REPORT
Bilingualism Affects Infant Cognition:
Insights From New and Open Data
Rodrigo Dal Ben*
, Hilary Killam , Sadaf Pour Iliaei*
, and Krista Byers-Heinlein
*Rodrigo Dal Ben is now at Ambrose University, Calgary. Sadaf Pour Iliaei is now at the Ontario Institute for Studies in Education,
University of Toronto.
Concordia University
un accès ouvert
journal
Mots clés: infancy, bilingualism, cognitive control, inhibitory control, anticipatory looking
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
.
/
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
ABSTRAIT
Bilingualism has been hypothesized to shape cognitive abilities across the lifespan. Ici, nous
examined the replicability of a seminal study that showed monolingual–bilingual differences
in infancy (Kovács & Mehler, 2009un) by collecting new data from 7-month-olds and 20-month-
olds and reanalyzing three open datasets from 7- to 9-month-olds (D’Souza et al., 2020;
Kalashnikova et al., 2020, 2021). Infants from all studies (N = 222) were tested in an
anticipatory eye-tracking paradigm, where they learned to use a cue to anticipate a reward
presented on one side of a screen during Training, and the opposite side at Test. To correctly
anticipate the reward at Test, infants had to update their previously learned behavior. Across
four out of five studies, a fine-grained analysis of infants’ anticipations showed that bilinguals
were better able to update the previously learned response at Test, which could be related to
bilinguals’ weaker initial learning during Training. Cependant, in one study of 7-month-olds, nous
observed the opposite pattern: bilinguals performed better during Training, and monolinguals
performed better at Test. These results show that bilingualism affects how infants process
information during learning. We also highlight the potential of open science to advance our
understanding of language development.
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
INTRODUCTION
Bilingual infants are unique because they must acquire two languages simultaneously. Con-
trary to early warnings about potential disadvantages of growing up bilingual (Epstein, 1905;
Macnamara, 1966; Yoshioka, 1929), there is evidence that bilingualism leads to improved
metalinguistic awareness (Edwards & Christophersen, 1988; Yelland et al., 1993) ainsi que
enhanced executive function and cognitive control (Bialystok, 2007; Bialystok & Barac,
2013). These adaptive impacts of bilingualism on cognitive function have been reported in
studies with children (Barac & Bialystok, 2011; Bialystok, 2010; Bialystok & Martine, 2004;
see reviews by Adesope et al., 2010; Barac et al., 2014), young and middle-aged adults (Costa
et coll., 2008), and older adults (Bialystok et al., 2007; Gold et al., 2013; Kavé et al., 2008).
De plus, studies have suggested that the age at which the two languages are acquired mod-
ulates this effect. Par exemple, bilingual adults who learned two languages in infancy show
enhancements in cognitive control and brain connectivity relative to adults who learned their
second language later in life (Kousaie et al., 2017).
Citation: Dal Ben, R., Killam, H., Pour
Iliaei, S., & Byers-Heinlein, K. (2022).
Bilingualism Affects Infant Cognition:
Insights From New and Open Data.
Open Mind: Discoveries in Cognitive
Science, 6, 88–117. https://est ce que je.org/10
.1162/opmi_a_00057
EST CE QUE JE:
https://doi.org/10.1162/opmi_a_00057
Supplemental Materials:
https://osf.io/ bz8jn
Reçu: 7 Décembre 2021
Accepté: 29 Avril 2022
Intérêts concurrents: The authors
declare no conflict of interest.
Auteur correspondant:
Krista Byers-Heinlein
k.byers@concordia.ca
droits d'auteur: © 2022
Massachusetts Institute of Technology
Publié sous Creative Commons
Attribution 4.0 International
(CC PAR 4.0) Licence
La presse du MIT
Effects of Bilingualism on Cognition Dal Ben et al.
Cependant, other studies have raised concerns about the robustness of these findings, call-
ing into question the existence of cognitive enhancements in bilinguals (par exemple., de Bruin et al.,
2015; Fernández-López & Perea, 2019; Leivada et al., 2020; Paap et al., 2015; Paap et al.,
2018; Paap & Greenberg, 2013; Struys et al., 2018). Studies with infants have yielded mixed
résultats, with some showing support for bilingualism’s impact on early cognition and others
failing to do so (Brito & Barr, 2012; Brito et al., 2015; Brito et al., 2020; Comishen et al.,
2019; D’Souza et al., 2020; Ibánez-Lillo et al., 2010; Kalashnikova et al., 2020; Kovács &
Mehler, 2009un, 2009b; Molnar et al., 2014; Poulin-Dubois et al., 2011; Singh et al., 2014;
Tsui & Fennell, 2019).
Ainsi, a careful examination of bilingualism’s effects on cognition, especially during
infancy, is crucial for building comprehensive theories of linguistic and cognitive develop-
ment. Par exemple, demonstrating a monolingual–bilingual difference in preverbal infants
would indicate that cognitive differences can be driven by factors related to receptive lan-
guage, rather than solely by those related to productive language. Here we examined the rep-
licability of a seminal study (Kovács & Mehler, 2009un) using new data from 7-month-olds and
20-month-olds as well as by reanalyzing three open datasets from 7- to 9-month-olds.
Effects of Bilingualism on Cognition
Executive functions are mental processes that help individuals pay attention, flexibly ignore
unnecessary information, and quickly adapt to changing circumstances (Diamond, 2013).
One component of executive function—inhibitory control—has historically been the major
focus of research on monolingual–bilingual cognitive differences (for reviews see Barac &
Bialystok, 2011; Hilchey & Klein, 2011). According to early views, since bilinguals typically
only speak one language at a time, they constantly need to select representations from their
target language while inhibiting the other language, which in turn enhances domain-general
inhibitory processes (Vert, 1998; Philipp et al., 2007). Cependant, more recent accounts pro-
pose that bilingual experience serves to strengthen other aspects of executive function as well
and that a single cognitive selection mechanism may underpin the ability to use different lan-
guages in different contexts (Blanco-Elorrieta & Caramazza, 2021). This is supported by studies
finding that bilinguals outperform monolinguals in congruent trials of conflict tasks that require
inhibitory control, as well as in the incongruent trials that do not (Hilchey & Klein, 2011).
According to Bialystok (2017), attention is at the core of executive function employed in these
tasks, and bilingual experience provides the basis for the development of a more flexible system
of attention as bilinguals recurrently need to switch their attention between two languages.
Other theorists have conceptualized monolingual–bilingual differences in terms of neuroplasti-
city, while still emphasizing the impact of bilingualism on executive function (Baum & Titone,
2014). Understanding the developmental trajectory of monolingual–bilingual differences,
especially during infancy, can shed light on these and other theories.
Effects of Bilingualism on Infant Cognition
In a seminal study attempting to detect monolingual–bilingual cognitive differences at a much
younger age than had been previously demonstrated, Kovács and Mehler (2009un) compared
the performance of 7-month-old monolingual and bilingual infants in three eye-tracking
experiments. Each experiment consisted of nine training and nine test trials assessing infants’
anticipatory eye movements to visual and speech cues. Training trials began with a visual or
auditory cue, followed by a 1,000-ms anticipatory interval, and then a visual reward displayed
consistently on one side of the screen (par exemple., gauche). Infants were expected to learn that the cue
OPEN MIND: Discoveries in Cognitive Science
89
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
.
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
predicted the reward and anticipate its appearance by looking toward the trained side during
the anticipatory interval. Test trials had the same structure, except that the side of the reward
was switched (par exemple., droite). To successfully anticipate the reward at test, infants had to update
their previously learned response. Results revealed that in all experiments, monolinguals and
bilinguals performed similarly during the training phase and learned to anticipate the reward.
At Test, cependant, only bilinguals displayed an increase in correct anticipatory looks over the
course of the nine trials. The authors concluded that perceiving and processing two languages
from birth affects bilinguals’ cognitive control beyond the language domain, allowing bilinguals
to suppress previously learned responses and update predictions on the experimental task. Con-
trary to some theoretical accounts (par exemple., Vert, 1998; Philipp et al., 2007), this finding indicates
that monolingual–bilingual cognitive differences can be observed even prior to the onset of
speech production, suggesting that bilingual infants’ experience processing and building sepa-
rate mental representations of two languages is sufficient for enhancing executive function.
Given its theoretical importance, there have been several attempts to conceptually replicate
this initial finding. Four teams of researchers, testing infants from 7 et 10 months of age in
tasks similar (but not identical) to the original study, did not observe differences between
monolinguals and bilinguals (D’Souza et al., 2020; Ibánez-Lillo et al. 2010; Molnar et al.,
2014; Tsui & Fennell, 2019). Cependant, a paper that included more training and test trials
reported that bilinguals outperformed monolinguals in one of two studies, although only after
post hoc exclusion of some trials (Comishen et al., 2019). A more recent investigation reported
that at Test bilinguals performed either similarly to or better than monolinguals in one condi-
tion (depending on the analysis), but performed worse in another (see Kalashnikova et al.,
2020; revised analysis 2021, https://onlinelibrary.wiley.com/doi/10.1111/desc.13139). Là
are multiple possibilities that could explain why these replication attempts have reported dif-
fering results, including sampling error in the context of small sample sizes, methodological
differences across studies, differences between the bilingual populations tested, and different
analytic choices. Dans l'ensemble, out of six teams that have attempted conceptual replications of
Kovács and Mehler (2009un), only two have reported evidence supporting enhanced cognitive
control in bilingual infants in this paradigm.
