly showed that at and above 60% of exposure
opmental bilingualism as we did not consider the differ- to English, bilingual children are comparable with their
ent degrees of linguistic proximity to English, as well as monolingual peers and can be assessed using the mono-
the differences in the social prestige of the additional lan- lingual norms so that a child with language disorder
guage (e.g. Welsh in Wales as compared with Punjabi in can be identified. For the children with less than 60%
Plymouth). However, we hold that it is in fact a strength of exposure to English, the findings are less clear given
of the current study, firstly because early years’ workers in the variability of the CDIs available in the other lan-
the UK do not usually see bilingual toddlers learning one guages. Nevertheless, the Arabic CDI data gathered from
particular additional language. Rather, they encounter the subsample of Arabic–English children showed that
a multitude of different additional languages. Secondly, the amount of exposure to English was again a strong
any variability that this heterogeneity causes should lead (but inversely related) predictor of Arabic language
to weaker relationships between the exposure to English performance.
and performance on language tests. Thus, the fact that
we found significant relationships is a point in favour of
our approach.
We argue that it is counterproductive to restrict the Acknowledgements
comparative study to one particular homogeneous ad- Many thanks to Amanda Norman, Dayal Dihman, Naomi Kilgore,
ditional language versus English language, particularly Zelah Hunt, Katherine Kraft, Natalie Fairhurst, Emma Metcalf,
if this produces an extremely large number of dual lan- Samantha Momber, Emily Cooper and Hayley Wallington for as-
sistance in testing, and to the parents who brought their children
guage comparisons resulting from the thousands of lan- into the laboratories for testing. Many thanks also to our colleagues
guages spoken in the world (Katzner 2002). Indeed, it who so kindly provided copies of their normed or pilot versions of
does seem possible to consider the child as an English the CDI, which they have developed or are in the process of devel-
learner with a particular amount of exposure. Our work oping, for other languages, and to Larry Fenson for helping track
suggests that the data collected from children with one these down: Gisela Szagun, Ciara O’Toole, Magdalena Smoczynska,
Ursula Stephany, Elisabet Serrat Sellabona (Catalan), Paula Fikkert
homogeneous additional language pattern in the same (Dutch), Sabah Safi, Nalah Dashash, Fatima Ba-Saffar (Arabic) and
way as data from children with a variety of additional Sophie Kern (French). Finally, many thanks to Anna Katzidaki,
languages. Overall, this approach offers the prospect of a Obioha Ukoumunne and Olwen Cockell for very helpful comments
new way of thinking about the early assessment of bilin- on earlier versions of this work. Preliminary versions of this work
gual 2;6-year-olds. For 2;6-year-olds who are dominant were presented at the 12th International Congress for the Study of
Child Language, and the 4th UK Cognitive Linguistics Conference,
in one of their two languages, assessment in one lan- London, UK. Declaration of interest: The authors report no con-
guage only and with simple tools seems possible. This flicts of interest. The authors alone are responsible for the content
could allow early years’ workers to make an early evalu- and writing of the paper.
ation regarding a child’s language at 30 months of age,
