Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228658354

Effects of input on the early grammatical development of bilingual children

Article  in  International Journal of Bilingualism · December 2010


DOI: 10.1177/1367006910370917

CITATIONS READS

83 311

1 author:

Elma Blom
Utrecht University
86 PUBLICATIONS   1,616 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Linguistic Aspects of Texting and its Impact on Cognitive and Social Development View project

Children and Language Mixing: developmental, psycholinguistic and sociolinguistic aspects (CALM) View project

All content following this page was uploaded by Elma Blom on 03 August 2015.

The user has requested enhancement of the downloaded file.


422 ‘International
INTERNATIONAL Journal of Bilingualism’
JOURNAL • Volume 14• 14
OF BILINGUALISM Number
(4) 4 • 2010, 422–446

Effects of input on the early grammatical


development of bilingual children
Elma Blom
University of Alberta, Canada and University of Amsterdam, The Netherlands

Abstract Key words


This study investigates the language development of 2- to 3-year-old child
Turkish–Dutch bilingual children with different amounts of input quan- bilingualism
tity. Developmental patterns in spontaneous speech data of the bilingual
children are compared to those of monolingual children of the same age. Dutch
It is found that low input quantity leads to slower grammatical develop-
ment, but only if input is clearly reduced. The observation that not only
mean length of utterance but also the development of finiteness can show
finiteness
pronounced delays in bilingual language acquisition contradicts matu-
rational views of grammatical development. However, such overall delays grammatical
are expected given input-based theories of grammatical acquisition. All development
four bilingual children show difficulties in establishing the relation between
finiteness and expression of grammatical subjects in Dutch. It is argued input
that cross-linguistic influence, driven by surface overlap between Turkish
and Dutch, may account for this observation. Turkish

1 Introduction
Whereas for the development of the lexicon it is commonly accepted that input is relevant
(Hoff & Naigles, 2002), this is less clear for the development of grammar. According to
one viewpoint, maturation of innate knowledge of language, and not input, determines
grammatical development (e.g. Crain, 1991; Hornstein & Lightfoot, 1981; Wexler, 1991).
Others argue for a much more prominent role of the input (Newport & Aslin 2004,
Thompson & Newport, 2007; Tomasello, 2003). The question of whether input influences
grammatical development is particularly relevant for bilingual children, because these
children show extensive variation in how much they are exposed to the languages they are
learning. Moreover, they are likely to have less exposure to both of their languages than
their monolingual peers (Paradis & Genesee, 1996: 20). Therefore, this study focuses on
the early language development of four Turkish–Dutch bilingual children with different
input situations. It is found that input quantity influences children’s early grammatical
development, resulting in differences between bilingual and monolingual children.
However, such differences were only found for the language in which a bilingual child

Address for correspondence


Dr Elma Blom, Linguistics Department, 4–32 Assiniboia Hall, University of Alberta, Edmonton, Alberta,
Canada T6G 2E7. [email: w.b.t.blom@uva.nl or eblom@ualberta.ca]

The International Journal of Bilingualism


Copyright © 2010 the Author/s 2010, Vol 14 (4): 422–446; ID no 370917;
DOI; 10.1177/1367006910370917 http://Ijb.sagepub.com
Blom: Input, bilingualism and grammatical development 423

received clearly reduced input (that is, the weaker language). There was one aspect of
grammatical development that categorically distinguished bilingual and monolingual
children, namely the use of subjects. This difference may be caused by cross-linguistic
influence between Turkish and Dutch.

2 Bilingual language development and input quantity


A simple prediction of the hypothesis that amount of input influences grammatical
development is that bilingual children show a protracted development, because they
receive less input in one language than monolingual children of the same age. Overviews
in the literature seem to contradict this hypothesis (Genesee, 2001; Meisel, 2001; Nicoladis
& Genesee, 1997; Paradis & Genesee, 1996). In this section, we will discuss a number
of studies showing that amount of input does influence the grammatical development
of bilingual children.
Nearly all studies on the early grammatical development of bilingual children
have focused on the development of the dominant language, and little is known of how
the weaker language of bilingual children develops (Meisel, 2007: 496; Paradis, Crago,
Genesee, & Rice, 2003: 3). The few studies that investigated the weaker language – which
is usually the language to which a bilingual child is less frequently exposed – report
differences with monolingual development. In a study of word order properties in the
speech production of 2- and 3-year-old Swedish–French bilingual children, Schlyter
(1993) and Schlyter & Håkansson (1994) found that the weaker language of young
bilingual children is characterized by a lower mean length of utterance, suggesting a
slower development. They also report the occurrence of atypical word order errors in the
weaker language. Recently, Meisel (2007) confirmed the lower mean length of utterance
for the weaker language of German–French bilinguals, but he did not find non-native
error profiles in the weaker language.
Paradis, Nicoladis, and Crago (2007) compared the development of the dominant/
weaker language, but their study included older children who ranged in age from 4,0–5,5
(mean age 4,9). The French–English bilingual children in this study were divided into
an English-dominant (n = 14) and French-dominant (n = 11) group, depending on the
language in which they received most input. Amount of input was measured through
length of exposure since birth and exposure in the home as well as at the daycare/
preschool/kindergarten. All children were tested on their use of regular and irregular
past tense in English and French. For both languages, it was found that children’s perfor-
mance on regulars was better than on irregulars. In English, English-dominant children
performed equal to monolinguals for regular verbs but for irregulars, their performance
was lower. The French-dominant children scored lower than monolingual English chil-
dren on both English regular and irregular verbs. In French, French-dominant children
outperformed English-dominant children. The observation that English-dominant
children scored lower than the monolinguals on English irregulars only, whereas the
French-dominant children scored lower on both English regulars and irregulars indicates
that dominance, hence input quantity, is a relevant factor. This conclusion is confirmed
by a comparison of the bilingual groups on the French past tense. The results of the
English past tense also show that some structures are affected by input quantity more
easily than other structures.
The International Journal of Bilingualism
424 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

Gathercole (2002a, 2002b, 2002c) measured amount of input by instructional


method at school, the language spoken at home and socio-economic status. Using a
grammaticality judgment task with correction, she tested the mass/count distinction
(English), grammatical gender (Spanish) and that-trace structures (Spanish/English) in
Spanish–English 2nd- and 5th-grade children. For all three structures, it was found that
monolingual children outperformed the bilinguals. In addition, differences between
bilingual groups confirmed the relevance of the amount of input. For instance, with
respect to the mass/count distinction in English, it was found that in grade 2, bilingual
participants of the immersion programme outperformed bilingual participants of the
two-way method. With respect to grammatical gender in Spanish, the 2nd-grade Spanish-
dominant group performed better than all other bilingual groups at that age in the
judgment task. In the 5th grade, the bilingual group with least Spanish input made fewer
grammatical corrections than children of the same age in the other groups.
The studies of Paradis et al. (2007) and Gathercole (2002a, 2002b, 2002c) support
the hypothesis that input quantity influences the grammatical development of older
bilingual children. However, the participants in these studies may have an age at which
their (innate) ability to acquire grammar is already fading out (Meisel, 2008: 58–59). As
a consequence, the bilingual children in these studies might rely more on distributional
properties of the input than younger bilingual children do. The participants’ ages may
also have caused a certain selectiveness of input effects: at time of testing the children
may already have caught up with respect to the more transparent target properties (e.g.
regular past tense). For these reasons, it is important to look more closely at effects of
input quantity in very young bilingual children. Therefore, we investigated the gram-
matical development of 2- to 3-year-old bilingual children. Specific attention will be
given to the development of finiteness, which is a transparent target property that has
been approached from a nativist perspective as well as from an input-driven perspective.
In the following section, we will further expand on both views.