Other tasks have also been used to examine the impacts of bilingualism on infant cognition,
and findings have been mixed. Par exemple, in a visual habituation task, 6-month-old bilin-
guals showed more efficient visual stimulus encoding and stronger recognition of stimuli than
monolinguals (Singh et al., 2014). In another series of studies, bilinguals tested between 7 et
9 months of age outperformed monolinguals in two of three cognitive flexibility tasks (D’Souza
et coll., 2020). Enhancements in memory generalization were found in 6- and 18-month-old
bilinguals (Brito & Barr, 2012; Brito et al., 2015; Brito et al., 2020), but unexpectedly not in
trilingual infants (Brito et al., 2015). Dans une autre étude, 12-month-old bilinguals were able to
associate different syllabic patterns (par exemple., AAB as in lo-lo-vu vs. ABA as in lo-vu-lo) avec
rewards on different sides of the screen, whereas monolinguals were not (Kovács & Mehler,
2009b). Executive functions were found to be enhanced in bilinguals at 24 months old on a
Shape Stroop task, although there were no monolingual–bilingual differences in any of four
other executive function tasks tested (Poulin-Dubois et al., 2011). De plus, while some of
these studies used tasks designed to tap into executive functions (Kovács & Mehler, 2009b;
Poulin-Dubois et al., 2011), others have found monolingual–bilingual differences in tasks that
are not clearly related to executive functions (Brito & Barr, 2012; Brito et al., 2015; Brito et al.,
2020; D’Souza et al., 2020). Such findings have motivated new theories to explain
monolingual–bilingual cognitive differences, Par exemple, that bilingual infants’ more variable
language environments promote greater exploratory behavior (D’Souza et al., 2020).
OPEN MIND: Discoveries in Cognitive Science
90
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
.
/
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
Dans l'ensemble, evidence has been inconsistent regarding whether and how bilingualism impacts
domain-general cognition in infancy. En outre, there have been recent concerns about
reproducibility in psychological science (Open Science Collaboration, 2015; Simmons
et coll., 2011), bilingualism research (Bolibaugh et al., 2021; de Bruin et al., 2015; Fernández-
López & Perea, 2019; Leivada et al., 2020), and infancy research (Frank et al., 2017). Given these
concerns, replicating and extending previous work on this matter is crucial.
CURRENT RESEARCH
The primary goal of the current study was to further investigate the existence and development
of a bilingual–monolingual cognitive difference in infancy. We tested monolingual and bilin-
gual infants in a simplified version of the paradigm used by Kovács and Mehler (2009un), en utilisant
a single audiovisual cue for all infants. Whereas previous studies have tested infants at a single
âge, we tested both preverbal 7-month-olds (Étude 1) and early verbal 20-month-olds (Étude 3)
to examine developmental effects. In an effort to increase statistical power and better charac-
terize infants’ performance, we employed logistic mixed-effects regression to model perfor-
mance at multiple time points per trial (DeBolt et al., 2020). This analytic approach has not
previously been used to analyze data from this anticipation paradigm, and it allowed us to
make fuller use of the fine-grained eye-tracking data to characterize infants’ moment-to-
moment performance within and across trials. This was done as a complement to the analytic
approche (ANOVAs [analysis of variance] on data averaged across three-trial blocks) used in
the original Kovács and Mehler (2009un) étude. As our data collection began prior to the pub-
lication of the replication attempts discussed in the previous section, we predicted that our
results would replicate the original study: monolinguals and bilinguals would show similar
performance during Training, but at Test bilinguals would outperform monolinguals when
updating an anticipatory response.
En plus, we conducted exploratory reanalyses (Studies 2a, 2b, and 2c) of open data
from two recently published papers (D’Souza et al., 2020, Experiment 1; Kalashnikova
et coll., 2020, 2021, Visual Condition, Auditory Condition). Tableau 1 illustrates key aspects of
participants and experimental design for each of these studies. Despite using almost identical
experimental paradigms to Kovács and Mehler (2009un) and larger sample sizes, the authors
reported conflicting results with samples of 7- to 9-month-old bilinguals and monolinguals.
D’Souza et al. (2020) reported a failure to replicate the original findings, detecting no
monolingual–bilingual differences at Test. Kalashnikova et al. (2020, 2021) reported results
in their corrigendum for the Visual Condition which they interpreted as showing no
monolingual–bilingual difference, although some analyses suggested that bilinguals may have
outperformed monolinguals at Test. The authors found a different pattern in their Auditory
Condition: monolinguals outperformed bilinguals at Test. Given the similarity of these studies
to our Study 1, we applied the same fine-grained analytical approach (logistic mixed-effects
models) to their data, aiming to capture additional effects that might have been undetected in
their original analyses.
STUDY 1: 7-MONTH-OLDS, NEW DATA
Method
This research was conducted according to the Declaration of Helsinki, and was approved by
the Human Research Ethics Board of Concordia University, Montréal (certificates UH2011-
041, 10000439). Parents gave informed consent prior to participation. Stimuli, data, et
OPEN MIND: Discoveries in Cognitive Science
91
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Tableau 1.
Comparison of Datasets Used for Analyses
Étude
1
Original
Authors
Dal Ben et al.
(2022;
this article)
Age in
mois
7
Native
languages
M.: Fr or En;
B: Fr–En or
Fr/En–other
2un
D’Souza et al.
7–9
2b
2c
(2020;
Étude 1)
Kalashnikova
et autres. (2020;
Visual
Condition)
Kalashnikova
et autres. (2020;
Auditory
Condition)
7
7
M.: En;
B: En–other
M.: Sp or Bq;
B: Sp–Bq
M.: Sp or Bq;
B: Sp–Bq
3
Dal Ben
20
et autres. (2022;
this article)
M.: Fr or En;
B: Fr–En
Unfiltered N
72 (36M., 36B)
Filtered N
43 (21M., 22B)
#
Trials/phase
9
Exemple
Cue
Results:
Training
phase
M > B
Results:
Test
phase
B > M
102 (51M., 51B)
53 (29M., 24B)
9
M > B
B > M
70 (40M., 30B)
41 (25M., 16B)
12
M > B
B > M
67 (38M., 29B)
44 (25M., 19B)
12
B > M
M > B
72 (34M., 38B)
41 (22M., 19B)
9
M = B
B > M
Note. M denotes monolinguals; B denotes bilinguals. En denotes English; Fr denotes French; Sp denotes Spanish, Bq denotes Basque; = and > denote the relative
performance of monolinguals and bilinguals.
E
F
F
e
c
t
s
o
F
B
je
je
je
n
g
toi
un
je
je
s
m
o
n
C
o
g
n
je
t
je
o
n
D
un
je
B
e
n
e
t
un
je
.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Ô
P.
E
N
M.
N
D
je
:
je
D
s
c
o
v
e
r
je
e
s
je
n
C
o
g
n
je
t
je
v
e
S
c
e
n
c
e
je
9
2
Effects of Bilingualism on Cognition Dal Ben et al.
analysis scripts are available at https://osf.io/bz8jn. Data were collected between January 2017
and July 2018.
Participants. A total of 108 7-month-old infants were tested. We filtered out 29 infants from the
main analysis because they failed to provide sufficient data, defined as gazing at the areas of
interest for at least 50% of the analyzed anticipation period, on at least five of the nine trials for
both the Training and the Test phases. This inclusion criterion was set prior to data collection
to ensure that infants were paying attention, and was based on our experience testing infants in
a variety of tasks. An additional 36 infants were excluded due to fussiness (n = 6), health issues
(n = 5), failure to meet language criteria for being classified as monolingual or bilingual (n =
18; see next paragraph for exact criteria), technical issues with the eye-tracker (n = 6), et
being older than the specified age range (n = 1).
The final sample consisted of 43 infants (Mage = 7m 11d, range: 6m 21d–8m 6d, 23 girls).
Twenty-one infants were monolingual (Mage = 7m 10d, range: 6m 23d–8m 1d, 8 girls), clas-
sified as hearing either French (n = 12) or English (n = 9) at least 90% of the time from birth.
The other 22 infants were bilingual (Mage = 7m 11d, range: 6m 21d–8m 6d, 15 girls), classified
as hearing each of two languages from 25% et 75% of the time from birth (Pearson et al.,
1993). Twelve bilinguals (55%) were learning English and French, and the other 10 étaient
learning different language pairs (see the Supplemental Materials for language backgrounds).
Participants lived in Montreal, Canada, and were recruited from government birth lists, com-
munity events, and social media. They were all healthy full-term infants (at least 37 weeks’
gestation) with no reported developmental, vision, or hearing impairments. Monolingual and
bilingual samples were predominantly of mid- to high-socioeconomic status (as estimated via
maternal educational attainment) and were comparable to each other: mothers of monolin-
guals had an average of 15.8 years of education and mothers of bilinguals had an average
de 16.3 années. There was no significant difference in estimated years of maternal education
across the two groups, t(40) = .72, p = .475, d = 0.11. Datasets and scripts are available at
https://osf.io/bz8jn.
Measures. To evaluate infants’ exposure to different languages, we used the Language Expo-
sure Questionnaire (LEQ; Bosch & Sebastián-Gallés, 2001) in conjunction with the Multilin-
gual Approach to Parent Language Estimates (MAPLE; Byers-Heinlein et al., 2019). The LEQ is
a semistructured interview that asks parents about their family language background and the
languages spoken directly to the child over the course of typical weekdays and weekends,
both at home and in other environments such as daycare. This allows the calculation of the
percentage of time infants are exposed to each language from birth. MAPLE provides guide-
lines for eliciting reliable information from families in a culturally sensitive manner.