allowing those who need it to be referred early and elim-
inating those who are not in need of referral from the Notes
waiting lists.
1. When parents differed with respect to their educational back-
ground, the highest level attained was considered (Bello et al.
2012, Caselli et al. 2007, Doblhammer et al. 2009).
Conclusions 2. During extensive piloting and in this study we did not encounter
any cases of families in which English is the only language spoken
The current study provides unprecedented information at home with an additional language being spoken in a nursery,
regarding bilingual language development relating to for example; therefore it was not included in the calculations.
the challenges faced by SLPs in the UK in the clinical However, if during the conversation with the experimenter this
assessment of diverse children speaking a vast array of had emerged, the child would have been excluded from partici-
languages. Not surprisingly, we found that simultane- pation. It would be very easy to include an additional question
in a future version of the English exposure questionnaire. In ad-
ous bilingual children as a group underperformed their dition, for reasons of simplicity at this stage, we did not provide
English monolingual peers (matched with regards to de- calculations for families in which more than two languages are
mographic characteristics) in English proficiency. This spoken.
3. The translation of the German test was initially piloted at the CAESAR, L. and KOHLER, L., 2007, The state of school-based bilin-
universities of Manchester and Leipzig, and equivalent perfor- gual assessment: actual practice versus recommended guide-
mance was found in the raw scores for 30-month-old German lines. Language, Speech, and Hearing Services in Schools, 38,
and English children (Dittmar 2009). 190–200.
4. As for Arabic and Polish CDIs where there is an adapta- CAMILLERI, B. and LAW, J., 2007, Assessing children referred to
tion in progress but no public data available, to find and to speech and language therapy: static and dynamic assess-
contact the researcher responsible for the adaptation in other ment of receptive vocabulary. International Journal of Speech–
language visit the official website CDIs in other languages: Language Pathology, 9, 312–322.
sci.sdsu.edu/cdi/adaptations_ol.htm CASELLI, M. C. and CASADIO, P., 1995, Il Primo Vocabolario del
5. One parent was a Czech speaker but understood Slovak, which Bambino (Milan: Franco Angeli).
is a dialectal variant and was thus instructed to check off if the CASELLI, M. C., PASQUALETTI, P. and STEFANINI, S., 2007, Parole e
child said the Czech equivalent of any of the Slovak words. Frasi nel ‘Primo Vocabolario del Bambino’. Nuovi dati normativi
6. The Punjabi translation of the Oxford CDI was only used for fra i 18 e i 36 mesi e forma breve del questionario [Words
the analysis in the additional language. and Sentences in ‘The First Vocabulary of the Child’: New
7. Similar regression analyses carried out using the raw scores for the Normative Data from 18 and 36 Months and Short Form of
BPVS III, PLS 4 and English SETK-2 reported equal outcomes. the Questionnaire] (Milan: Franco Angeli).
8. The variance differed until the cut-off point was reached, i.e. CONBOY, B. T. and THAL, D. J., 2006, Ties between the lexicon and
between the group of toddlers with less than 60% exposure to grammar: cross-sectional and longitudinal studies of bilingual
English, on the one hand, and, on the other hand, the mono- toddlers, Child Development, 77, 712–735.