3 Maturational and input-driven theories on the


development of finiteness
Children learning a non pro-drop language tend to omit finite verbs and initially produce
nonfinite utterances even though this is marked in the ambient language (Wexler, 1998).
In order to explain this ‘Optional Infinitive stage’, it has been proposed that particular
knowledge of the grammar is subject to maturation (Radford 1990, Rizzi 1993/1994,
Wexler, 1998, 2003). For instance, Wexler (2003) argued that children employ a single
computational operation where the target grammar requires two operations because
children obey the ‘Unique Checking Constraint’. Amongst other things, Wexler argues
that this allows children to use both finite and nonfinite utterances. As children grow
older they lose the Unique Checking Constraint, and hence the typical characteristics of
the Optional Infinitive stage. According to Rizzi (1993/1994), children lack knowledge
of the Root Principle, providing them with the possibility to project partial structures.
If the Root Principle matures, which, according to Rizzi (1993/1994: 384, fn. 4) happens
between ages 2 and 3, the use of nonfinite sentences will quite rapidly diminish.
The foregoing theories explain similarities between child language and target
language from the knowledge that children have from birth (e.g. a full-fledged syntactic
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 425

structure, relations between words). Differences between child language and the target
language are explained as effects of biological maturation, which may either cause
knowledge to fall away (e.g. the Unique Checking Constraint) or come in (e.g. the Root
Principle) at a particular age.
According to another point of view, children’s early clauses reflect input distri-
butions. Children start out with word-specific constructions, which they generalize
to more general frames as a function of more input (Tomasello 2003). With respect to
children’s early nonfinite clauses, it has been proposed that this can be analyzed as the
combined effect of the frequency of nonfinite verb forms together with their final posi-
tion in the input in languages such as Dutch and German. The assumption underlying
this view is that for young children, the final sentence-position is relatively salient. In
a pro-drop language like Spanish, nonfinite verbs are not in final position, causing the
absence of an early stage with nonfinite clauses in this language (Freudenthal, Pine,
Aguado-Orea, & Gobet, 2007; Freudenthal, Pine, & Gobet 2006). In English, it has been
argued that the high number of questions and sentences with do-support in the input
exposes children to relatively many sentence-final nonfinite verbs (Theakston, Lieven,
& Tomasello, 2003).
The two types of theories appear to be indistinguishable in various respects.
Both theories predict optional infinitives in non pro-drop languages. In addition, both
theories predict a contingency between verb form and place, that is, both morphologically
finite and nonfinite verbs are in their target position. Maturational accounts predict
this because of the premise that children have, apart from those aspects that have to
mature, full grammatical competence. According to input-based accounts it is expected
because children follow input patterns. Another prediction is that children will show
a tendency to drop subjects, and do this more often in nonfinite clauses than in finite
clauses. According to maturational accounts, constraints on the sentence structure
will license subject drop in nonfinite sentences. For Dutch, one of the target languages
studied here, input-based theories will presumably predict the same because in Dutch
subjects are adjacent to finite verbs whereas they are not adjacent to nonfinite verbs.
Given the input, it is thus more likely that children store units that include subject and
finite verb than units with subject and nonfinite verb.
The two types of accounts make different developmental predictions, however.
Although maturational views leave some room for experience-related (Wexler, 1998:
26) and environmental factors (Rice, 2004: 231), their prediction would be that effects
of input quantity are rather minimal, and certainly less pronounced than predicted by
input-based accounts. Maturational accounts thus predict that bilingual children will
show minimal delays in the development of finiteness whereas input-based accounts
predict more pronounced delays. In particular, bilingual children with small amounts
of input will need more time to accumulate a sufficient amount of linguistic experience.
Also, maturational accounts may predict developmental asymmetries or selective delays
in typically developing children. That is, whereas delays will be minimal for those parts
of the grammar that mature (e.g. finiteness), delays may be more pronounced for other
aspects of language (e.g. mean length of utterance), especially those that are commonly
assumed to be input-dependent (e.g. vocabulary). A more global delay would be more
consistent with input-based accounts, however.
The International Journal of Bilingualism
426 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

4 Method
In this study, longitudinal spontaneous speech data of four Turkish–Dutch bilingual
children were analyzed. The children were selected because of their different input
situations (Backus & Van der Heijden, 1998, 2002). The corpus consists of 42 transcripts
of sessions that were aimed at collecting Dutch production data and 42 transcripts of
sessions that were aimed at the collection of Turkish production data. In addition there
are data from two monolingual Dutch and two monolingual Turkish controls.

Data collection
Data were collected over a period of a year and a half, when the children were between
the ages of 2,0 and 3,06. The timepoints of the transcriptions correspond with ages that
were held constant across all children. For each child, seven transcriptions in CHAT
format (MacWhinney, 2008) are available with generally 3-month intervals in between.
Each type of recording (Dutch, Turkish) was made by a native speaker of the target
language. Recordings of participants who were brought up at home were made in the
family’s home. The recordings of the participants who attended a daycare were at the
daycare. In the homes, the mother of the child was present most of the time, sometimes
accompanied by the child’s siblings, father or occasional visitors. At the daycare centers,
the child was surrounded by the other children attending the center, caretakers, personnel
and sometimes also by visiting parents. The data were collected using a jacket with a
wireless transmitter and a microphone that the children wore on top of their normal
clothing. The length of the recordings varied considerably, depending on the talkativeness
of the participant and location. Recordings at home varied in length between 45–160
minutes (mean ≈ 90 minutes). The length of the daycare recordings varied between 70–220
minutes (mean ≈ 150 minutes). Though these recordings are considerably longer, they
generally contain more and longer episodes during which the informant hardly talked.

Participants
The following five criteria were used in selecting bilingual children: (1) the mother
tongue of the child’s father and mother is Turkish; (2) the parents of the child are of a
low educational level; (3) the parents of the child originate from rural sites in Anatolia;
(4) the child is a girl; (5) the child is born in the Netherlands. Gender is kept constant
across participants in order to exclude differences between participants as an effect of
gender. For instance, Craig and Washington (2004) have shown that among African
American children boys show greater density of effects of African American English in
their speech than do girls. Also, the input to boys and girls may be different.
Moreover, the participants were selected based on expected language dominance
patterns. The details of this will be given later. We will refer to the participants using codes
that are informative with respect to their language situation: ‘BD’ stands for ‘bilingual,
predominantly exposed to Dutch’, ‘BT’ stands for ‘bilingual, predominantly exposed to
Turkish’, ‘BU’ will be used for a bilingual child that has a rather unclear input pattern
that is neither clearly Dutch-dominant nor clearly Turkish-dominant. If a code applies
to more than one child, the code will be followed by a number.
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 427

BD has been to an international daycare center since she was 10 weeks old, for five
days a week, from 9 a.m. to 5 p.m. The center had appointed a Turkish-speaking and at
least one native Dutch-speaking caretaker for every group. During the entire period she
went to daycare, BD was the only child of Turkish origin visiting the center regularly
and almost all of the children in BD’s group were of Dutch origin. Consequently, the
Turkish-speaking caretakers did not use Turkish very often. Though the official language
policy was claimed to be the ‘one person one language’ strategy, Turkish-speaking
caretakers regularly spoke Dutch to BD. Input of the caretakers contained standard
varieties of Turkish and Dutch. Only some phonological characteristics displayed the
regional background of the caretakers. All Turkish-speaking caretakers involved had
acquired a (near) native proficiency in Dutch. At the daycare center, much attention
was paid to creating a stimulating environment for language acquisition: children were
repeatedly asked to tell about events they had experienced, caretakers told stories, songs
(some in Turkish) were sung at least once a day, ritual activities were verbalized and in
the toddlers’ group, picture books were read and discussed, both with the whole group
and with individual children. BD’s parents had been living in the Netherlands for about
7 years. Both parents had a higher level of proficiency in Turkish than in Dutch. The
parents reported they usually spoke Turkish to BD, only her mother occasionally used
some Dutch. Outside of the daycare center BD was reported to play most of the time
with her cousins.
BU1 had been to an international daycare center ever since she was 6 weeks old.
Until the age of 2 she attended the center three days a week. From then on, her parents
decided to bring her to the daycare center for five days a week from 9 a.m. to 5 p.m.
BU1 went to the daycare center regularly. In all the groups BU1 attended, a Turkish-
speaking caretaker and at least one native Dutch-speaking caretaker were present. The
ethnic composition of the groups BU1 attended was mixed, Turkish children being the
largest ethnic group. Turkish-speaking caretakers were supposed to use Turkish when
speaking to a (group of) Turkish child(ren), and Dutch when addressing a Dutch child or
a mixed group of children. In practice, this language policy was not always adhered to.
Turkish caretakers used Turkish most of the time when addressing a (group of) Turkish
child(ren). Dutch caretakers now and then used several Turkish words, especially key
words to regulate the behavior of Turkish children. The language input by the caretakers
consisted of standard varieties of Turkish and Dutch. Only phonological characteristics
of the regional dialect were expressed by the Dutch caretakers. The Turkish caretakers
involved had a near-native level of proficiency in Dutch. Verbal interaction was to a much
larger extent left to the child’s initiative than was the case in the daycare center discussed
earlier. BU1’s parents had been living in the Netherlands for about 20 years. Both of them
were more proficient in Turkish than they were in Dutch. At the onset of data collection,
BU1 had two elder sisters and one elder brother, aged 16, 14, and 13 respectively. Her
siblings seemed to be proficient in both Turkish and Dutch. According to BU1’s mother,
Turkish was the main language of communication in the family. Furthermore, the family
lived in a neighborhood with a high number of Turkish immigrants.
BU2 was being brought up at home. Only at the end of the period of data collection,
from the age of 3,4 onwards, did she go to a daycare center for two half days a week.
BU2’s group at the daycare consisted of 3- and 4-year-olds. Though there were quite
a few Turkish (and other minority) families living in the neighborhood of the daycare
The International Journal of Bilingualism
428 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