Stimuli and Apparatus. Stimuli were developed based on those used by Kovács and Mehler
(2009un) and Reuter et al. (2018), with the goal of creating a simple and compelling nonlinguis-
tic task (all stimuli are available at https://osf.io/bz8jn). In their Visual (nonlinguistic) Condi-
tion, Kovács and Mehler presented infants with a series of shapes following an AAB or ABB
pattern, with the specific shapes changing on different trials. Reuter et al. were able to elicit
similar anticipatory behavior by cuing infants with only a looming circle accompanied by a
whistle sound, which was presented on all trials. Since we aimed for a conceptual rather than
an exact replication of Kovács and Mehler, and there was no theoretical reason why a partic-
ular cue structure would be necessary to observe monolingual–bilingual differences, we chose
to use a simple and consistent cue that might elicit more robust anticipations. Ainsi, our stimuli
were similar to those used by Reuter et al.
OPEN MIND: Discoveries in Cognitive Science
93
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
Structure of the trial sequence. Trials started with a display of a central visual fixation
Chiffre 1.
cue (a blue circle) against a black background and flanked by two white squares on the right and left
sides of the screen. The anticipatory period began after the offset of the cue, where only the white
squares were visible. At the end of the anticipatory period, a visual reward (a butterfly) appeared
inside either the left or the right square. It was always displayed inside the same square during the
Training phase (9 trials) and in the other square during the Test phase (9 trials). The analyzed antic-
ipation interval is offset from the stimulus presentation by 150 ms, to account for the time it takes
infants to initiate a gaze change.
Our cue consisted of a looming blue circle accompanied by a whistle sound, ce qui était
displayed for the first 2,000 ms of each trial. It was presented centrally on a black background
and flanked by two white squares. After the offset of the cue, a 1,000-ms anticipatory period
began, in which infants saw only the two squares. Enfin, an audiovisual reward (a spinning
butterfly accompanied by a tinkling sound) appeared inside either the left or the right square
pour 2,000 ms. Chiffre 1 depicts the structure of the trial sequence. During the Training trials, le
reward appeared consistently on one side of the screen (par exemple., in the left square), and during the
Test trials it switched sides (par exemple., in the right square). The side where the reward appeared was
counterbalanced across participants, so that infants were randomly assigned to one of two
experimental orders (par exemple., right during Training and left during Test, or left during Training
and right during Test). Trials were presented on a 24″ Tobii T60XL eye-tracker and eye gaze
data were collected at 60 Hz using Tobii Studio Software.
Procedure. During the study, infants sat on their parent’s lap on a chair in a sound-attenuated
room, environ 60 cm away from the eye-tracker. Parents wore darkened sunglasses and
headphones playing masking music, and were instructed not to talk to their child during the
étude. The experiment started with the eye-tracker calibration, using a 5-point infant calibra-
tion routine. Suivant, infants completed 9 Training and 9 Test trials. The total duration of the
experiment was approximately 90 s. Parents completed questionnaires either prior to or fol-
lowing the eye-tracking experiment. At the end of the session, parents were thanked for their
participation and children received a small gift and an honorary diploma.
Data Analysis. The main dependent variable was infants’ anticipatory eye movements. Nous
defined anticipatory eye movements as looks to either of the white squares where the reward
could appear during the 1,000-ms time window between the cue and the visual reward (fol-
lowing Kovács & Mehler, 2009un; McMurray & Aslin, 2004). Chiffre 1 shows the trial sequence
OPEN MIND: Discoveries in Cognitive Science
94
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
.
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
and highlights the anticipation period. The anticipation period occurred between 2,000 ms
et 3,000 ms after trial onset, immediately following the offset of the audio-visual cue, et
immediately prior to the onset of the visual reward. Following Kovács & Mehler (2009un), le
analyzed anticipation window was shifted to begin 150 ms after the offset of the cue and to
end 150 ms after the onset of the reward, to account for the time necessary for infants to initiate
an eye movement (Canfield et al., 1997). Ainsi, our analyzed anticipatory period was 2,150–
3,150 ms after trial onset.
Infants’ looking was measured within three areas of interest (AOIs), corresponding to the
squares on each side of the screen and the central fixation area where the blue circle
appeared, using a square approximately 2 cm larger than the visual stimuli. As mentioned
in the Participants section, prior to conducting the study, we planned both trial-level and
infant-level inclusion criteria to ensure that only infants who were on task were retained
for analysis. Although Kovács and Mehler (2009un) did not specifically mention applying such
criteria, this is a common, although variable, approach in infant research (D’Souza et al.,
2020; Kalashnikova et al., 2020), as infants who are not engaged in the task are unlikely
to provide informative data (see Byers-Heinlein et al., 2020, and ManyBabies Consortium,
2020, for a discussion of missing data in infant looking time paradigms). Given our interest
in how anticipatory behavior unfolds over time, we planned, prior to data collection, to limit
analyses to trials with at least 500 ms total looking time to any of the AOIs (gauche, center, droite)
during the 1-s analyzed anticipatory period. En outre, infants who contributed fewer than
five out of nine trials for both Training and Test phases were also excluded. To better under-
stand the effects of this strict inclusion criteria, we also conducted parallel analyses using an
unfiltered dataset, which did not impose any trial-level or infant-level exclusion criteria.
Dans l'ensemble, we found a similar pattern of results, although in some cases significance patterns
changed. Further discussion of the effect of data filtering, full models, and unfiltered data are
available in the Supplemental Materials.
We implemented two analytical approaches (datasets and scripts are available at https://osf
.io/bz8jn). The first used ANOVAs to analyze data averaged across blocks, following Kovács
and Mehler (2009un) as closely as possible. The second approach used logistic mixed-effects
models to harness the rich eye-tracking data (models were fit using the lme4 package for R;
Bates et al., 2015).
Overview of ANOVA Approach. Each trial was coded as correct (when infants looked longer to
the white square where the reward would appear than to the other square) or incorrect (quand
infants looked longer to the square where the reward would not appear, or did not look at
either square but looked only toward the central fixation AOI). Trials were grouped into
three blocks (first/middle/last) in each of the two phases. This approach analyzed a maxi-
mum of 3 blocks/phase * 2 phases = 6 data points per infant.
A correct anticipation score was calculated for each block by dividing the number of trials
with correct anticipations by the total number of valid trials. Note that in this context 50%
cannot be considered “chance,” as correct performance involves comparing looks to the target
side to looks at both the distractor side and the central fixation area combined. En tant que tel, infants
responding randomly would be expected to have “incorrect” looks on more than 50% of trials
(Par exemple, if they perseverate on the central fixation area). ANOVA tables are presented in
the Supplemental Materials, and for brevity this information is not repeated in-text.
Overview of Logistic Mixed-Effects Regression Approach. We used logistic mixed-effects models
to account for fixed and random effects arising from the repeated measures study design (par exemple.,
OPEN MIND: Discoveries in Cognitive Science
95
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
Barr, 2008; Dixon, 2008; Humphrey & Swingley, 2018). Our analysis allowed us to investigate
how anticipatory behavior unfolded within each learning instance (within trials) and how it
accumulated across the experiment (across trials)—a major improvement over block-level
analyses. Dans cette approche, rather than categorizing each trial as correct or incorrect overall,
we calculated the weighted proportion of looks to the correct side at each of five 200-ms time
bins per trial.1 We omitted the first trial of each phase, since at that point infants did not have
information with which to make a correct anticipation. This approach analyzed up to 5 temps
bins/trial * 8 trials/phase * 2 phases = 80 data points per infant. En tant que tel, it provided a richer
characterization of infants’ performance over time and greater statistical power (DeBolt et al.,
2020).
Our models tested the effects of language group, time bin, trial number, and their interac-
tions on the proportion of looking to the correct side during each phase (Training and Test)
separately. We aimed for a consistent random effects structure across models to facilitate com-
parisons. Our final models included only the random intercept for participants, as more com-
plex models that included random slopes for trial number and time bin did not converge for all
models. As mentioned, the first trial from both phases was excluded prior to fitting the models.
Trial number was then scaled so that the reference trial would be the first analyzed trial (trial
number − 2) and time bin was centered so that the middle time bin would be the reference
time bin, as by then many infants might have made an anticipatory look (time bin − 2). Mono-
linguals were the reference language group. Ainsi, the model intercept describes the odds of
making a correct anticipation for monolinguals at the middle time bin of Trial 2.
We report our results in odds ratios. Changes in the relative odds of making a correct antic-
ipation as a function of language group, time bin, and trial number indicate how learning
unfolded across the experiment. Spécifiquement, the effect of time bin indexes infants’ anticipa-
tions within trials. The effect of trial number indexes learning across trials. The interaction
between time bin and trial number indexes the speed of making correct anticipations across
trials. Surtout, our models also included main effects and interactions of these terms
with language group to capture different patterns of performance between monolinguals
and bilinguals. For brevity, estimates and statistics for all studies are presented in Table 2
(Training) and Table 3 (Test), and are not repeated in the text. Model visualizations are
presented in the Supplemental Materials.
Results
ANOVA. One participant was excluded from the ANOVAs due to missing data. During Train-
ing, un 2 (language group) × 3 (block) mixed ANOVA showed no significant interaction, nor
significant main effects of language or block (Chiffre 2; ANOVA tables available in the Supple-
mental Materials). The latter was surprising because performance should improve from the first
to the last block as learning unfolds. At Test, there was a significant main effect of block, dem-
onstrating improved performance over the course of the Test phase, but no main effect of lan-
guage group or interaction of block with language group. Dans l'ensemble, under the ANOVA
approach we did not detect any monolingual–bilingual differences, and thus found no support
for our prediction that bilinguals would outperform monolinguals at Test.
1 Weighted proportion looking to the correct side was the dependent variable. The proportion of looking to
the correct side (ranging from 0 à 1) was calculated for each time bin by averaging the values of all samples
(binary, either 0 ou 1) in a given bin. This resulted in a quasi-binomial distribution and the number of samples
was used to weight the proportions between 0 et 1 (see analysis scripts at https://osf.io/bz8jn).