lingual children plus the bilingual toddlers with 60% or above CONTI-RAMSDEN, G., CRUTCHLEY, A. and BOTTING, N., 1997, The
exposure to English. Extent to which psychometric tests differentiate subgroups of
9. This particular analysis for the PLS 4 scores was not carried out, children with SLI. Journal of Speech, Language, and Hearing
as 15 of 36 bilingual toddlers did not complete this test. Research, 40, 765–777.
10. To make sure that this effect was not due to chance, we also CRUTCHLEY, A., 2000, Bilingual children in language units: does
carried out the procedure, removing children’s data one by one, having ‘well-informed parents’ make a difference? Interna-
until there were only 10 bilingual toddlers left. No significant tional Journal of Language and Communication Disorders, 35,
t-test value was observed. 65–81.
DALE, P. S., PRICE, T. S., BISHOP, D. V. M. and PLOMIN, R.,
2003, Outcomes of early language delay: I. Predicting per-
References sistent and transient language difficulties at 3 and 4 years.
Journal of Speech, Language, and Hearing Research, 46, 544–
AMERICAN SPEECH-LANGUAGE-HEARING ASSOCIATION (ASHA), 560.
1999, Guidelines for the Roles and Responsibilities of School- DE HOUWER, A., 2009, Bilingual First Language Acquisition (Bristol:
Based Speech–Language Pathologists (Rockville, MD: Ameri- Multilingual Matters).
can Speech–Language–Hearing Association). DITTMAR, M., 2009, Acquiring the transitive construction: the de-
BAKER, C., 2011, Bilingualism and cognition. In C. Baker (ed.), velopment of understanding word order and case marking
Foundations of Bilingual Education and Bilingualism, 5th edn cues. Unpublished doctoral dissertation zur Erlangung des
(Bristol: Multilingual Matters), pp. 138–162. akademischen Grades Doktorin der Philosophie. Fachbereich
BEDORE, L. M. and PEÑA, E. D., 2008, Assessment of bilingual Erziehungswissenschaft und Psychologie der Freien Univer-
children for identification of language impairment: current sität Berlin, pp. 77–78.
findings and implications for practice. International Journal DOBLHAMMER, D., HOFFMANN, R., MUTH, R., WESTPHAL, C. and
of Bilingual Education and Bilingualism, 11, 1–29. KRUSE, A., 2009, A systematic literature review of studies
BEDORE, L. M., PEÑA, E. D., GARCÍA, M. and CORTEZ, C., 2005, analysing the effects of sex, age, education, marital status,
Conceptual versus monolingual scoring: when it does make a obesity, and smoking on health transition, Demographic Re-
difference? Language, Speech, and Hearing Services in Schools, search, 20, 37–64.
36, 188–200. DUNN, L. M. and DUNN, L. M., 1997, Examiner’s Manual for the
BELLO, A., GIANNANTONI, P., PETTENATI, P., STEFANINI, S. and PPVT-III: Peabody Picture Vocabulary Test Third Edition (Cir-
CASELLI, M. C., 2012, Assessing lexicon: validation and de- cle Pines, MN: American Guidance Service).
velopmental data of the picture naming game (PiNG), a new DUNN, L. M., DUNN, D. M. and NATIONAL FOUNDATION FOR ED-
picture naming task for toddlers. International Journal of Lan- UCATIONAL RESEARCH, 2009, The British Picture Vocabulary
guage and Communication Disorders, 47, 589–602. Scale Third Edition BPVS III (London, GL Assessment).
BIALYSTOK, E., 2007, Cognitive effects of bilingualism: how linguis- ELLIS WEISMER, S., 2000, Language intervention for young chil-
tic experience leads to cognitive change. International Journal dren with language impairments. In L. Watson, E. Crais and
of Bilingual Education and Bilingualism, 10, 210–223. T. Layton (eds), Handbook of Early Language Impairment in
BIALYSTOK, E., 2009, Bilingualism: the good, the bad, and the indif- Children: Assessment and Treatment (Albany, NY: Delmar),