center, BU2 was the only child of non-Dutch origin. The caretaker was of Dutch origin
as well, but a trainee visiting the daycare for half a day a week was of Turkish origin.
At the daycare, a stimulating environment for language acquisition was created. BU2’s
parents had been living in the Netherlands for about 9 years. BU2’s parents, especially
her mother, were highly proficient in Dutch. In daily communication their level of
proficiency in Dutch was nearly or completely equal to their level of Turkish. While
BU2’s parents used the standard variety of Dutch, their variety of Turkish was clearly
regionally marked (Burdur dialect). BU2’s family had contacts with both Turkish and
Dutch people. BU2 had one elder sister, who was 7 years of age at the beginning of the
data collection and who attended primary school.
During the whole period of data collection, BT was being brought up at home. She
had two older sisters, the eldest of whom was 11 years of age and went to primary and
secondary education in the period of data collection; the other girl was 4 years of age
and attended kindergarten. BT’s mother had lived in the Netherlands for about 15 years.
BT’s father did not live with his family. BT’s family lived in a middle-class neighborhood
with mostly Dutch inhabitants. There was only one other Turkish family living nearby.
The family’s closest contacts were with their relatives. Some of them lived in their house
during part of the data collection, when BT was aged 2,0 to 2,5. Furthermore, BT’s family
had frequent contact with one of their Dutch neighbors. Especially in summer, BT and
her sisters used to play with the children of this family. BT’s mother was more proficient
in Turkish than in Dutch. Her level of proficiency in Dutch satisfied her needs in daily
conversations. To her children, BT’s mother spoke Turkish most of the time, though now
and then she turned to Dutch as well. She wanted her children to speak Turkish at home,
but reported that especially her oldest daughter refused to speak Turkish.
Table 1 summarizes the four different input situations of the four bilingual children.

Table 1
Input situations of bilingual participants

Child Daycare situation Home situation Input quantity Dutch

BD Attends a Turkish–Dutch Parents are more proficient in Large


daycare 5 days a week; Turkish than in Dutch. Turkish is
Dutch-dominant; much the main language used at home by
stimulation of verbal the parents to address the children.
interaction.
BU1 Attends a Turkish–Dutch Parents are more proficient in Medium
daycare 3–5 days a week; Turkish than in Dutch. Turkish is
moderate stimulation of the main language used at home by
verbal interaction. the parents to address the children.
BU2 Brought up at home until Parents, especially mother, are Medium
age 3,4. highly proficient in both Turkish
and Dutch.
BT Brought up at home; Mother is more proficient in Small
Turkish-dominant. Turkish than in Dutch.

The following four criteria were used in selecting monolingual Dutch participants:
(1) the mother tongue of the child’s father and mother is Dutch; (2) the parents of the
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 429

child are of a low educational level; (3) the child is a girl; (4) the child was born in the
Netherlands. As the control children are confronted with only one language, going to
a daycare center or being brought up at home does not affect the balance of languages
in the input the children are confronted with. The code ‘MD’ that will be used stands
for ‘monolingual Dutch’. Likewise ‘MT’ will be used to refer to monolingual Turkish
children later on. MD1 and MD2 attended the same daycare center from the age of 2
years onwards. With the exception of one Indonesian/Dutch child and one Surinamese
trainee, all children and staff were of Dutch origin. Data from two monolingual Turkish
participants, MT1 and MT2, were collected in Turkey. The participants met the following
criteria: (1) the mother tongue of the child’s father and mother is Turkish; (2) the parents
of the child are of a low educational level; (3) the parents of the child originate from
rural sites in Anatolia; (4) the child is a girl. Both monolingual Turkish participants
were raised at home.
The small number of participants in this study prevents us from grouping. Yet,
the dataset has a number of important advantages. Data have been collected at seven
different timepoints between ages 2,0 and 3,06. The longitudinal design allows for
comparisons of developmental patterns across children. All children have the same age,
gender and parental level of education (hence socio-economic status, see: Ensminger
& Fothergill 2003). Given that socio-economic status does influence children’s gram-
matical development (e.g. length of utterances and scores on standardized tests), this is
a relevant property of the database (Arriaga, Fenson, Cronan, & Pethick, 1998, Snow
1999). Finally, across children, similar procedures have been used for collecting data.

Predictions
Using input quantity in Dutch as an independent variable, we will compare the chil-
dren’s developmental patterns for two general developmental measures and one specific
morphosyntactic measure. The two general measures are D (Malvern & Richards, 1997)
for vocabulary development and MLU-W (Brown, 1973) for grammatical develop-
ment. The dataset comprises transcripts of different sizes. Therefore, D is much more
reliable than type-token ratio, which is a well-known measure of lexical diversity as
well (MacWhinney, 2008: 113). Based on previous research, we expect that the vocabu-
lary development will show effects of input quantity (Hoff & Naigles, 2002; Pearson,
Fernandez, Lewedeg, & Oller, 1997). Hence, we expect that the four Turkish–Dutch
bilingual children (BD, BU1, BU2, BT) would have lower D scores than the Dutch (MD1,
MD2) and Turkish (MT1, MT2) monolingual controls. We might also expect differences
in D between bilingual children as an effect of differences in input quantity between
these children. MLU-W is considered to be a reliable general measure of the grammatical
development of pre-school children. If grammatical development is influenced by input
quantity, we expect that the four bilingual children would have lower scores for MLU-W
than the monolingual children. In addition there may be differences between bilingual
children in MLU-W that relate to the amount of input.
The specific morphosyntactic feature in this study is finiteness; this will only be
investigated in Dutch. Dutch is, unlike Turkish, a non pro-drop language, and displays
the typical characteristics of an Optional Infinitive language (e.g. early omission of
finite verbs and overuse of infinitival verbs; see Section 3 for details). Maturational
accounts predict that the bilingual children will show minimal delays in the development
The International Journal of Bilingualism
430 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

Figure 1
D in Dutch

of Dutch finiteness and develop at approximately the same rate as their monolingual
peers. Input-driven accounts would predict pronounced delays. Within the group of
bilinguals, the prediction is moreover that BD will be most similar to the monolingual
children. BT, on the other hand, is expected to exhibit the largest age gap with the Dutch
monolinguals. The two bilingual children BU1 and BU2 are expected to be in between
these two more extreme cases. Moreover, input-based accounts predict overall slower
development, whereas the finding that there are no effects of input quantity for finiteness
in combination with input effects for MLU-W and D (hence selective delays) would be
more consistent with maturational approaches.

5 Results
General developmental measures
Utterances in the data of the bilinguals that contained code-switches were excluded
(Backus & Van der Heijden 2002). With respect to the calculation of D exclusion of
mixed utterances is obvious. However, with respect to the calculation of MLU-W this
strategy is reasonable as well. Children may expand their morphosyntactic abilities
in the weaker language by code-mixing from the dominant language (Bernardini &
Schlyter 2004). Moreover, the inclusion of utterances with code-switches may make it
difficult, or even impossible, to assign an utterance to a particular language (Paradis
2010). Figures 1 and 2 display the development of D and MLU-W in Dutch. Note that
MD1 and MD2 are the two monolingual controls. Recall that BD is the bilingual child
who within the group of bilinguals receives most Dutch input quantity whereas BT
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 431

Figure 2
MLU-W in Dutch

receives least Dutch input quantity. In between are the bilingual children with medium
input quantity, BU1 and BU2. Calculation of D requires a minimum of 50 tokens per
transcript. Following Brown’s criteria (1973: 54), MLU-W has only been calculated for
transcripts that contain at least 100 utterances. Empty cells represent those timepoints/
transcripts with insufficient numbers of word tokens or utterances.
The results show a difference between the monolingual children, on the one hand,
and the bilinguals, on the other hand: D of the monolinguals ranges between 48 and 112,
whereas D of the bilinguals ranges between 8 and 53. There is thus very little overlap
in the ranges of D of the monolingual and bilingual children who have the same age.
The bilingual child with large Dutch input quantity (BD) has a mean lexical diversity
(M = 35.2, SD = 11.5) that comes closest to those of the monolingual children MD1
(M = 75.5, SD = 18.4) and MD2 (M = 65.1, SD = 11.7). However, BD’s score does not
differ markedly from the score of BT, who has a small Dutch input quantity (M = 28.7,
SD = 13.7). Also, BT has a higher D score in mean over time than the two children with
medium input quantity: BU1 (M = 17.8, SD = 6.1) and BU2 (M = 21.4, SD = 8.44).
Both monolingual children MD1 (M = 2.36, SD = .38) and MD2 (M = 2.47, SD = .49)
show a gradually increasing MLU-W. The bilingual children BD and BU2 have at T1 an
MLU-W that is similar to the MLU-W of the two monolinguals. Their mean MLU-W
is lower, though (BD (M = 2.14, SD = .41), BU2 (M = 2.11, SD = .44). A comparison
of the developmental patterns suggests that this may be caused by a somewhat slower
development. The MLU-W of the bilingual children BU1 (M = 1.59, SD = .21) and BT
(M = 1.64, SD = .3) is clearly below that of the other four children, and their MLU-W
hardly develops (BU1) or seems to develop very slowly (BT). At T6, which corresponds
The International Journal of Bilingualism
432 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