OPEN MIND: Discoveries in Cognitive Science
96
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
.
/
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Fixed Effects as Odds Ratios From the Final Model [looking proportion ∼ language group * time bin * trial number + (1 | participant)] for the Training Phase of
Ô
P.
E
N
M.
N
D
je
:
je
D
s
c
o
v
e
r
je
e
s
je
n
C
o
g
n
je
t
je
v
e
S
c
e
n
c
e
je
Tableau 2.
All Studies
Prédicteurs
(Intercept)
langue [bilingual]
time bin
Interpretation
The reference odds of
monolinguals correctly
anticipating during the
middle time bin during
Trial 2
The relative odds
(compared to
monolinguals) de
bilinguals correctly
anticipating during
the middle time bin
during Trial 2
The change in odds of
monolinguals correctly
anticipating during Trial
2 for each unit change
in time bin
trial number
The change in odds of
monolinguals correctly
anticipating during the
middle time bin for
each unit change in
trial number
The relative change in
odds of bilinguals
correctly anticipating
during Trial 2 with each
unit change in time bin
The relative change in
odds of bilinguals
correctly anticipating
during the middle time
bin with each unit
change in trial number
langue [bilingual] *
time bin
langue [bilingual] *
trial number
9
7
Étude 1
Odds Ratios
[CI] p
0.08
[0.02–0.25]
***
Study 2a
Odds Ratios
[CI] p
0.09
[0.04–0.19]
***
Study 2b
Odds Ratios
[CI] p
0.06
[0.03–0.13]
***
Study 2c
Odds Ratios
[CI] p
0.23
[0.14–0.38]
***
Étude 3
Odds Ratios
[CI] p
1.09
[0.65–1.82]
ns
0.39
[0.07–2.02]
ns
0.73
[0.25–2.18]
ns
0.61
[0.19–2.01]
ns
0.59
[0.27–1.28]
ns
0.84
[0.40–1.80]
ns
6.47
[5.67–7.40]
***
2.06
[1.94–2.19]
***
2.67
[2.52–2.83]
***
1.43
[1.37–1.50]
***
3.15
[2.88–3.44]
***
1.12
[1.08–1.16]
***
1.14
[1.12–1.16]
***
1.16
[1.15–1.18]
***
1.05
[1.04–1.06]
***
0.88
[0.86–0.91]
***
0.43
[0.36–0.51]
***
0.91
[0.83–0.99]
***
0.83
[0.75–0.92]
***
1.51
[1.40–1.63]
***
0.83
[0.74–0.94]
**
1.04
[0.99–1.10]
ns
1.14
[1.11–1.17]
***
0.96
[0.94–0.98]
***
1.19
[1.17–1.21]
***
1.07
[1.03–1.11]
**
E
F
F
e
c
t
s
o
F
B
je
je
je
n
g
toi
un
je
je
s
m
o
n
C
o
g
n
je
t
je
o
n
D
un
je
B
e
n
e
t
un
je
.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
.
/
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Interpretation
The change in odds of
monolinguals correctly
anticipating with each
unit change in both time
bin and trial number
ensemble (difference in
odds between earlier
time bins in earlier
trials and later time bins
in later trials)
The relative change in
odds of bilinguals
correctly anticipating
with each unit change
in both time bin and
trial number together
Prédicteurs
time bin *
trial number
langue [bilingual] *
time bin *
trial number
Random Effects
N
Observations
Tableau 2.
(a continué )
Étude 1
Odds Ratios
[CI] p
0.91
[0.89–0.94]
***
Study 2a
Odds Ratios
[CI] p
0.96
[0.95–0.97]
***
Study 2b
Odds Ratios
[CI] p
0.94
[0.93–0.95]
***
Study 2c
Odds Ratios
[CI] p
1.06
[1.05–1.07]
***
Étude 3
Odds Ratios
[CI] p
1.01
[0.99–1.03]
ns
1.11
[1.07–1.15]
***
1.01
[1.00–1.03]
ns
1.04
[1.02–1.05]
***
0.92
[0.91–0.93]
***
0.97
[0.94–1.00]
ns
43id
1,436
53id
1,734
41id
1,908
44id
2,122
41id
1,370
Marginal R2 / Conditional R2
0.267 / 0.772
0.106 / 0.594
0.160 / 0.593
0.178 / 0.456
0.325 / 0.532
Note. Tables 1 et 2 present the odds ratio estimates from logistic mixed-effects models. Odds ratios greater than 1 mean an outcome is more likely to happen; odds
ratios less than 1 mean an outcome is less likely to happen; odds ratios of exactly 1 mean an outcome is as likely to happen as not (chance). Odds ratios are expo-
nentiated from the logit, which renders effects that would be additive in the direct model output multiplicative in these tables. When interpreting the effects for inter-
action terms in the tables, the displayed odds ratio must be multiplied with the reference odds (or reference change in odds) to arrive at a final effect size.
ns = not significant, p > .05. *p ≤ .05. **p ≤ .01. ***p ≤ .001.
E
F
F
e
c
t
s
o
F
B
je
je
je
n
g
toi
un
je
je
s
m
o
n
C
o
g
n
je
t
je
o
n
D
un
je
B
e
n
e
t
un
je
.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Ô
P.
E
N
M.
N
D
je
:
je
D
s
c
o
v
e
r
je
e
s
je
n
C
o
g
n
je
t
je
v
e
S
c
e
n
c
e
je
9
8
Fixed Effects as Odds Ratios From the Final Model [looking proportion ∼ language group * time bin * trial number + (1 | participant)] for the Test Phase of All
Ô
P.
E
N
M.
N
D
je
:
je
D
s
c
o
v
e
r
je
e
s
je
n
C
o
g
n
je
t
je
v
e
S
c
e
n
c
e
je
Tableau 3.
Études
Prédicteurs
[Intercept]
langue [bilingual]
time bin
Interpretation
The reference odds of
monolinguals correctly
anticipating during the
middle time bin during
Trial 2
The relative odds
[compared to
monolinguals] de
bilinguals correctly
anticipating during
the middle time bin
during Trial 2
The change in odds of
monolinguals correctly
anticipating during
Trial 2 for each unit
change in time bin
trial number
The change in odds of
monolinguals correctly
anticipating during the
middle time bin for
each unit change in
trial number
The relative change in
odds of bilinguals
correctly anticipating
during Trial 2 with each
unit change in time bin
The relative change in
odds of bilinguals
correctly anticipating
during the middle time
bin with each unit
change in trial number
langue [bilingual] *
time bin
langue [bilingual] *
trial number
9
9
Étude 1
Odds Ratios
[CI] p
0.01
[0.00–0.03]
***
Study 2a
Odds Ratios
[CI] p
0.04
[0.02–0.11]
***
Study 2b
Odds Ratios
[CI] p
0.06
[0.04–0.11]
***
Study 2c
Odds Ratios
[CI] p
0.19
[0.10–0.34]
***
Étude 3
Odds Ratios
[CI] p
0.49
[0.32–0.76]
**
2.02
[0.23–17.42]
ns
1.59
[0.43–5.86]
ns
0.74
[0.31–1.78]
ns
0.31
[0.12–0.78]
*
2.53
[1.33–4.81]
**
1.26
[1.11–1.42]
***
0.81
[0.77–0.86]
***
1.65
[1.55–1.75]
***
2.47
[2.34–2.60]
***
1.67
[1.54–1.80]
***
1.05
[1.00–1.10]
*
0.94
[0.92–0.96]
***
1.12
[1.10–1.13]
***
1.09
[1.08–1.11]
***
0.98
[0.95–1.01]
ns
2.47
[2.05–2.98]
***
1.93
[1.78–2.09]
***
1.28
[1.15–1.43]
***
0.61
[0.56–0.67]
***
1.09
[0.98–1.22]
ns
1.12
[1.05–1.19]
***
1.32
[1.28–1.36]
***
1.14
[1.11–1.17]
***
1.07
[1.05–1.09]
***
0.88
[0.85–0.92]
***
E
F
F
e
c
t
s
o
F
B
je
je
je
n
g
toi
un
je
je
s
m
o
n
C
o
g
n
je
t
je
o
n
D
un
je
B
e
n
e
t
un
je
.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
.
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Ô
P.
E
N
M.
N
D
je
:
je
D
s
c
o
v
e
r
je
e
s
je
n
C
o
g
n
je
t
je
v
e
S
c
e
n
c
e
je
Interpretation
The change in odds of
monolinguals correctly
anticipating with each
unit change in both
time bin and trial
number together
[difference in odds
between earlier time bins
in earlier trials and later
time bins in later trials]
The relative change in
odds of bilinguals
correctly anticipating
with each unit change
in both time bin and
trial number together
Prédicteurs
time bin *
trial number
[langue [bilingual] *
time bin] *
trial number
Random Effects
N
Observations
Tableau 3.
(a continué )
Étude 1
Odds Ratios
[CI] p
1.05
[1.02–1.09]
***
Study 2a
Odds Ratios
[CI] p
1.12
[1.11–1.14]
***
Study 2b
Odds Ratios
[CI] p
0.99
[0.98–1.00]
**
Study 2c
Odds Ratios
[CI] p
0.95
[0.94–0.95]
***
Étude 3
Odds Ratios
[CI] p
1.04
[1.02–1.06]
***
0.95
[0.91–1.00]
*
0.90
[0.89–0.92]
***
0.99
[0.98–1.01]
ns
1.09
[1.07–1.10]
***
1.01
[0.98–1.04]
ns
43id
1383
53id
1676
41id
1672
44id
1846
41id
1272
Marginal R2 / Conditional R2
0.108 / 0.807
0.095 / 0.666
0.139 / 0.451
0.161 / 0.510
0.202 / 0.393
Note. ns = not significant, p > .05. *p ≤ .05. **p ≤ .01. ***p ≤ .001.