ferent. Bilingualism: Language and Cognition, 12, 3–11. pp. 173–198.
BISHOP, D. V., 2006, What causes specific language impairment FENSON, L., DALE, P. S., REZNICK, J. S., THAL, D., BATES, E., HAR-
in children? Current Directions in Psychological Sciences, 15, TUNG, J. P., PETHICK, P. S. and REILLY, J. S., 2007, The
217–221. MacArthur Communicative Development Inventories: User’s
BORNSTEIN, M. H., HAHN, C. S., SUWALSKY, J. T. D. and HAYNES, O. Guide and Technical Manual, Second Edition (San Diego, CA:
M., 2003, Socioeconomic status, parenting and child devel- Singular).
opment: the Hollingshead Four-Factor Index of Social Status GATHERCOLE, V. C. M., 2002, Command of the mass/count distinc-
and the Socioeconomic Index of Occupations. In M. H. Born- tion in bilingual and monolingual children: an English mor-
stein and R. H. Bradley (eds), Socioeconomic Status, Parenting phosyntactic distinction. In D. K. Oller and R. E. Eilers (eds),
and Child Development (Mahwah, NJ: Lawrence Erlbaum As- Language and Literacy in Bilingual Children (Tonawanda, NJ:
sociates), pp. 29–82. Multilingual Matters), pp. 175–206.
GATHERCOLE, V. C. M., 2007, Miami and North Wales, so MILLIGAN, G. W. and SOKOL, L. M., 1980, A two-stage clustering
far and yet so near: Constructivist account of morpho- algorithm with robust recovery characteristics, Educational
syntactic development in bilingual children, International and Psychological Measurement, 40, 755–759.
Journal of Bilingual Education and Bilingualism, 10(3), NELSON, H. D., NYGREN, P., WALKER, M. and PANOSCHA, R., 2006,
224–247. Screening for speech and language in preschool children: Sys-
GATHERCOLE, V. C. M. and THOMAS, E. M., 2009, Bilingual first- tematic evidence review for the US preventive services task
language development: dominant language takeover, threat- force. Pediatrics, 117(2), e298–e319.
ened minority language take-up. Bilingualism: Language and NICHOLLS, R. J., EADIE, P. A. and REILLY, S., 2011, Monolingual
Cognition, 12, 213–237. versus multilingual acquisition of English morphology: what
GRIMM, H., 2000, SETK-2: Sprachentwicklungstest fu¨r zweijährige can we expect at age 3? International Journal of Language and
Kinder (2.0–2.11). Diagnose rezeptiver und produktiver Communication Disorders, 46, 449–463.
Sprachverarbeitungsfähigkeiten (Gottingen: Hogrefe). OFFICE FOR NATIONAL STATISTICS (ONS), 2011, Parents’ Country
GUTIÉRREZ-CLELLAN, V. and PEÑA, E., 2001, Dynamic assessment of Birth, England and Wales, 2010. ONS Statistical Bul-
of diverse children: a tutorial. Language, Speech, and Hearing letin, 25 August (available at: http//www.ons.gov.uk/ons/
Services in Schools, 32, 212–224. dcp171778_230307.pdf).
HAMILTON, A., PLUNKETT, K. and SCHAFER, G., 2000, Infant vo- O’TOOLE, C. and FLETCHER, P., 2010, Validity of a parent report
cabulary development assessed with a British communicative for Irish-speaking toddlers. First Language, 30, 199–217.
development inventory, Journal of Child Language, 27, 689– PANCSOFAR, N. and VERNON-FEAGANS, L., 2006, Mother and fa-
705. ther language input to young children: contributions to later
HASSON, N., CAMILLERI, B., JONES, C., SMITH, J. and DODD, B., language development. Journal of Applied Developmental Psy-
2013, Discriminating disorder from difference using dynamic chology, 27, 571–587.
assessment with bilingual children. Child Language Teaching PARADIS, J., 2010, Bilingual children’s acquisition of English verb
and Therapy, 29, 57–75. morphology: effects of language exposure, structure complex-