Figure 3
D in Turkish

to age 3,03, their MLU-W resembles the MLU-W of the other four children at T1 (age
2,0/2,01). These results suggest that input quantity is relevant with respect to the develop-
ment of MLU-W: the bilingual child BD received most exposure to Dutch and shows a
development of MLU-W that is rather similar to that of the monolinguals whereas the
bilingual child BT, who received a relatively small amount of input quantity in Dutch
has clearly lower MLU-W scores.
Suppose that the measured effects are not the result of the input, but that BD is
in general the most linguistically advanced child, and that BT shows generally weak
linguistic abilities. In that case, we would expect this to be reflected in their Turkish.
Figures 3 and 4 show D and MLU-W in the Turkish-aimed sessions.
Overall, the Turkish D-scores are higher than the Dutch D-scores. This may be
an effect of the agglutinative character of Turkish. In a fusional language like Dutch,
various functions are encoded in one affix whereas in an agglutinative language each affix
encodes one function. The means of the Turkish monolingual children MT1 (M = 168.8,
SD = 41.6) and MT2 (M = 78.6, SD = 25.5) vary greatly, and all scores of the bilinguals
fall in the monolingual range (41–211). In other words, we do not observe any effects of
input quantity. There is little variation between bilingual children. BT (M = 2.18, SD = .42)
and BU2 (M = 2.09, SD = .3) have mean MLU-W scores that do not differ from those
of the monolingual children MT1 (M = 2.11, SD = .215) and MT2 (M = 2.07, SD = .47),
whereas the mean MLU-W of BU1 (M = 1.96, SD = .28) and BD (M = 1.82, SD = .28) are
lower and the development is less steady in these children. Focusing on the two extremes,
these data confirm that BD is Dutch-dominant whereas BT is Turkish-dominant. The
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 433

Figure 4
MLU-W in Turkish

reverse ordering of BT and BD in Dutch and Turkish shows that general verbal weakness
or strength cannot explain the differences between the bilingual children.
This section focused on the relation between input quantity, on the one hand,
and lexical diversity (D) and grammatical development (MLU-W), on the other. Both
measures reflected effects of input quantity, albeit in a different fashion: whereas in
Dutch, D distinguished between monolingual children, on the one hand, and bilinguals,
on the other, MLU-W also made more fine-grained distinctions between different bilin-
gual children. A comparison of the MLU-W of the Dutch and the Turkish data within
each bilingual child disconfirmed that individual differences reflect general linguistic
weaknesses or strengths. There also were a number of observations that could not be
explained by input quantity. First of all, the bilingual child BT, who had a small amount
of input quantity in Dutch, had an unexpectedly high lexical diversity score in Dutch.
Secondly, lexical diversity in Turkish did not reflect effects of input quantity. We will
return to these issues in the general discussion.

Finiteness
Before turning to the results, we first explain the basic properties of finiteness in Dutch.
Dutch finiteness is marked in syntax through placement of the verb in second position:
in declarative main clauses, nonfinite verbs are placed in final position whereas finite
verbs are in second position (Verb Second). In addition, finiteness is expressed through
verb morphology: infinitives consist of Vstem+en whereas finite verbs in singular contexts
are either bare (first person) or Vstem+t (second/third person singular). Finite plural
forms are homophonous with infinitives:
The International Journal of Bilingualism
434 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

Figure 5
Growth curves of
finite utterances
(Dutch)

(1) a. Ik vlieg niet graag


I fly-1SG not eagerly
‘I do not like to fly’
b. De vogel vliegt door de lucht
The bird fly-3SG through the sky
‘The bird is flying through the sky’
c. Vogels vliegen van boom tot boom
Birds fly-PL from tree to tree
‘Birds fly from tree to tree’
d. Vogels kunnen hoog vliegen
Birds can-FIN high fly-INF
‘Birds can fly high’

In the transcripts, all declarative utterances were coded for verb placement, verb
form and subject realization. ‘Finite’ were verbs in first/second position and in non-final
position. ‘Nonfinite’ were verbs in final position in utterances that contain more ma-
terial than only a subject. Two-word sentences with a subject in first position were coded
as ‘unclear’, irrespective of verb form, as were all other sentences in which it could not
be determined if the verb was in finite or nonfinite position. Additionally, codes were
assigned for the correctness of a certain verb form in a position.
Figure 5 shows the development of finite utterances. As a cut-off point we analyzed
data from timepoints with a denominator that was equal to 5, or greater. Finite utterances
comprise utterances with simplex (finite main verb) and complex verbs (finite auxili-
ary + nonfinite main verb). The number of finite utterances (numerator) is a proportion of
the sum of finite and nonfinite declarative utterances that contained a verb (denominator).
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 435

Table 2
Development of finite main verb, numbers of types (tps) and tokens (tkns) between T1 and T7

MD1 MD2 BD BU1 BU2 BT


Timepoint tps/tkns tps/tkns tps/tkns tps/tkns tps/tkns tps/tkns

T1 4/5 8/21 3/3 1/1 4/4 1/1


T2 6/15 10/30 6/12 1/2 1/1 2/2
T3 8/30 9/33 4/28 2/2 3/7 0/0
T4 14/42 8/34 11/33 3/5 4/5 1/1
T5 12/41 14/45 12/28 4/6 10/49 1/1
T6 11/56 13/63 12/75 3/7 11/27 1/1
T7 22/93 17/64 12/58 4/5 11/47 1/1

Monolingual Dutch children generally show a steady increase of finite utterances


(Blom, 2008a; Wijnen, 2000; Wijnen & Bol, 1993). In our sample, the increase in the
proportion of finite utterances is most clearly visible in the data of the monolingual
children MD1 and MD2, and the bilinguals BD and BU1. Between T1 and T4, the
proportion of finite utterances in the data of BU2 seems to develop in a typical fashion.
However, between T4 and T7 her proportion of finite utterances does not increase any
further. At T4, the proportion of finite utterances in the data of BT appears to be at a
monolingual level. However, data from later stages indicate that there is no development
at all and that at ages 3,03 and 3,06, her proportion of finite utterances is similar to those
of MD1, MD2, BD and BU2 1 year (or even more) earlier.
The foregoing data provide information on the growth of finite utterances. It has
been argued, however, that verb type frequency is more telling with respect to produc-
tivity of grammatical rules (Bybee, 2001, 2010; Clark, 1993), and is more revealing with
respect to developmental delays (Rieckborn, 2006, cited in Meisel, 2007). Intuitively,
one might suspect that sample size influences the number of verb types, hence that verb
type development expressed in raw numbers reflects differences in sample size rather
than input quantity. However, type:token ratios become unreliable if samples are of
unequal size, indicating that type frequency is relatively independent of token frequency
(MacWhinney, 2008). Therefore, Table 2 displays the raw verb type frequency of finite
verbs. The token frequency of finite verbs is added as well. The results in Table 2 are
limited to finite main verbs, and do not include auxiliary verbs. The reason for excluding
auxiliary verbs in the type (and hence token) counts is that this is a small, closed class
of verbs with distinctive morpho-syntactic properties.
The monolingual children show an increase of both the frequency of finite main verb
types as well as tokens; the number of verb tokens grows disproportionally in comparison
to the number of verb types. As a consequence, the numbers of finite verb types increase,
whereas during the same period the type:token ratios of finite verbs declines. Despite
fluctuations, BD displays a growth of finite verb types. Moreover, the size of her finite
verb lexicon is similar to that of the monolinguals. In all, her development of finiteness
does not differ markedly from that of the monolingual children. Whereas Figure 5 gave
the impression that BU2’s development of finiteness stagnated between T4 and T7, this
conclusion is contradicted by the verb type data: between T4 and T7 BU2 starts to use
more verb types, until her number of verb types resembles that of the monolingual
The International Journal of Bilingualism
436 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