E
F
F
e
c
t
s
o
F
B
je
je
je
n
g
toi
un
je
je
s
m
o
n
C
o
g
n
je
t
je
o
n
D
un
je
B
e
n
e
t
un
je
.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
.
/
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
1
0
0
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
Chiffre 2. Correct anticipation by trial for Study 1 (âge 7 mois). Visualization of Study 1 (7 mois) under Analytic Approach 1
(ANOVAs), similar to Figure 2 from Kovács and Mehler (2009un). Symbols represent the proportion of infants with correct anticipatory
looks, with monolinguals plotted in red (circles and solid lines) and bilinguals plotted in blue (triangles and dashed lines). Lines show
best linear fit for each group.
Logistic Mixed-Effects Regression. During Training (Tableau 2; Chiffre 3), we unexpectedly found
that 7-month-old monolinguals outperformed bilinguals, evidenced by more robust anticipa-
tions both within trials and across trials. Both groups showed evidence of learning across
trials, and interaction effects indicated that bilinguals approached similar levels of perfor-
mance as monolinguals by the end of Training (Trial 9). Inversement, at Test (Tableau 3;
Chiffre 3), bilinguals showed much stronger evidence of correct anticipations, increasing
their looking to the target more than monolinguals both within trials and across trials, avec
a sharper increase in later trials. This indicated that bilinguals were better at updating their
previously learned association.
As predicted, our main analyses showed that 7-month-old bilinguals were better than
monolinguals at updating their old behavior with a new one. Unexpectedly, our results suggest
that these monolingual–bilingual differences might be due to different learning trajectories dur-
ing the Training phase. Monolinguals initially learned to anticipate the reward faster than bilin-
guals (although bilinguals caught up by Trial 9), but struggled to update their behavior when
the environment changed at Test. The reverse pattern was found for bilinguals. Surtout, pre-
vious studies have only found a difference at Test (par exemple., Kalashnikova et al., 2020, 2021;
Kovács & Mehler, 2009un). Ainsi, our finding that performance also differs during training sug-
gests that the impacts of bilingualism on domain-general cognition in infancy are more com-
plex than previously observed, in that it might involve both the ability to initially learn new
information and to update existing knowledge.
STUDY 2: 7- TO 9-MONTH-OLDS, REANALYZED DATA
Recently, close replications of Kovács and Mehler (2009un) were published by D’Souza et al.
(2020) and Kalashnikova et al. (2020, 2021). Both studies tested whether 7- to 9-month-old
bilinguals would be better at inhibiting a learned anticipation compared to matched monolin-
guals. D’Souza et al., despite using a near-identical task, a more sensitive trial-level analysis,
and a larger sample than the original study, found no differences between groups’ anticipatory
looks. In a corrected analysis (2021) of their original data (2020), Kalashnikova et al., on the
other hand, reported that Basque–Spanish bilinguals performed similarly to monolinguals
when tested using visual cues, although some analyses suggested that bilinguals showed better
performance at Test. Depending on interpretation, this could be seen as replicating Kovács and
OPEN MIND: Discoveries in Cognitive Science
101
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Chiffre 3. Time course of infant looking for Study 1 (âge 7 mois). Time course of proportion looking for Study 1 (7 mois). Red indicates
monolingual infants and blue indicates bilingual infants. Solid lines indicate looks to the correct side. Dashed lines indicate looks to the
incorrect side. Dotted lines indicate looks to the location where the central fixation cue had appeared. Yellow backgrounds indicate the ana-
lyzed anticipation period time frame (2,150–3,150 ms after trial onset) used for our analyses.
Mehler’s original findings. Cependant, when testing the same infants using auditory cues (en utilisant
a within-subjects design), Kalashnikova et al. found that monolinguals outperformed bilinguals
during the Test phase, thus failing to replicate the original study.
Given the close similarities in experimental design between our Study 1, D’Souza et al.’s
(2020) Experiment 1, and Kalashnikova et al.’s (2020) Visual and Auditory Conditions, we con-
ducted a reanalysis of their data using the same logistic mixed-effects regression we used with
our own data, to see if there were additional effects that might have been undetected in their
original analyses. This was possible thanks to the authors’ engagement with open science
practices—their data were openly available and they were responsive when contacted for
additional information.
It is worth noting that although all three of these studies were similar in design to Kovács
and Mehler (2009un), the authors used different inclusion criteria for infant attention. D’Souza
et autres. (2020) excluded infants who did not have valid data on at least 75% of trials, alors que
Kalashnikova et al. (2020, 2021) excluded infants who had more than 40% gaze loss for
OPEN MIND: Discoveries in Cognitive Science
102
Effects of Bilingualism on Cognition Dal Ben et al.
the entire task. To maximize comparability between results, we applied our stricter filtering
criteria (c'est à dire., at least 50% looking during the anticipation period in more than half of trials
in both phases) to their data before reanalyzing it. This filtering resulted in smaller sample sizes
than reported in the original experiments. To ensure that our results were not due to our stricter
filtering and differences in sample size, we also ran the models on the unfiltered datasets. UN
comparison between the filtered and unfiltered samples is available in the Supplemental Mate-
rials, and an overall summary appears in Table 4.
Tableau 4. Differences Between Filtered and Unfiltered Samples in Our Analyses
Étude
1
n
(Unfiltered)
72 (36M.,
36B)
n
(Filtered)
43 (21M.,
22B)
Sample
reduction Analysis
Phase
40%
ANOVA Training
Différence
yes
Details
The unfiltered sample, but not the filtered sample,
showed a group difference with monolinguals
outperforming bilinguals.
2un
2b
2c
3
102 (51M.,
51B)
53 (29M.,
24B)
70 (40M.,
30B)
41 (25M.,
16B)
Test
LMEM
Training
Test
48%
LMEM
Training
Test
41%
LMEM
Training
Non
Non
Non
Non
Non
Non
Test
yes
In the unfiltered sample only, the odds of
bilinguals correctly anticipating increased less
over the course of the baseline trial than the
odds of monolinguals, but this difference was
mediated to some extent by a significant three-
way bilingual–trial number–time bin
interaction. Ainsi, the odds of bilinguals
correctly anticipating in later time bins
increased as trial number increased from the
baseline trial.
67 (38M.,
29B)
44 (25M.,
19B)
72 (34M.,
38B)
41 (22M.,
19B)
34%
LMEM
Training
Test
43%
ANOVA Training
Test
LMEM
Training
Non
Non
Non
yes
yes
At Test, the effect of language group was no
longer significant in the unfiltered sample.
In the unfiltered sample, the trial number–time
bin interaction changed from nonsignificant to
decreasing odds for monolinguals.
Test
yes
Results from the unfiltered dataset were
somewhat different: the only significant effect
related to language group was its three-way
interaction with trial number and time bin.
Note. M denotes monolinguals; B denotes bilinguals. LMEM = logistic mixed-effects model; ANOVA = analysis of variance.
OPEN MIND: Discoveries in Cognitive Science
103
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
STUDY 2A: D’SOUZA ET AL. (2020) REANALYSIS
D’Souza et al.’s (2020) experimental design was very similar to our Study 1. Training and Test
phases had 9 trials each. Following Kovács and Mehler (2009un, Experiment 3), trials started
with an attention-getter stimulus (displayed for 500 ms), followed by a sequence of silent AAB
or ABB visual cues (3,000 ms), an anticipation period (1,000 ms), and a reward (left or right
side; 2,000 ms). Bilingual (n = 51) and monolingual (n = 51) infants were closely matched on
âge, genre, and parental socioeconomic status, and were counterbalanced across visual cue
sequences and reward sides. To reanalyze D’Souza et al. (2020), we applied the inclusion
criteria for attentiveness used in Study 1, which left 29 monolinguals and 24 bilinguals from
the original sample. Both groups were still comparable in age, genre, and parental socioeco-
nomic status.
Using our logistic mixed-effects model approach, we found a very similar pattern to Study 1
(Chiffre 4). During Training, monolinguals outperformed bilinguals (Tableau 2, see details in the
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Chiffre 4. Time course of infant looking for Study 2a (age 7–9 months). Time course of proportion looking for Study 2a (reanalysis of D’Souza
et coll., 2020). Red indicates monolingual infants and blue indicates bilingual infants. Solid lines indicate looks to the correct side. Dashed lines
indicate looks to the incorrect side. Dotted lines indicate looks to the location where the central fixation cue had appeared. Yellow back-
grounds indicate the analyzed anticipation period time frame (3,150–4,150 ms after trial onset) used for our analyses.
OPEN MIND: Discoveries in Cognitive Science
104
Effects of Bilingualism on Cognition Dal Ben et al.
Supplemental Materials). Spécifiquement, bilinguals struggled more than monolinguals with cor-
rectly anticipating during initial Training trials, but their performance did catch up to mono-
linguals by Trial 9, a finding also reported by D’Souza et al. (2020) in their original analysis. Dans
the Test phase, bilinguals outperformed monolinguals across both time bin and trial number
(Tableau 3). Significant interactions indicated that only bilinguals showed improvement over the
course of the trials, while monolinguals’ performance actually decreased, suggesting that only
bilinguals were able to learn the new association.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
.
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Chiffre 5. Time course of infant looking for Study 2b (âge 7 mois; Visual Condition). Time course of proportion looking for Study 2b,
Visual Condition (reanalysis of Kalashnikova et al., 2020). Red indicates monolingual infants and blue indicates bilingual infants. Solid lines
indicate looks to the correct side. Dashed lines indicate looks to the incorrect side. Dotted lines indicate looks to the location where the central
fixation cue had appeared. Yellow backgrounds indicate the analyzed anticipation period time frame (3,650–4,650 ms after trial onset) used for
our analyses.