HASSON, N. and JOFFE, V., 2007, The case for dynamic assessment ity, and task type. Language Learning, 60, 651–680.
in speech and language therapy, Child Language Teaching and PARADIS, J., EMMERZAEL, K. and DUNCAN, T. S., 2010, Assessment
Therapy, 23, 9–25. of English language learners: using parent report on first lan-
HEMSLEY, G., HOLM, A. and DODD, B., 2010, Patterns in diversity: guage development. Journal of Communication Disorders, 43,
lexical learning in Samoan–English bilingual children. Inter- 474–497.
national Journal of Speech–Language Pathology, 12, 362–374. PARRA, M., HOFF, E. and CORE, C., 2011, Relations among language
HOFF, E., CORE, C., PLACE, S., RUMICHE, R. SENOR, M. and PARRA, exposure, phonological memory, and language development
M., 2012, Dual language exposure and early bilingual devel- in Spanish–English bilingually developing 2-year-olds. Jour-
opment, Journal of Child Language, 39, 1–27. nal of Experimental Child Psychology, 108, 113–125.
JUNKER, D. A. and STOCKMAN, I. J., 2002, Expressive vocabulary PAUL, R., 1996, Clinical implications of the natural history of slow
of German–English bilingual toddlers, American Journal of expressive language development. American Journal of Speech–
Speech–Language Pathology, 11, 381–395. Language Pathology, 5, 5–21.
KAPALKOVÁ, S., SLANČOVÁ, D., BÓNOVÁ, I., KESSELOVÁ, J. and PEARSON, B. Z., FERNANDEZ, S. C., LEWEDEG, V. and OLLER, D.
MIKULAJOVÁ, M., 2010, Hodnotenie komunikačných schopnostı́ K., 1997, The relation of input factors to lexical learning by
detı́ v ranom veku (Bratislava: Slovenská asociácia logopédov). bilingual infants. Applied Psycholinguistics, 18, 41–58.
KATZNER, K., 2002, The Languages of the World (London: Rout- PEARSON, B. Z., FERNANDEZ, S. C. and OLLER, D. K., 1993, Lexical
ledge). development in bilingual infants and toddlers: comparison to
KERN, S., 2007, Lexicon development in French-speaking infants, monolingual norms. Language Learning, 43, 93–120.
First Language, 27, 227–250. PEÑA, E. D. and HALLE, T. G., 2011, Assessing preschool dual
KOVACS, A. M. and MEHLER, J., 2009, Cognitive gains in 7-month- language learners: traveling a multiforked road. Child Devel-
old bilingual infants. Proceedings of the National Academy of opment Perspectives, 5, 28–32.
Sciences, USA, 106, 6556–6560. PLACE, S. and HOFF, E., 2011, Properties of dual language expo-
KRUGER, J. and DUNNING, D., 1999, Unskilled and unaware of it: sure that influence two-year-olds’ bilingual proficiency. Child
how difficulties in recognizing one’s own incompetence lead Development, 82, 1834–1849.
to inflated self-assessments. Journal of Personality and Social POULIN-DUBOIS, D., BLAYE, A., COUTYA, J. and BIALYSTOK, E.,
Psychology, 77, 1121–1134. 2011, The effects of bilingualism on toddlers’ executive func-
LÓPEZ ORNAT, S., GALLEGO, C., GALLO, P., KAROUSOU, A., tioning. Journal of Experimental Child Psychology, 108, 567–
MARISCAL, S. and MARTÍNEZ, M., 2005, MacArthur: In- 579.
ventario de desarrollo comunicativo. Manual y Cuadernillos PUNJ, G. and STEWART, D. W., 1983, Cluster analysis in marketing
(Madrid: TEA). research: review and suggestions for application. Journal of
LYYTINEN, P., 1999, Varhaisen kommunikaation ja kielen kehityksen Marketing Research, 20(2), 134–148.
arviointimenetelmä. Jyväskylän yliopiston Lapsitutkimuskeskus RESTREPO, M. A. and SILVERMAN, S., 2001, Validity of the Spanish
ja Niilo Mäki instituutti. Jyväskylä: Niilo Mäki Säätio (Uni- Preschool Language Scale-3 for use with bilingual children.
versity of Jyväskylä, Child Research Center and Niilo Mäki American Journal of Speech–Language Pathology, 10, 382–393.