children. Although Figure 5, and in particular the data of T5, suggested that BU1 catches
up, the results in Table 2 reveal that her expression of finiteness remains limited until
T7. Both BU1 and BT show neither a steady increase of finite verb types, nor do they
show the decline in the type:token ratio of finite verbs that is typical for monolingual
children, and that is also present in the data of BD and BU2: until the age of 3,06, BT’s
and BU1’s use of finite main verbs remains extremely low.
Monolingual Dutch children produce utterances that lack finite verbs, but they
infrequently err in using infinitival verbs in finite position and vice versa (De Haan, 1987;
Poeppel & Wexler, 1993). What do bilingual children do? It was counted how often the
children used the bare verbal stem or Vstem+t in finite position. Because Vstem+en is not
only the infinitival verb form in Dutch, but also the finite plural form, this analysis was
limited to singular contexts. The same counting was made for verb forms in nonfinite
final position. Data from several timepoints were pooled in order to be able to calculate
contingencies (using the Fisher Exact Probability Test): Period I is T1–T3, Period II is
T4–T5 and Period III is T6–T7. A period is thus simply a collection of transcripts of
subsequent timepoints. For BU1 there were enough analyzable data available in Periods
II and III, for BT only in Period III. To keep the analysis constant across children, data
from the monolinguals were pooled as well.
Participants MD1, MD2 and BD show highly significant relations between verb
form and placement in all three periods (see the contingency tables: Appendix 1). No
significant results were obtained for BU1 in Period III and BU2 in Period I. In Period
III, BU1 used hardly any Vstem+en forms (n = 1), which renders the p-value of the
period unreliable (one cell of the contingency table contains zero observations, and the
observations in the cells are very uneven). In Period I, BU2 used Vstem+en relatively often
incorrectly in finite position. In order to detect whether this occurs in early monolingual
development as well, more detailed monolingual data from different timepoints were
analyzed. At T1, MD1 shows no significant relation between verb placement and form
(p = .06). Thus, the results of BU2 are not deviant from those of a monolingual child;
both may indicate typical variation early in development.
In early child Dutch, utterances containing a nonfinite verb most often lack subjects
whereas children do realize subjects in finite utterances (Haegeman, 1994; Krämer,
1993). The same contingency is present in the data of the monolingual children in our
sample (Appendix 1), though in the beginning there seems to be some development in
this respect: analyses of the timepoints in this period reveal that MD1 and MD2 show
at T1 a weak significance (p = .025 and p = .03), and MD1’s data show no contingency at
T2 (p = .315). At all other timepoints, the results are significant. However, the same does
not hold for the bilingual children: BD only shows a contingency in Period III (and not
before), BU1 in Period II but not in Period III, BU2 in none of the three periods and in
the only period that we could analyze for BT, Period III, there was no relation between
finiteness and realization of the subject.

Discussion
We set out to investigate to what extent input quantity influences children’s early gram-
matical development. To this end, longitudinal data of four Turkish–Dutch bilingual
children with different input situations were analyzed and compared to monolingual
control data. Apart from two general measures (lexical diversity and mean length of
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 437

Table 3
Overview of findings for the bilingual children based on all measures applied

Measure BD BU1 BU2 BT

D-Dutch – ! ! !
MLU-Dutch "#! ! "#! !
(development but slower) (development
but slower)
D-Turkish " " " "
MLU-Turkish "#! "#! " "
(lower mean, less steady (lower mean, less
development) steady development)
Growth of finite " "#! "#! !
sentences (lower number) (stagnation)
Number of finite " ! " !
verb types

utterance) for Turkish and Dutch, we studied the development of finiteness in Dutch.
Investigating finiteness, we aim to discern between maturational accounts and input-
driven accounts of early grammatical development. Whereas maturational accounts are
more compatible with minimal delays, in particular for the Optional Infinitive stage,
input-based accounts predict more pronounced and more global delays.
We start with an overview of the main findings on all measures applied, and continue
with an interpretation of these findings. Table 3 shows whether each of the bilingual
children is similar (+) or different (–) from the monolingual controls on a certain measure.
In case there is similarity in some respects, but a difference in others, +/– is assigned. In
this case, a brief explanation is given.
In Dutch, the four bilingual children had lower scores for lexical diversity than the
two monolingual controls. This difference held consistently across ages 2,0 and 3,6. This
finding is in line with other studies that compared the vocabulary development of mono-
lingual and bilingual children (Pearson et al. 1997), and is compatible with the hypothesis
that input quantity influences vocabulary development. Between bilingual children,
input quantity does not account for the differences in lexical diversity. Moreover, lexical
diversity in Turkish did not distinguish between monolingual and bilingual children.
Results for MLU-W in Dutch suggested a three-way split: the monolingual children
MD1 and MD2 had the highest MLU-W, followed by BD and BU2, who, in turn, were
followed by BT and BU1. MD1, MD2, BD and BU2 showed a similar increase of their
MLU-W whereas BT and BU1 showed hardly any development. At the age of 3,06 their
MLU-W in Dutch was still below 2, and resembled the MLU-W of the other children at
the age of 2,0/2,01. In Dutch, BD had the highest and BT (together with BU1) the lowest
MLU-W. In their Turkish, BD had the lowest and BT the highest MLU-W of all four
bilingual children. The bilingual child BD can thus be considered Dutch-dominant in
her language development whereas BT shows a Turkish-dominant language development,
as expected given their input profiles. With respect to these two more extreme cases, the
The International Journal of Bilingualism
438 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

results of this study confirm that higher input quantity leads to a relatively high MLU-W
whereas small input quantity results in a low MLU-W.
The finiteness data revealed that both monolingual controls and the bilingual
children BD and BU2 develop productive use of finiteness. Between ages 2,0 and 3,6 BU1
shows an increasing proportion of finite utterances, but during this period her number
of finite main verb types hardly grows. BT has throughout this period a low proportion
of finite utterances and a low number of finite main verb types. At the age of 3,06, BU1
and BT behave like the other children at ages 2,0/2,01. Hence, BU1 and BT do not only
have a lower MLU-W than BD and BU2, but also show pronounced delays with respect
to the development of finiteness. Note that both BU1 and BT had extremely low numbers
for type and token frequencies of finite verbs. Considering the numbers of utterances
in Dutch produced by the two children, the numbers for finite verbs are disproportion-
ally low (Appendix 2). Thus, until a rather later age, both children used in their Dutch
many one-word utterances and multi-word utterances without a verb. This behavior
also indicates a slower development in Dutch, the weaker language of these children.
Finally, we investigated contingencies between verb form and verb position and
between finiteness and subjects. The bilingual children differed from the monolin-
gual children with regard to the contingency between finiteness and subject use. This
contingency was steadily present in the data of the monolinguals whereas it was most
of the time absent in the bilingual data. Before we turn to the discussion of possible
explanations and implications of this observation, we will first attempt to determine
to what extent the behavior of the bilinguals can be considered to be extreme, and falls
outside the normal range for monolingual children.
Previous studies on monolingual child Dutch have shown that proportions of
realized subjects in finite utterances and nonfinite clauses vary over time (Blom & Van
Geert, 2004). In early development, the contingency can thus be weak or even absent, as
confirmed by recalculation of data reported by Blom and Van Geert (Appendix 3). The
data from MD1 and MD2 conform to this pattern. Therefore, we can establish, based
on data of eight monolingual children, a range for age and MLU from which typically
developing monolingual Dutch children have a reliable contingency between finiteness
and subject realization. The age ranges between 1,10 and 2,05 (mean 2,02) and MLU
ranges between 1.4 – 2.23 (M = 1.77, SD = .4). Turning to the four bilingual children
in our sample, we find that only BD has at the age of 3,03/3,06 a significant relation
between finiteness and realization of subjects. At this age, her MLU is 2.5/2.7. That is,
BD establishes the relation at an age and MLU above the monolingual range. BU2’s
data do not contain a relation between finiteness and realization of subjects throughout.
Whenever BU2 establishes this relation, it will necessarily be at a later age than monolin-
gual children, and her MLU will be higher as well. The results of BU1 and BT are more
difficult to interpret. BU1 has a rather unstable profile, showing only a relation in Period
II. This unstable profile, as well as the absence of the contingency at the age of 3,03/3,06,
is not what we find in the monolingual data. BT shows no contingency throughout as
does BU2, but at the age of 3,06, the MLU of BT is still below 2. BT will thus certainly
be older than monolingual children when she establishes a relation between subject use
and finiteness, but at this age her MLU might still fall within the monolingual range.
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 439