OPEN MIND: Discoveries in Cognitive Science
105
Effects of Bilingualism on Cognition Dal Ben et al.
These results reinforce the findings from Study 1. We observed further evidence for
monolingual–bilingual differences in establishing and updating a learned behavior, pointing
to the importance of the initial learning trajectories during the Training phase. Our fine-grained
analytic approach captured additional effects that were undetected in the original analysis,
which was performed with trial-level means.
STUDY 2B: KALASHNIKOVA ET AL. (2020) VISUAL CONDITION
The experimental design of Kalashnikova and colleagues’ (2020, 2021) Visual Condition was
very similar to Kovács and Mehler (2009un, Experiment 3). In both studies, infants were cued
with a series of three geometric shapes presented silently in an AAB or ABB pattern before
a reward appeared consistently on one side of the screen during Training, switching sides at
Test. Kalashnikova et al. (2020, 2021) ran 12 trials in each experimental phase and reported
block-level analyses. Each block summarized four trials and each phase included three blocks.
After applying our participant-level filtering criteria, we retained 41 infants (monolingual = 25,
bilingual = 16) from the original sample of 70. We then used our logistic mixed-effects models
to analyze the data.
During Training (Tableau 2; Chiffre 5), monolinguals performed slightly better than bilinguals
within trials, an effect not detected by the original block-level analysis. Monolinguals also
outperformed bilinguals when measured across trials. Cependant, the difference between
monolingual and bilingual performance here was much weaker than the pattern found during
Training in Studies 1 and 2a, despite the procedural similarity between studies. At Test (Tableau 3;
Chiffre 5), bilinguals were faster than monolinguals in updating their previously learned antic-
ipation both within and across trials.
In sum, our reanalysis of Study 2b aligns with the findings from Studies 1 and 2a. Although
the effects were less pronounced, they show that monolinguals in this study initially learned to
anticipate more quickly than bilinguals, but bilinguals were better at updating their responses
when the environment changed.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
.
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
STUDY 2C: KALASHNIKOVA ET AL. (2020) AUDITORY CONDITION
Dans cette étude, we reanalyzed data from Kalashnikova et al.’s (2020) Auditory Condition.
Following Kovács and Mehler’s (2009un, Experiment 1) conception, on each trial, infants heard
an auditory cue consisting of a different sequence of three syllables in an AAB or ABB pattern
that had identical phonetic realizations in Spanish and Basque (the bilinguals’ languages).
After applying our filtering criteria for infant attention, we analyzed data from 25 monolinguals
et 19 bilinguals.
Using our logistic mixed-effects model approach, we found an opposite trend to the one
found in the Training phase of Studies 1, 2un, and 2b (Tableau 2; Chiffre 6). Bilinguals outper-
formed monolinguals within trials and across trials.2 However, monolinguals’ odds increased
in later time bins of later trials, while bilinguals’ odds decreased slightly. The significant
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
2 We experienced convergence issues with the Study 2c Training phase model. Cependant, the model con-
verged when the trial number variable was rescaled (trial number/2), resulting in very similar direction and
magnitude for all estimates. To be consistent, here we report the unscaled model, and interested readers can
see the scaled model results in the analysis code scripts.
OPEN MIND: Discoveries in Cognitive Science
106
Effects of Bilingualism on Cognition Dal Ben et al.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
.
/
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Chiffre 6. Time course of infant looking for Study 2c (âge 7 mois; Auditory Condition). Time course of proportion looking for Study 2c
(reanalysis of Kalashnikova et al., 2020, Auditory Condition). Red indicates monolingual infants and blue indicates bilingual infants. Solid lines
indicate looks to the correct side. Dashed lines indicate looks to the incorrect side. Dotted lines indicate looks to the location where the central
fixation cue had appeared. Yellow backgrounds indicate the analyzed anticipation period time frame (2,350–3,350 ms after trial onset) used for
our analyses.
interaction term indicated that by the end of Training, monolinguals’ performance approached
bilinguals’.
Also contrary to what was found in Studies 1, 2un, and 2b, but consistent with what was
reported by Kalashnikova et al. (2020), monolinguals outperformed bilinguals at Test (Tableau 3;
Chiffre 6). Monolinguals were substantially faster in updating their previously learned antici-
pation within trials. Both monolinguals and bilinguals improved across trials, although the
effect was slightly stronger for bilinguals. The significant interaction term showed that
OPEN MIND: Discoveries in Cognitive Science
107
Effects of Bilingualism on Cognition Dal Ben et al.
bilinguals approached monolinguals’ performance by Trial 9. Dans l'ensemble, monolinguals were
more successful at updating their previously learned anticipation, although bilinguals did
show learning over the course of the Test phase.
In sum, the pattern of results in Kalashnikova et al.’s (2020) Auditory Condition was oppo-
site to the pattern found in the Visual Condition (Study 2b): bilinguals showed faster and more
robust learning during Training, but monolinguals were better at updating their response at
Test. Ensemble, the results from Studies 1 et 2 highlight a trade-off between performance in
the two phases: groups of infants who learn faster in the initial Training phase are not as able to
update their anticipation in the subsequent Test phase, and vice versa. This is especially clear
in Studies 2b and 2c, which used a within-subjects design. In the general discussion, we elab-
orate on the common patterns found in Study 1 et 2 (see Table 1 for a summary). In Study 3,
we examined developmental effects on the bilingual–monolingual differences we found in
Études 1 et 2, by testing older infants on the same task.
STUDY 3: 20-MONTH-OLDS, NEW DATA
The goal of this study was to investigate developmental effects on cognition in monolinguals
and bilinguals, by testing 20-month-old infants using the same methods as in Study 1. As in the
previous studies, we explored the effect of our filtering criteria (reported in the Supplemental
Materials and summarized in Table 4). Dans l'ensemble, the pattern of results was weaker with the unfil-
tered dataset than with the filtered dataset used for the main analyses (c'est à dire., fewer effects
reached statistical significance), a point we return to in the discussion.
Method
Participants. A total of 119 infants were tested. We filtered out 31 infants from the main anal-
ysis because they failed to provide sufficient data for both the Training and the Test phases, comme
defined in Study 1 based on our predetermined criteria. An additional 47 infants were
excluded due to fussiness (n = 6), health issues (n = 2), failure to meet language criteria for
being classified as monolingual or bilingual as described in Study 1 (n = 35), expérimental
error (n = 1), and parental interference (n = 3). All infants were born full-term (at least 37 weeks’
gestation). Data were collected from January 2017 and March 2019. Participants lived in
Montréal, Canada, and were recruited from government birth lists, community events, et
social media.
The final sample included 41 infants (Mage = 20m 14d, range: 19m 23d–21m 6d, 21 girls).
Twenty-two were monolinguals (Mage = 20m 15d, range: 19m 24d–21m 6d, 15 girls), exposed
at least 90% of the time to either French (n = 15) or English (n = 7). Nineteen were bilinguals
(Mage = 20m 13, range: 19m 23d–21m 5d, girls = 6), exposed to two languages at least 25% de
the time each (English and French, n = 16; other language pairs, n = 3), hearing both languages
from birth (see the Supplemental Materials for language backgrounds). The same criteria were
used for monolingualism and bilingualism as in Study 1. Monolingual and bilingual samples
both came from mid- to high-socioeconomic-status families. En moyenne, mothers of mono-
lingual infants had 15.5 years of education compared to mothers of bilinguals who had
17.3 années, a statistically significant difference (t(39) = −2.87, p = .007, d = −0.45),
although we note that both groups of mothers were highly educated, given that most
had completed a bachelor’s or advanced degree (par exemple., MA or PhD). Detailed measures
are reported in the Supplemental Materials.
In addition to measuring infants’ language exposure as in Study 1, parents
Measures.
completed a measure of infants’ productive vocabulary, using the MacArthur-Bates
OPEN MIND: Discoveries in Cognitive Science
108
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
.
/
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
Communicative Developmental Inventory (MCDI; Fenson et al., 1993) and/or its adaptation in
Québec French (Trudeau et al., 1999), depending on which language(s) their child was learn-
ing. Word vocabulary (the sum of unique word forms produced in any language; González-
Barrero et al., 2020; Hoff et al., 2012) was calculated for monolinguals and French–English
bilinguals. Word vocabulary was not calculated for 3 bilinguals learning other language pairs,
as the MCDI was only available in one of their languages. Infants with missing vocabulary data
were excluded from a preliminary model that tested for effects of vocabulary size (see the Sup-
plemental Materials), but were included in the main model.
Stimuli, Apparatus, and Procedure. The same stimuli, apparatus, and procedure as in Study 1
were used.
Results
Preliminary analyses indicated no effects or interactions with vocabulary size (details in the
Supplemental Materials), so we applied the same analytic strategies as in Study 1.
ANOVA. Five participants were excluded from ANOVA analyses due to missing data. During
Training (Chiffre 7), performance increased significantly from Block 1 to Block 2 before falling
for Block 3, which could indicate learning over the course of the trials. There were no other
significant main effects or interactions. During Test, there were significant main effects of both
block and language group, indicating that both language groups learned the association, mais
that bilinguals’ performance was better overall. Ainsi, some monolingual–bilingual differences
seen at 7 months of age seem to persist at 20 months of age.