Institute, Jyväskylä: Yliopistopaino). ROYAL COLLEGE OF SPEECH AND LANGUAGE THERAPISTS (RCSLT),
MARCHMAN, V. A., MARTÍNEZ-SUSSMANN, C. and DALE, P., 2004, 2007, Good Practice for Speech and Language Therapists
The language-specific nature of grammatical development: Working with Clients from Linguistic Minority Communities.
evidence from bilingual language learners. Developmental Sci- Guidelines Prepared by the RCSLT Specific Interest Group in
ence, 7, 212–224. Bilingualism (London: RCSLT) (available at: http://www.
MEINTS, K. and WOODFORD, A., 2011, Electronic Vocabulary rcslt.org/members/publications/linguistic_minorities.pdf)
Database Lincoln UK-CDI Infants and/or the Lincoln UK- (accessed on 30 April 2013).
CDI Toddlers. Computer Software and Manual (available at: SCHEELE, A. F., LESEMAN, P. P. M. and MAYO, A. Y., 2010, The home
http://www.lincoln.ac.uk/psychology/babylab.htm). language environment of monolingual and bilingual children
and their language proficiency. Applied Psycholinguistics, 31, THORDARDOTTIR, E., 2011, The relationship between bilingual ex-
117–140. posure and vocabulary development. International Journal of
SCHWARZ, N., BLESS, H., BOHNER, G., HARLACHER, U. and KELLEN- Bilingualism, 15(4), 426–445.
BENZ, M., 1991, Response scales as frames of reference: the THORDARDOTTIR, E., ROTHENBERG, A., RIVARD, M.-E. and NAVES,
impact of frequency range on diagnostic judgements. Applied R., 2006, Bilingual assessment: can overall proficiency be esti-
Cognitive Psychology, 5, 37–49. mated from separate measurement of two languages? Journal
SERRAT, E., SANZ-TORRENT, M., BADIA, I., AGUILAR, E., OLMO, R., of Multilingual Communication Disorders, 4, 1–21.
LARA, M. F., ANDREU, L. Y SERRA, M., 2010, La relación WARD, J., 1963, Hierarchical grouping to optimize an objective
entre el aprendizaje léxico y el desarrollo gramatical. Infancia function. Journal of the American Statistical Association, 58,
y Aprendizaje, 33(4), 435–448. 236–244.
SQUIRES, J. and BRICKER, D., 2009, Ages & Stages Questionnaires, WIIG, E. H., SECORD, W. and SEMEL, E. M., 1992, Clinical Evalua-
Third Edition (ASQ-3) (Baltimore, MD: Brookes). tion of Language Fundamentals—Preschool (San Antonio, TX:
STEINLEY, D., 2006, K-means clustering: a half-century synthesis. Psychological Corporation).
British Journal of Mathematical and Statistical Psychology, 59, WILLIAMS, C. J. and MCLEOD, S., 2012, Speech–language patholo-
1–34. gists’ assessment and intervention practices with multilingual
STOW, C. and DODD, B., 2003, Providing an equitable service to children. International Journal of Speech–Language Pathology,
bilingual children in the UK: a review. International Journal 14, 292–305.
of Communication Disorders, 38, 351–377. ZIMMERMAN, I. L., STEINER, V. G. and EVATT POND, R. E.,
SZAGUN, G., STUMPER, B. and SCHRAMM, A. S., 2009, Fragebogen 2009, Preschool Language Scale Fourth Edition—UK (London:
zur frühkindlichen Sprachentwicklung (FRAKIS) und FRAKIS- Pearson Assessment).
K (Kurzform) (Frankfurt: Pearson Assessment). ZINK, I. and LEJAEGERE, M., 2002, N-CDIs: Lijsten voor Com-
TARDIF, T., FLETCHER, P., ZHANG, Z. X., LIANG, W. L. and ZUO, municatieve Ontwikkeling. Aanpassing en hernormering van
Q. H., 2008, The Chinese Communicative Development Inven- de MacArthur CDIs van Fenson et al. Acco [A CDI User’s
tory (Putonghua and Cantonese versions): Manual, Forms, and Manual with Normative and Validity Data] (Leuven and
Norms (Beijing: Peking University Medical Press). Leusden).
Appendix A: English exposure questionnaire:

calculating the percentage of English and
non-English input
Evaluation of the amount of exposure to English and to an additional language

INSTRUCTIONS
Each parent will take a different route through this part of the questionnaire.
Can you write the answers in column C next to the answer which is correct for this particular child
Section A: Language(s) spoken in the home
Do you and your partner . . . .? (Can you circle your situation and go to the section indicated)
a) This child hears 1 language, English. Go to Section D
b) This child hears 2 languages, because both parents speak to her using another language Go to Section B
(for example, they both speak Russian).
c) This child hears 2 languages, because one of the parents speaks to her using another Go to Section C
language (for example, Mum speaks Spanish and Dad speaks English).
d) This child hears 3 languages, because each parent speaks a different language to the child (for Go to Section D
example, Mum speaks Spanish and Dad Russian).
e) This child hears 3 languages, because Mum and Dad speak another language to the child, but Go to Section D
also because another person (a grandparent or a childminder for example) speaks a third
language (for instance, Mum and Dad speak Spanish and the child has a French nanny).
Section B: Both parents speak the same additional language to the child (let’s refer to it as
additional language)
1 Can you please write here what is the additional language (e.g. Spanish)
2 Write the number of hours a week in average your child spends in an English speaking
nursery/day care/preschool/childminder/relative or friend.
3 Write the number of hours in average your child spends sleeping per 24 hours
4 Does the mother of this child . . . (please write 1 in the corresponding cell)
Always speak additional language to your child
Usually speak additional language to your child
Speak English to your child about half the time
Usually speak English to your child
Always speak English to your child
Continued
5 Does the father of this child . . . (please write 1 in the corresponding cell)
Always speaks additional language to your child
Usually speaks additional language to your child
Speaks English to your child about half the time
Usually speaks English to your child
Always speaks English to your child
6 When you and your partner are together with this child, who speaks most to the child?
(please write 1 in the appropriate cell)
Mother
Father
We both speak to this child an equal amount
7 If there are certain days or parts of certain days in a typical week when only you or your
partner are with your child (e.g. father always takes care of child on Saturday
afternoons).
Write the number of hours per week when your child is with her mother only.
Write the number of hours per week when your child is with her father only
Percentage of exposure to English
Please go to Section D
Section C: One parent speaks English, the other parent speaks an additional language to the
child (let’s refer to it as additional language)
1 Can you please write here what is the additional language (e.g. Spanish) French
2 Who speaks English? Please write 1 if it is the mother and 2 if it is the father. 2
3 Write the number of hours a week in average your child spends in an English speaking 21
nursery/day care/preschool/childminder/relative or friend.
4 Write the number of hours in average your child spends sleeping per 24 hours 11
5 The English speaking parent . . . .(please write 1 in the corresponding cell)
Always speaks English to your child 1
Speaks additional language to your child about half the time
Always speaks additional language to your child
6 The additional language speaking parent . . . (please write 1 in the corresponding cell)
Always speaks additional language to your child 1
Speaks English to your child about half the time
Always speaks English to your child
7 When you and your partner are together with this child, who speaks most to the child?
(please write 1 in the appropriate cell)
The English speaking parent
The additional language speaking parent 1
We both speak to this child an equal amount
8 If there are certain days or parts of certain days in a typical week when only you or your
partner are with your child (e.g. father always takes care of child on Saturday
afternoons).
Write the number of hours per week when your child is with the English speaking parent 10
only.
Write the number of hours per week when your child is with the additional language 15
speaking parent only.
Percentage of exposure to English 0.35
Please go to Section D
Section D All parents, please fill in this section
1 What is the mother’s highest educational qualification? Please write 1 after the
corresponding case.
No qualifications
Below standard for a pass on the school-leaving examination
O-levels
A-levels
Tertiary vocational qualifications
An undergraduate degree
A postgraduate degree 1
2 What is the father’s highest educational qualification? Please write 1 after the
corresponding case.
No qualifications
Below standard for a pass on the school-leaving examination
O-levels
A-levels
Tertiary vocational qualifications
An undergraduate degree 1
A postgraduate degree
3 What is the mother’s occupation? Teacher