In sum, it seems warranted to conclude that the bilingual children in our sample
acquire the relation between finiteness and subjects at a later age than monolingual
children. Unlike the other grammatical measures we investigated, this delay affects
all the bilinguals in our sample. In other words, there is an asymmetry between the
development of MLU and growth of finiteness, on the one hand, and establishing the
relation between finiteness and subject realization, on the other hand. Whereas the
former can be explained as a function of input quantity, which distinguishes between
bilingual children, the latter seems rather an effect of bilingualism, which distinguishes
the bilingual children from the monolinguals.
One factor that could be relevant here is cross-linguistic influence. It has been
found that cross-linguistic influence occurs more clearly from the dominant to the weak
language, than the other way around (Döpke, 1998; Yip & Matthews, 2000). Interestingly,
BD, the bilingual child who received most input in Dutch and least input in Turkish of
all four children, is the only bilingual child who establishes a target-like relation between
realization of the subject and finiteness in her Dutch. Also, in terms of proportions of
dropped subjects, dominance-patterns seem to correlate with the severity of subject
drop in Dutch. Earlier we concluded that BD is a Dutch-dominant child whereas BT
is a Turkish-dominant child. Focusing on Period III, which is the only age range that
has a sufficient number of analyzable utterances for all children, we observe that in the
speech of BD the proportion of dropped subjects is .23 (n = 177) whereas this proportion
is .62 (n = 16) in the speech of BT (see Appendix 1 for the relevant data).
It has been argued that cross-linguistic transfer can cause delays if there is surface
overlap between the two languages, which happens when there is superficial similarity
between two languages that have a different underlying grammar (Hulk & Müller, 2000).
To a certain degree we may indeed find such surface overlap between Dutch and Turkish.
Because Turkish is a pro-drop language (Lewis, 1967), subjects are generally left out in
finite sentences. Dutch is a non pro-drop language, hence subjects are usually realized
in finite sentences. However, if a subject functions as the topic of a sentence it can be
omitted (Jansen, 1981; Weerman, 1989). Note that although discourse requirements
(e.g. recoverability of the topic) do play an important role, topic drop in Dutch is also
a syntactic phenomenon given the positional restriction that constituents can only be
dropped from first position and the functional asymmetry between object drop, which
is restricted to drop of third person objects, and subject drop, for which no such restric-
tions exist. Thus, pro-drop in Turkish and topic-drop in Dutch both lead to allowance of
subject drop in the two languages, creating surface overlap between the two languages.
The overlap may facilitate cross-linguistic influence, more specifically, ‘overdrop’ in
the Dutch of Turkish–Dutch bilinguals. As a consequence, Turkish–Dutch bilingual
children may show delays with respect to the target-like use of subjects.

Conclusion
The observation that 3-year-old bilingual children do not only have a mean length of utter-
ance, but also a proportion of finite utterances and a finite verb type profile similar to that
of a 2-year-old monolingual child, confirms the hypothesis that less input quantity slows
down grammatical development in a non-selective manner. These differences are limited
to the weaker language of bilingual children, showing that children’s ability to acquire
grammar is robust: only clearly reduced input will result in a protracted development. Some
The International Journal of Bilingualism
440 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

of the bilingual children displayed a delay of (more than) 1 year, which is pronounced,
especially given the children’s young age. The pronounced, non-selective delays observed
in the present study do not support hypotheses that relate the development of finiteness
to maturation of knowledge of language. The results are compliant with input-driven
perspectives on language development (e.g. usage-based grammar, construction grammar),
however. All bilingual children in this study experienced difficulties in establishing the
relation between finiteness and subject use. It was argued that this is an effect of cross-
linguistic influence.

6 General discussion
Our study revealed effects of the language environment in which bilingual children
are being raised. More specifically, it was found that the amount of input affects bilin-
gual language development. The clear majority of studies that investigated the early
grammatical development of bilingual children did not find any differences between
monolingual and bilingual children (Genesee, 2001; Meisel, 2001; Nicoladis & Genesee,
1997; Paradis & Genesee, 1996). However, most of these studies discuss bilingual children
in a bilingual setting that is clearly different from the bilingual setting studied here.
Also, we compared children with different dominance patterns and analyzed data from
the weaker language of bilingual children, whereas most previous work is about the
dominant language only (Meisel, 2007: 496; Paradis et al., 2003: 3).
In this study, we applied two commonly used general measures of linguistic devel-
opment: MLU for grammatical development and D for vocabulary development. The
results for MLU tie in with observations reported elsewhere. The bilingual children
BD and BU2 (large/medium input quantity) had higher MLU scores than the bilingual
children BU1 and BT (medium/small input quantity). Lower MLUs for the weaker
language of young bilingual children are also reported by Schlyter (1993) and Meisel
(2007). Unlike Schlyter (1993) we did not find errors that resembled those made by non-
native late acquirers in the data of BU1 and BT. For instance, both children showed a
clear contingency between verb form and verb placement, even though this contingency
is typically absent in the Dutch of late, non-native acquirers who have Turkish as their
first language (Blom, 2008b). With respect to finiteness, BD and BU2 were not delayed,
like the balanced bilingual children studied by Paradis and Genesee (1996), but we did
find delays in those bilingual children whose weaker language was Dutch (BU1 and
BT). This result underscores the importance of controlling for dominance pattern and
either keeping dominance pattern constant or varying dominant and weak languages,
depending on the question that is being asked.
Another issue that deserves attention is the social setting of bilingual children,
because social setting is an important determinant of children’s linguistic environment
(see for an overview Hoff, 2006). In this study, we described the grammatical develop-
ment of Turkish–Dutch bilingual children in a social setting that is quite different
from that of the one-parent, one-language, middle-class setting of bilingual children
in many other studies. Most Turkish immigrants came to the Netherlands as a result of
labor migration and family reunification. Being part of a substantial ethnic minority,
Turkish–Dutch children are raised in a bilingual community, which implies that they
are exposed to code-switching, convergence and ethnic varieties. Addressing the social
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 441

characteristics of the acquisition setting and their influence on language development


goes beyond the scope of this contribution, but we do want to emphasize that, in future
research, it is important to ask whether these two settings, the bilingual home situation
in middle-class families and the bilingual setting situation in lower-class ethnic minority
families, are indeed equally representative of bilingual upbringing.
This brings us to the next point. Because of the design of our study, we limited our
questions and conclusions to input quantity and did not discuss the quality of the input.
Recall that MLU appeared more sensitive to effects of input quantity than lexical diversity
(D). In order to understand the seemingly unexpected behavior of D, we suggest that
input quality may be a relevant factor to take into account, though we want to emphasize
that any conclusions regarding input quality based on our study should be strengthened
in future research. In our study, D in Dutch only distinguished monolinguals from
bilinguals, and did not draw distinctions between bilingual children. As pointed out
in Oller & Eilers (2002), the language knowledge of bilinguals is ‘distributed’, with the
effect that the vocabulary of a bilingual is different in the two languages. Thus, if the
lexical diversity of bilingual children in one language is compared to that of monolingual
children learning the same language, as was done in our study, it is not surprising that
the monolinguals end up with greater lexical diversity than the bilinguals.
A closer look at effects of input quality might also help us to understand two
other findings. First, in the Turkish data, lexical diversity did not distinguish between
monolinguals and bilinguals. One factor may be that the context of exposure to Turkish
was constant across monolinguals and bilinguals (at home). Another factor is the large
range for D in monolingual Turkish, in particular the lower boundary of this range, set
by the monolingual child MT2. MT2’s mother, who was at home most of the time, was
not very talkative, and whenever she spoke to her children it usually was to command
or reprimand them. It is thus conceivable that the lower boundary was influenced by the
particular linguistic climate in which the monolingual child MT2 was raised. Secondly,
we found that the bilingual child with the smallest Dutch input quantity had a higher
D than the bilinguals with medium input quantity (BU1 and BU2) and had D scores
that resembled those of BD, who had large input quantity. Interestingly, BT is the only
bilingual child in the sample who lived in a neighborhood with few immigrants. Also,
BT’s family had intensive contact with Dutch neighbors and BT frequently played with
the children of this family. It might be that this situation influenced the development of
BT’s Dutch vocabulary, and explains why she outperformed BU1, for instance.
In future research it is thus important to explore effects of input quality and social
setting, and differential effects of these factors on the growth of the grammars and
vocabularies of bilingual children. The present study has highlighted the role of language
dominance and input quantity in the grammatical development of bilingual children.
An important outcome is that the grammar of the weaker language of bilingual children
shows a delayed development in comparison to the development of grammar in mono-
lingual children/the dominant language of bilingual children. This result supports the
view that environmental factors influence the development of grammar.
The International Journal of Bilingualism
442 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

Acknowledgements
The study reported in this article is made possible with research grants from the Niels
Stensen Stichting and the Marie Curie Actions (Proposal No. 219276 ‘ELSIC’). I am
very grateful to Hanneke van der Heijden who so generously shared with me the data
she collected and transcribed. This study would not have existed without her meticulous
work. Without the help of Ad Backus and Guus Extra from the University of Tilburg
I would not have been able to access this unique set of data. Johanne Paradis and the
reviewers gave helpful comments that considerably improved the quality of this article.