Logistic Mixed-Effects Regression. Dans l'ensemble, 20-month-old infants showed fast learning during
Training (Tableau 2; Chiffre 8). The model intercept was substantially higher than the intercepts
found in Studies 1 et 2, meaning that during the middle time bin of the reference trial (Trial 2),
20-month-old monolinguals made more correct anticipations than 7- to 9-month-olds. Once
again, the model odds ratios suggest that monolinguals learned more quickly than bilinguals
during Training. Unexpectedly, across trials, the odds of correctly anticipating decreased for
both groups, indicating that at 20 months old, infants learned the association quickly and then
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Chiffre 7. Correct anticipation by trial for Study 3 (âge 20 mois). Visualization of Study 3 (20 mois) under Analytic Approach 1 (ANOVAs),
similar to Figure 2 from Kovács and Mehler (2009un). Symbols represent the proportion of infants with correct anticipatory looks, avec
monolinguals plotted in red (circles and solid lies) and bilinguals plotted in blue (triangles and dashed lines). For consistency, lines show
best linear fit for each group, although data do not appear to follow a linear trend.
OPEN MIND: Discoveries in Cognitive Science
109
Effects of Bilingualism on Cognition Dal Ben et al.
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
.
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Chiffre 8. Time course of infant looking for Study 3 (20 mois). Red indicates monolingual infants and blue indicates bilingual infants. Solid
lines indicate looks to the correct side. Dashed lines indicate looks to the incorrect side. Dotted lines indicate looks to the location where the
central fixation cue had appeared. Yellow backgrounds indicate the analyzed anticipation period time frame (2,150–3,150 ms after trial onset)
used for our analyses.
possibly grew bored of the task. At Test (Tableau 3; Chiffre 8), bilinguals performed better than
monolinguals on average, and the two groups showed similar improvements within trials.
Bilinguals, cependant, had lower odds of correctly anticipating across trials, again, possibly
indicating boredom with the task.
In sum, monolinguals were slightly better at correctly anticipating during the Training phase
and bilinguals were slightly faster than monolinguals in updating their previously learned
anticipation at Test. This suggests that the effects of bilingualism on cognition observed in
younger infants (7- to 9-month-olds) are also observable at an older age (20 mois), bien que
they may be less pronounced.
GENERAL DISCUSSION
This research sought further evidence regarding the effects of bilingualism on infant cognition.
To this end, we used an anticipatory eye movement paradigm adapted from Kovács and
Mehler (2009un): trials began with a central visual cue, progressed to an anticipatory period,
OPEN MIND: Discoveries in Cognitive Science
110
Effects of Bilingualism on Cognition Dal Ben et al.
and ended with a visual reward displayed consistently on one side of the screen. At Test, le
reward switched to the other side. We collected data from preverbal (7-month-olds; Étude 1)
and early verbal (20-month-olds; Étude 3) bilingual and monolingual infants, which allowed
us to examine the impacts of bilingualism at two points in infant development. We also rea-
nalyzed data from three similar studies with 7- to 9-month-old infants for direct comparison
of our statistically powerful analytic approach using logistic mixed-effects models (c'est à dire.,
D’Souza et al., 2020, Experiment 1, which we present in Study 2a; Kalashnikova et al.,
2020, Visual Condition, which we present in Study 2b; Kalashnikova et al., 2020, Auditory
Condition, which we present in Study 2c).
Confirming our predictions, in four out of the five datasets we analyzed (Études 1, 2un, 2b,
et 3), 7- à 9- and 20-month-old bilinguals were faster and more accurate than monolinguals
at updating their anticipatory behavior at Test. Unexpectedly, we found that in three of these
same studies, 7- to 9-month-old bilinguals were slower than monolinguals to learn the initial
anticipatory response during Training. For 20-month-olds, we saw similar but less pronounced
monolingual–bilingual differences during the Training phase. Dans l'ensemble, our results provide
compelling evidence that bilingualism affects infant cognitive development, both at younger
(7–9 months) and older (20 mois) ages. These studies both replicate and extend Kovács
and Mehler’s (2009un) étude, as we show that this finding is generalizable across methodo-
logical variations in the basic paradigm. Par exemple, whereas D’Souza et al. (2020) et
Kalashnikova et al.’s (2020) Visual Condition cued infants with a series of static geometric
shapes in AAB or ABB patterns presented in silence, in our studies infants were cued with a
looming circle paired with a whistle sound. Plus loin, the same pattern of results generalized
across diverse bilingual samples, who were learning a variety of language pairs, exposed to
a varying prevalence of bilingualism in their communities, and tested in different labs.
Contrary to our predictions, we found the reverse pattern of results in one dataset:
Kalashnikova et al.’s (2020) Auditory Condition (Study 2c). Dans cette étude, bilinguals outper-
formed monolinguals during training, but monolinguals outperformed bilinguals at Test.
How can we explain this result? Kalashnikova et al. proposed that individual patterns of
language exposure—for example, whether infants’ exposure is balanced and whether they
come from a bilingual community—could contribute to performance differences in this task.
Cependant, if that were the case, we would expect similar performance in the Auditory and
Visual Conditions, as the same infants were tested on both conditions in a within-subjects
conception. Another possible explanation pertains to the nature of the Auditory versus Visual
Conditions. Perhaps for linguistic stimuli, which were used in Kalashnikova et al.’s Auditory
Condition but not in the other studies analyzed here, bilinguals are faster than monolinguals at
initially encoding the information, whereas for nonlinguistic stimuli they are slower (see also
Hilchey & Klein, 2011, for a discussion of monolingual–bilingual differences on cognitive
tasks with linguistic vs. nonlinguistic stimuli). Cependant, Kalashnikova et al.’s Auditory Condi-
tion was a close replication of Kovács and Mehler’s (2009un) Experiment 2, where the opposite,
more prevalent pattern was found: bilinguals outperformed monolinguals at Test (bien que
contrary to what we observed, there were no reported differences during Training). While
the results of the Auditory Condition raise the possibility that bilingual adaptations to changes
in the environment are domain-specific, more research is needed to determine if the pattern of
results from this study will replicate with diverse samples and methods.
A consistent finding across Studies 1 et 2 was a trade-off in performance between the
Training and Test phases. Groups that showed stronger performance (c'est à dire., faster learning,
and perhaps stronger encoding) during Training had more difficulty learning the new asso-
ciation at Test, and vice-versa. In three out of four studies, all with nonlinguistic stimuli,
OPEN MIND: Discoveries in Cognitive Science
111
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
.
/
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
monolinguals were faster than bilinguals to learn the contingency between the cue and the
reward during the Training phase. This would have given them more time to strengthen
their anticipatory response, which in turn could have led to more difficulty updating their
response to a new contingency at test. Inversely, bilinguals might have had weaker initial
representations and responses, and thus found it easier to learn a new contingency. Le
reverse could be true for the one study that used linguistic stimuli (Study 2c). These dif-
ferences in learning strategies between bilinguals and monolinguals are in line with previ-
ous studies that have also reported trade-offs for bilingual children (Struys et al., 2018) et
adultes (Leivada et al., 2020).
Dans l'ensemble, our results are difficult to explain under theories that view executive functions as
the locus of monolingual–bilingual cognitive differences (par exemple., Bialystok, 2017). Kovács and
Mehler (2009un) interpreted their original results as showing that bilingualism enhances cogni-
tive control even in infancy, arguing that bilinguals were more able than monolinguals to
inhibit their original response and learn a new response. Cependant, this interpretation rested
on the two groups performing similarly in the Training phase—a finding that we did not rep-
licate. Plutôt, we found that monolinguals and bilinguals already showed behavioral differ-
ences during Training, a simple learning task that does not place any particular demands on
inhibition or cognitive control. While these findings do not negate the potential role of exec-
utive function in monolingual–bilingual differences, such theories do not provide a complete
explanation for the pattern of results we observed.
A compelling hypothesis that is more consistent with our data was advanced by D’Souza
et autres. (2020): bilingual infants might become more active in sampling multiple sources of
information as they interact with a more variable linguistic (and possibly sociocultural) envi-
ronment. It could be the case that bilinguals required more information during the Training
phase before responding, and were thus slower in associating the cue and the reward (at least
when nonlinguistic stimuli were used). This experience-dependent adaptive pattern could
explain overall differences between monolinguals’ and bilinguals’ anticipations and might
impact different areas of cognition (domain-general; Kovács & Mehler, 2009un). En effet,
monolingual–bilingual differences have been reported for a range of abilities, Par exemple,
in stimulus encoding and recognition (Singh et al., 2014), memory generalization (Brito & Barr,
2012; Brito et al., 2015; Brito et al., 2020), associative learning (Kovács & Mehler, 2009b), et
executive function (Poulin-Dubois et al., 2011). Surtout, the hypothesis proposed by
D’Souza et al. shifts the focus from the specific, and hard to define, cognitive ability that
the anticipatory looking paradigm measures (par exemple., inhibition, cognitive flexibility, attentional
processes) to a broader behavioral pattern (exploration) and to environmental variables that
might explain it. In this direction, future research could use infant-controlled paradigms to
more directly compare how monolinguals and bilinguals explore environments with different
levels of complexity (par exemple., Kidd et al., 2012).
Our findings have implications for future replications and open science in general. Many
recent open science endeavors in infant research (par exemple., ManyBabies Consortium, 2020) have
focused on reducing false positives (Type I error) that might arise from a number of factors,
which is important given the typical small samples in infant research. Cependant, our research
provides several directions for increasing statistical power independent of the number of
infants tested.
D'abord, we show that some standard analytic approaches may inadvertently increase the
chance of false negatives (Type II error; Jaeger, 2008). The nuanced pattern of results from
all five studies was only found when using an analytic approach that harnesses the rich
OPEN MIND: Discoveries in Cognitive Science
112
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
moment-to-moment eye-tracking data (up to 80–110 data points per infant; see also Barr,
2008; Humphrey & Swingley, 2018, for related approaches; and Wood, 2017, for a nonlinear
approche), which was not revealed in the original trial-level analysis reported by D’Souza
et autres. (2020, Experiment 1; up to 18 data points per infant) or the block-level analysis reported
by Kalashnikova et al. (2020; up to 6 data points per infant).