4 What is the father’s occupation? Estate Agent
5 Does your child have older siblings? Please write the ages of the older siblings: Age
Sibling 1 4
Sibling 2
Sibling 3
Sibling 4
6 Please enter your child’s date of birth: 01/01/2008
Please enter today’s date: 30/06/2010
7 Please enter your child’s gender (1 = girl, 2 = boy): 1
8 Does your child have any identified hearing problem? (1 if yes, and please write more
below)
9 Was your child more than 6 weeks premature? (1 if yes)
10 Does your child have any identified developmental delay? (1 if yes, and please write more
below)
11 Where was your child born? UK
12 How long have you been living in an English-speaking country for? 4 years
Identification code (internal use)
Appendix B: Details of the calculation of the If M (or F) = Always AL then ME (or FE) = 100
percentage of English exposure in a typical If M (or F) = Usually AL then ME (or FE) = 75
week for a toddler in its last year of life If M (or F) = Equally AL and English then ME (or FE)
= 50
If M (or F) = usually English then ME (or FE) = 25
Input from the parents If M (or F) = always English then ME (or FE) = 0
Number of hours a week in an English-speaking nurs-
ery/childminder/playgroup = N (2) Correct HM and HF to give more weight to the
Number of sleeping hours per night = S time spent with the Mother, as it is found usu-
Does the mother always speak the additional language ally that fathers tend to produce less verbal out-
(AL) to the child, or usually, or equally often English put to their child, therefore directly impacting on
and the AL, or usually English, or always English (five the amount of exposure in English and the addi-
possible responses) = M tional language (e.g. Pancsofar and Vernon-Feagans
Does the father always speak the additional language to 2006):
the child, or usually, or equally often English and the
AL, or usually English, or always English (five possible Corrected time with Mother = CHM = HM∗ 4/3
responses) = F Corrected time with Father = CHF = HF∗ 2/3
When together, who speaks most to the child? Mother,
father or both = Most
Number of hours per week spent with mother only = (3) Assign a value (MI to Most), to give more weight
HM to the Mother’s input. What is obtained corre-
Number of hours per week spent with father only = HF sponds to the percentage of the Mother’s input
during the time when both parents are with the
child:
What does the calculation entail?
(1) Assign a percentage to M and F to estimate the If Most = Mother then MI = 90
proportion of English in each parent’s input to the If Most = Father then MI = 50
child: If Most = Both then MI = 70
(4) Calculate the number of hours per week with both (6) Calculate the percentage of exposure to English:
parents together:
E
TBP = 7(24 − S) − N − CHM − CHF P =
7(24 − S)
(5) Calculate the total number of hours of En-
glish exposure in a week (E) with the following Appendix C: Adapted English SETK-2
formula:
Objects
1 Key
E = English from mother when mother alone 2 Doll, dolly, baby, child
3 Knife
+ English from father when father alone 4 Ball, football
5 Pencil, pen, felt tip, crayon, colourer
+ English from mother when both parents together 6 Book, picture book
+ English from father when both parents together Pictures
7 Car, types of car, e.g. VW
+ English from nursery or equivalent 8 Chair, seat
9 House, hut, villa, home, flat
10 Clock, alarm clock, tick tock
CHM(100 − ME) CHF(100 − FE) 11 Swing
E= + 12 Tree
100 100 13 Apple
MI(100 − ME) 14 Fork
+ N + 0.01∗ TBP∗ 15 Scissors, snip snip
100 16 Eyes, eye
17 Duck, goose, quack quack
TBP(100 − MI)(100 − F E)
+ 0.01∗ 18 Cup, beaker
100 19 Pig, oink oink, piglet, sow
20 Bus
with: 21 Butterfly
English from mother when mother alone = CHM(100 22 Pear
23 Comb
– ME)/100 24 Star
English from father when father alone = CHF(100 25 Cake, muffin, bun
– FE)/100 26 Bear, teddy, teddy bear, polar bear
English from mother when both parents together = 27 Train,
0.01∗ TBP∗ MI(100 – ME)/100 28 Brush
29 Fridge
English from father when both parents together = 30 Petrol station, garage
0.01∗ TBP(100 – MI)(100 – FE)/100

How Much Exposure To English Is Necessary For A Bilingual Toddler To Perform

Uploaded by

Copyright:

Available Formats

How Much Exposure To English Is Necessary For A Bilingual Toddler To Perform

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

How Much Exposure To English Is Necessary For A Bilingual Toddler To Perform

Uploaded by

Copyright:

Available Formats

INT J LANG COMMUN DISORD, NOVEMBER–DECEMBER 2014,

VOL. 49, NO. 6, 649–671

Keywords: bilingualism, lexicon, language delay and assessment.

What this paper adds?

What this paper adds?

Introduction (health visitors and early years’ practitioners) are not

Oxford Communicative Development Inventory (Oxford

Procedure Comparison between monolingual and bilingual

BPVS III PLS 4 SETK-2 CDI Comp CDI Prod

Coefficient Coefficient Coefficient Coefficient Coefficient

English Word Comprehension

BPVS III raw scores

English Word ProducƟon

English PLS 4 English CDI CDI Other CDI

English PLS 4 English CDI CDI Other CDI

Percentage of produced words

Appendix A: English exposure questionnaire:

Evaluation of the amount of exposure to English and to an additional language

You might also like