References
ARRIAGA, R. K., FENSON, L., CRONAN, T., & PETHICK, S. J. (1998). Scores on the MacArthur
Communicative Development Inventory of children from low- and middle-income families.
Applied Psycholinguistics, 19(2), 209–225.
BACKUS, A., & VAN DER HEIJDEN, A. (1998). Life and birth of a bilingual: The mixed code of
bilingual children and adults in the Turkish community in the Netherlands. In L. Johanson
(Ed.), The Mainz Meeting: Proceedings of the Seventh International Conference of Turkish
Linguistics (pp. 527–551). Wiesbaden: Harrassowitz Verlag.
BACKUS, A., & VAN DER HEIJDEN, A. (2002). Language mixing by young Turkish children
in the Netherlands. Psychology of Language and Communication, 6(1), 55–73.
BERNARDINI, P., & SCHLYTER, S. (2004). Growing syntactic structure and code-mixing in the
weaker language: The Yvy hypothesis. Bilingualism, Language and Cognition, 7(1): 49–70.
BLOM, E. (2008a). The acquisition of finiteness. Berlin: Mouton de Gruyter.
BLOM, E. (2008b). Testing the Domain-by-Age Model: Inflection and placement of Dutch verbs.
In B. Hazendar & E. Gavruseva (Eds.), Current trends in child second language acquisition
(pp. 271–300). Amsterdam and Philadelphia, PA: John Benjamins.
BLOM, E., & VAN GEERT, P. (2004). Signs of a developing grammar: Subject drop and inflection
in early child Dutch. Linguistics, 42(1), 195–234.
BROWN, R. (1973). A first language: The early stages. Cambridge MA: Harvard University Press.
BYBEE, J. (2001). Phonology and language use. Cambridge: Cambridge University Press.
BYBEE, J. (2010). Language, usage and cognition. Cambridge: Cambridge University Press.
CLARK, E. V. (1993). The lexicon in acquisition. Cambridge: Cambridge University Press.
CRAIG, H. K., & WASHINGTON, J. A. (2004). Grade-related changes in the production of African
American English. Journal of Speech, Language, and Hearing Research, 47(April), 450–463.
CRAIN, S. (1991). Language acquisition in the absence of experience. Behavioral and Brain
Sciences, 14, 597–612.
DE HAAN, G. (1987). A theory-bound approach to the acquisition of verb placement in Dutch.
In G. de Haan & W. Zonneveld (Eds.), Formal Parameters of Generative Grammar: OTS
Yearbook 1987 (pp. 15–30). Dordrecht: ICG.
DE HOUWER, A. (1995). Bilingual language acquisition. In P. Fletcher & B. MacWhinney (Eds.),
The handbook of child language (pp. 219–250). Oxford: Blackwell.
DÖPKE, S. (1998). Competing language structures: The acquisition of verb placement by bilingual
German–English children. Journal of Child Language, 25(3), 555–584.
ENSMINGER, M. E., & FOTHERGILL, K. (2003). A decade of measuring SES: What it tells us
and where to go from here. In M. H. Bornstein & R. Bradley (Eds.), Socioeconomic status,
parenting, and child development (pp. 13–27). Mahwah, NJ: Lawrence Erlbaum.
FREUDENTHAL, D., PINE, J. M., AGUADO-OREA, J., & GOBET, F. (2007). Modelling the
developmental patterning of finiteness marking in English, Dutch, German and Spanish
using MOSAIC. Cognitive Science, 31(2), 311–341.
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 443

FREUDENTHAL, D., PINE, J.M., & GOBET, F. (2006). Modelling the development of children’s use
of optional infinitives in English and Dutch using MOSAIC. Cognitive Science, 30(2), 277–310.
GATHERCOLE, V. C. M. (2002a). Command of the mass/count distinction in bilingual and mono-
lingual children: An English morphosyntactic distinction. In D. K. Oller & R. E. Eilers (Eds.),
Language and literacy in bilingual children (pp. 175–206). Clevedon: Multilingual Matters.
GATHERCOLE, V. C. M. (2002b). Grammatical gender in bilingual and monolingual children:
A Spanish morphosyntactic distinction. In D. K. Oller & R. E. Eilers (Eds.), Language and
literacy in bilingual children (pp. 207–219). Clevedon: Multilingual Matters.
GATHERCOLE, V. C. M. (2002c). Monolingual and bilingual acquisition: Learning different treat-
ments of that-trace phenomena in English and Spanish. In D. K. Oller & R. E. Eilers (Eds.),
Language and literacy in bilingual children (pp. 220–254). Clevedon: Multilingual Matters.
GENESEE, F. (2001). Bilingual first language acquisition: Exploring the limits of the language
faculty. In M. McGroarty (Ed.), Annual Review of Applied linguistics: Language and
Psychology Vol. 21 (pp. 153–168). New York: Cambridge University Press.
HAEGEMAN, L. (1994). Root infinitives, tense and truncated structures. Language Acquisition,
4(3), 205–255.
HOFF, E. (2006). How social contexts support and shape language development. Developmental
Review, 26(1), 55–88.
HOFF, E., & NAIGLES, L. (2002). How children use input to acquire a lexicon. Child Development,
73(2), 418–433.
HORNSTEIN, N., & LIGHTFOOT, D. (1981). Introduction. In N. Hornstein & D. Lightfoot
(Eds.), Explanation in Linguistics: The Logical Problem of Language Acquisition (pp. 9–31).
London: Longman.
HULK, A., & MÜLLER, N. (2000). Bilingual first language acquisition at the interface between
syntax and pragmatics. Bilingualism: Language and Cognition, 3(3), 227–244.
JANSEN, F. (1981). Syntaktische konstrukties in gesproken taal. Amsterdam: Huis aan Drie
Grachten.
KRÄMER, I. (1993). The licensing of subjects in early child language. In C. Phillips (Ed.), Papers
on case and agreement II (MIT working papers in linguistics 19: pp. 197–212), Cambridge,
MA: MIT Deptartment of Linguistics and Philosophy.
LEWIS, G. L. (1967). Turkish grammar. Oxford: Clarendon.
MACWHINNEY, B. (2008). URL (consulted 3 September 2009): http://childes.psy.cmu.edu/
manuals/clan.pdf
MALVERN, D. D., & RICHARDS, B. J. (1997). A new measure of lexical diversity. In A. Ryan
& A. Wray (Eds.), Evolving models of language (pp. 58–71). Clevedon: Multilingual Matters.
MEISEL, J. M. (2001). The simultaneous acquisition of two first languages: Early differentiation
and subsequent development of grammars. In J. Cenoz & F. Genesee (Eds.), Trends in
bilingual acquisition (pp. 11–41). Amsterdam: John Benjamins.
MEISEL, J. M. (2007). The weaker language in early child bilingualism: Acquiring a first language
as a second language? Applied Psycholinguistics, 28(3), 495–514.
MEISEL, J. M. (2008). Child second language acquisition or successive first language acquisition?
In B. Hazendar & E. Gavruseva (Eds.), Current trends in child second language acquisition
(pp. 55–80). Amsterdam and Philadelphia, PA: John Benjamins.
NEWPORT, E. L., & ASLIN, R. N. (2004). Learning at a distance: Statistical learning of non-
adjacent dependencies. Cognitive Psychology, 48, 127–162.
NICOLADIS, E., & GENESEE, F. (1997). Language development in preschool bilingual children.
Journal of Speech Language Pathology and Audiology, 21(4), 258–270.
OLLER, D. K., & EILERS, R. E. (Eds.) (2002). Language and literacy in bilingual children.
Clevedon: Multilingual Matters.
PARADIS, J. (2010). Comparing typically developing children and children with specific language
impairment. In E. Blom & S. Unsworth (Eds.) Experimental methods in language acquisition
research. Amsterdam and Philadelphia, PA: John Benjamins.
The International Journal of Bilingualism
444 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

PARADIS, J., CRAGO, M., GENESEE, F., & RICE, M. (2003). French–English bilingual children
with SLI: How do they compare with their monolingual peers? Journal of Speech, Language
and Hearing Research, 46(February), 113–127.
PARADIS, J., & GENESEE, F. (1996). Syntactic acquisition in bilingual children: Autonomous
or interdependent? Studies in Second Language Acquisition, 18(1), 1–25.
PARADIS, J., NICOLADIS, E., & CRAGO, M. (2007). French–English bilingual children’s
acquisition of the past tense. Proceedings of BUCLD 31. Somerville, MA: Cascadilla Press.
PEARSON, B. Z., FERNÁNDEZ, S. C., LEWEDEG, V., & OLLER, D. K. (1997). The relation
of input factors to lexical learning by bilingual infants. Applied Psycholinguistics, 18, 41–58.
POEPPEL, D., & WEXLER, K. (1993). The full competence hypothesis of clause structure in
early German. Language, 69(1), 1–33.
RADFORD, A. (1990). Syntactic theory and the acquisition of English syntax. Oxford: Blackwell.
RICE, M. (2004). Growth models of developmental language disorders. In M. Rice & S. Warren (Eds.),
Developmental language disorders (pp. 207–204). Mahwah, NJ and London: Lawrence Erlbaum.
RIZZI, L. (1993/1994). Some notes on linguistic theory and language development: The case of
root infinitives. Language Acquisition, 3(4), 371–393.
SCHLYTER, S. (1993). The weaker language in bilingual Swedish–French children. In K. Hyltenstam
& A. Viberg (Eds.), Progression and regression in language: Sociocultural, neuropsychological and
linguistic perspectives (pp. 289–308). Cambridge: Cambridge University Press.
SCHLYTER, S., & HÅKANSSON, G. (1994). Word order in Swedish as the first language,
second language, and weaker language in bilinguals. Scandinavian Working Papers on
Bilingualism, 9, 49–66.
SNOW, C. E. (1999). Social perspectives on the emergence of language. In B. MacWhinney (Ed.),
The emergence of language (pp. 257–276). Mahwah, NJ: Lawrence Erlbaum.
THEAKSTON, A., LIEVEN, E., & TOMASELLO, M. (2003). The role of input in the acquisi-
tion of third-person singular verbs in English. Journal of Speech, Language and Hearing
Research, 46(August), 863–877.
THOMPSON, S.P., & NEWPORT, E.L. (2007). Statistical learning of syntax: The role of transi-
tional probability. Language Learning and Development, 3(1), 1–42.
TOMASELLO, M. (2003). Constructing a language: A usage-based theory of language acquisition.
Cambridge, MA: Harvard University Press.
WEERMAN, F. (1989). The V2 Conspiracy: A synchronic and diachronic analysis of verbal positions
in the Germanic languages. Dordrecht: Foris.
WEXLER, K. (1991). On the argument from the poverty of the stimulus. In A. Kasher (Ed.), The
Chomskyan Turn (pp. 252–270). Oxford: Basil Blackwell.
WEXLER, K. (1998). Very early parameter setting and the unique checking constraint: A new
explanation of the optional infinitive stage. Lingua, 106(1–4), 23–79.
WEXLER, K. (2003). Lenneberg’s dream: Learning, normal language development and specific
language impairment. In Y. Levy & J. Schaeffer (Eds.), Language competence across popu-
lations: Towards a definition of specific language impairment (pp. 11–61). Mahwah NJ and
London: Lawrence Erlbaum.
WIJNEN, F. (2000). Input, intake and sequence of syntactic development. In M. Beers, B. van den
Bogaerde, G. Bol, J. de Jong & C. Rooijmans (Eds.), From sound to sentence (pp. 163–186).
Groningen: Center for Language and Cognition.
WIJNEN, F., & BOL, G. (1993). The escape from the optional infinitive stage. In A. de Boer,
J. de Jong & R. Landeweerd (Eds.), Language and Cognition 3: Yearbook of the research
group for theoretical and experimental linguistics of the University of Groningen (pp. 105–118).
Groningen: Center for Language and Cognition.
YIP, V., & MATTHEWS, S. (2000). Syntactic transfer in a Cantonese–English bilingual child.
Bilingualism: Language and Cognition, 3(3), 193–208.
The International Journal of Bilingualism
Blom: Input, bilingualism and grammatical development 445

Appendix 1
Contingency tables

stem+en bare stem/stem+t P

MD1 Period I V2 11 46 p < .0001


Vfinal 52 3
Period II V2 1 80 p < .0001
Vfinal 38 1
Period III V2 2 144 p < .0001
Vfinal 49 1
MD2 Period I V2 1 78 p < .0001
Vfinal 44 8
Period II V2 2 74 p < .0001
Vfinal 25 0
Period III V2 1 125 p < .0001
Vfinal 28 1
BD Period I V2 0 43 p < .0001
Vfinal 16 1
Period II V2 4 61 p < .0001
Vfinal 38 15
Period III V2 1 133 p < .0001
Vfinal 12 0
BU1 Period II V2 1 11 p = .0004
Vfinal 6 0
Period III V2 0 12 p = .14
Vfinal 1 1
BU2 Period I V2 4 12 p = .55
Vfinal 2 1
Period II V2 7 54 p < .0001
Vfinal 17 3
Period III V2 6 74 p < .0001
Vfinal 24 3
BT Period III V2 0 2 p = .0065
Vfinal 13 3

"subject !subject P

MD1 Period I "fin 49 21 p < .001


!fin 17 38
Period II "fin 119 15 p < .001
!fin 8 28
Period III "fin 186 24 p < .001
!fin 2 48
MD2 Period I "fin 81 24 p < .001
!fin 6 37
Period II "fin 123 3 p < .001
!fin 5 20
Period III "fin 165 8 p < .001
!fin 9 18
BD Period I "fin 22 25 p = .15
!fin 4 12
Period II "fin 57 36 p = .2542
!fin 19 20
Period III "fin 132 33 p = .0068
!fin 5 7

The International Journal of Bilingualism


446 INTERNATIONAL JOURNAL OF BILINGUALISM 14 (4)

Appendix 1
(Continued)

"subject !subject P

BU1 Period II "fin 9 2 p = .018


!fin 1 5
Period III "fin 6 7 p = 1.0
!fin 0 1
BU2 Period I "fin 6 9 p = 1.0
!fin 1 1
Period II "fin 30 6 p = .764
!fin 36 10
Period III "fin 59 32 p = .617
!fin 13 10
BT Period III "fin 1 2 p = 1.0
!fin 5 8

Appendix 2
Numbers of Dutch utterances in transcripts (excluding code-mixed utterances)

Timepoint MD1 MD2 BD BU1 BU2 BT

T1 374 448 153 26 100 46


T2 364 340 149 117 15 141
T3 470 394 138 23 120 30
T4 453 323 384 123 155 76
T5 407 321 333 45 309 148
T6 287 320 351 104 189 256
T7 580 423 247 71 282 132

Appendix 3
Age ranges, MLU and p-values calculated for the contingency between finiteness and realization
of the subject, based on data in Blom & Van Geert (2004)

ABEL N/A 1;10.30 – 1;11.26 2;01.02 – 2;03.02 2;05.17 – 2;07.29


MLU = 1.31 MLU = 1.87 MLU = 2.19
p = .02 p < .0001 p < .0001
DAAN 1;08.21 – 1;10.16 2;00.22 – 2;01.21 2;04.14 – 2;05.11 2;08.13 – 2;09.10
MLU = 1.10 MLU = 1.41 MLU = 2.07 MLU = 2.73
p = 1.0 p = .003 p < .0001 p < .0001
JOSSE N/A 2;00.7 – 2;00.21 2;03.28 – 2;04.11 2;07.20 – 2;08.18
MLU = 1.32 MLU = 1.98 MLU = 2.32
p = .403 p < .19 p < .0001
LAURA 1;09.04 – 1;09.18 2;00.05 – 2;01.02 2;04.01 – 2;06.10 3;02.09 – 3;04.06
MLU = 1.33 MLU = 1.42 MLU = 1.87 MLU = 2.84
p = .18 p = .0095 p < .0001 p < .0001
MATTHIJS 1;09.30 – 1;10.13 1;11.10 – 2;00.09 2;04.24 – 2;05.01 2;10.22 – 2;11.19
MLU = 1.07 MLU = 1.45 MLU = 1.83 MLU = 2.55
p = 1.0 p = .86 p = .016 p < .0001
PETER 1;07.18 1;09.20 – 1;10.03 2;00.28 – 2;01.28 2;03.07 – 2;03.21
MLU = 1.00 MLU = 1.41 MLU = 2.21 MLU = 3.01
p = 1.0 p = .69 p = .008 p < .0001

The International Journal of Bilingualism

View publication stats

You might also like