Deuxième, we were able to examine the effects of applying strict versus loose inclusion cri-
teria (c'est à dire., analyzing data from all infants versus only trials and infants with a minimum level of
attention; see Table 4; see also ManyBabies Consortium, 2020, for evidence that infants who
contribute more trials show larger effect sizes). Pour 7- to 9-month-olds, both strict and loose
criteria led to similar effect size estimates and patterns of statistical significance in most cases.
Cependant, for 20-month-olds, who appeared less engaged in the task, some effects only
reached significance when strict inclusion criteria were applied. Applying strict filtering cri-
teria meant higher exclusion rates (reducing sample sizes by 34–48%), but it ensured that ana-
lyzed data came from an on-task sample that had enough experience with the task to learn its
contingencies. The effects of filtering data in this way likely depend on the study design, le
age of the infants, and the analytic technique. It will be helpful for future studies to also sys-
tematically compare filtered versus unfiltered data.
Troisième, we observed striking differences in the datasets of 7- to 9-month-olds we analyzed,
both in terms of effect sizes (par exemple., the reference, or monolingual, odds of correctly anticipating
with time bin ranging from 1.43 à 6.47, with nonoverlapping confidence intervals) and pro-
portions of variance explained across different studies (10–32% of fixed effects; 45–81% of
fixed and random effects). This supports what most infant researchers know from experience:
seemingly small details of experimental design and/or infant populations can have a large
impact on infant performance. Systematic comparisons of different design variables, au-delà
small pilot studies, will be important in further increasing statistical power in infant research
(Bergmann et al., 2018). En outre, the high level of measurement error common in infant
recherche, together with small sample sizes, increases the rates of false positives in this field,
while also making it more difficult to detect true effects (Byers-Heinlein, Bergmann, & Savalei,
2021). Larger samples or multiple replications using smaller samples (the approach we took
ici) are crucial for drawing strong conclusions (Bergmann et al., 2018).
Enfin, our reanalyses were only possible because other researchers shared their datasets
and answered our queries—open science practices that allowed us to make new discoveries
(see Bolibaugh et al., 2021, and Dal Ben et al., 2022, for recent discussions of open data and
materials in bilingualism research). We hope that future researchers also benefit from the data-
sets we have generated and shared. In an open science spirit, further large-scale, pre-registered
experimentation with infants from a range of ages, cultures, languages, and using standardized
stimuli and analytic approaches could also be performed (par exemple., Byers-Heinlein et al., 2020;
Byers-Heinlein, Tsui, et coll., 2021; ManyBabies Consortium, 2020). This approach could help
to better understand experimental-level moderators that might affect performance during data
collection (par exemple., different sets of stimuli, conception expérimentale) and infant-level moderators that
might affect learning and cognition over development (par exemple., different sociolinguistic contexts).
Bilingual infants experience a complex linguistic environment as they navigate between
their languages. Our findings further demonstrate that this experience affects bilinguals’ cog-
nitive abilities as early as age 7 mois. In two original and three reanalyzed open datasets, nous
show that bilingual and monolingual infants might have different learning trajectories. Four out
of the five datasets we analyzed support the idea that bilinguals build less rigid initial repre-
sentations of the world, which in turn are easier to update when circumstances change. On the
OPEN MIND: Discoveries in Cognitive Science
113
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
.
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
other hand, monolinguals seem to be faster in building and strengthening initial representa-
tion, making it harder to update them when circumstances change. Cependant, in one dataset
we observed the reverse pattern, leaving open questions for future research. Ainsi, our results
cannot be easily explained under the traditional notion of enhancements in cognitive control
or executive function in bilingual infants. Plutôt, we believe that these bilingual–monolingual
differences are adaptations that arise from dealing with environments with different degrees of
linguistic (and probably social) variability, and as such can be viewed as either a help or a
hindrance, depending on the context. The origins of such adaptations, moderators, and their
impact later in life are interesting open questions. Large-scale data collection from diverse
bilingual infants, together with analyses that tap into the dynamics of infant learning, have
great potential to contribute to the complex and exciting debate of bilingualism’s effects in
infancy.
REMERCIEMENTS
We thank Dean D’Souza and Marina Kalashnikova for their responsiveness in providing addi-
tional demographic information for the reanalyses conducted in Study 2 (un, b, and c). We also
thank the members of our lab for their assistance with data collection and feedback on previ-
ous versions of the manuscript.
INFORMATIONS SUR LE FINANCEMENT
KBH, Natural Sciences and Engineering Research Council of Canada, Award ID: 402470-
2011. KBH, Natural Sciences and Engineering Research Council of Canada, Award ID:
2018-04390. SPI, Fonds de Recherche du Québec-Société et Culture (https://dx.doi.org/10
.13039/100008240). SPI, Concordia University (https://dx.doi.org/10.13039
/501100002914). RDB, Concordia University (https://dx.doi.org/10.13039/501100002914).
CONTRIBUTIONS DES AUTEURS
RDB and HK share first authorship and are listed alphabetically. RDB: Conceptualisation:
Equal; Conservation des données: Equal; Analyse formelle: Equal; Logiciel: Equal; Validation: Equal; Writ-
ing – original draft: Equal; Rédaction – révision & édition: Equal. HK: Conservation des données: Lead; Pour-
mal analysis: Lead; Logiciel: Lead; Validation: Equal; Visualisation: Lead; Écriture – originale
brouillon: Equal; Rédaction – révision & édition: Equal. SPI: Conceptualisation: Lead; Enquête:
Lead; Méthodologie: Equal. KBH: Conceptualisation: Lead; Acquisition de financement: Lead; Pro-
ject administration: Lead; Ressources: Lead; Surveillance: Lead; Rédaction – ébauche originale: Equal;
Rédaction – révision & édition: Equal.
RÉFÉRENCES
Adesope, Ô. O., Lavin, T., Thompson, T., & Ungerleider, C. (2010).
A systematic review and meta-analysis of the cognitive correlates
of bilingualism. Review of Educational Research, 80(2), 207–245.
https://doi.org/10.3102/0034654310368803
Barac, R., & Bialystok, E. (2011). Cognitive development of bilin-
gual children. Language Teaching, 44(1), 36–54. https://doi.org
/10.1017/S0261444810000339
Barac, R., Bialystok, E., Castro, D. C., & Sanchez, M.. (2014). Le
cognitive development of young dual language learners: A criti-
cal review. Early Childhood Research Quarterly, 29(4), 699–714.
https://doi.org/10.1016/j.ecresq.2014.02.003, PubMed:
25284958
Barr, D. J.. (2008). Analyzing “visual world” eyetracking data using
multilevel logistic regression. Journal of Memory and Language,
59(4), 457–474. https://doi.org/10.1016/j.jml.2007.09.002
Bates, D., Mächler M., Bolker B., & Walker S. (2015). Fitting linear
mixed-effects models using lme4. Journal of Statistical Software,
67(1), 1–48. https://doi.org/10.18637/jss.v067.i01
Baum, S., & Titone, D. (2014). Moving toward a neuroplasticity
view of bilingualism, executive control, and aging. Applied
OPEN MIND: Discoveries in Cognitive Science
114
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
.
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
.
/
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
Psycholinguistics, 35(5), 857–894. https://est ce que je.org/10.1017
/S0142716414000174
Bergmann, C., Tsuji, S., Piccinini, P.. E., Lewis, M.. L., Braginsky, M.,
Frank, M.. C., & Cristia, UN. (2018). Promoting replicability in
developmental research through meta-analyses: Insights from
language acquisition research. Child Development, 89(6),
1996–2009. https://doi.org/10.1111/cdev.13079, PubMed:
29736962
Bialystok, E. (2007). Cognitive effects of bilingualism: How linguis-
tic experience leads to cognitive change. Revue internationale de
Bilingual Education & Bilingualism, 10(3), 210–223. https://est ce que je
.org/10.2167/beb441.0
Bialystok, E. (2010). Global–local and trail-making tasks by mono-
lingual and bilingual children: Beyond inhibition. Developmen-
tal Psychology, 46(1), 93–105. https://doi.org/10.1037
/a0015466, PubMed: 20053009
Bialystok, E. (2017). The bilingual adaptation: How minds accom-
modate experience. Psychological Bulletin, 143(3), 233–262.
https://doi.org/10.1037/bul0000099, PubMed: 28230411
Bialystok, E., & Barac, R.. (2013). Cognitive effects. In F. Grosjean &
P.. Li (Éd.), The Psycholinguistics of Bilingualism (pp. 192–213).
Wiley-Blackwell.
Bialystok, E., Craik, F. JE., & Freedman, M.. (2007). Bilingualism as
a protection against the onset of symptoms of dementia. Neuro-
psychologia, 45(2), 459–464. https://est ce que je.org/10.1016/j
.neuropsychologia.2006.10.009, PubMed: 17125807
Bialystok, E., & Martine, M.. M.. (2004). Attention and inhibition in
bilingual children: Evidence from the dimensional change card
sort task. Developmental Science, 7(3), 325–339. https://doi.org
/10.1111/j.1467-7687.2004.00351.x, PubMed: 15595373
Blanco-Elorrieta, E., & Caramazza, UN. (2021). A common selection
mechanism at each linguistic level in bilingual and monolingual
language production. Cognition, 213, Article 104625. https://est ce que je
.org/10.1016/j.cognition.2021.104625, PubMed: 33608129
Bolibaugh, C., Vanek, N., & Marsden, E. J.. (2021). Towards a
credibility revolution in bilingualism research: Open data and
materials as stepping stones to more reproducible and replica-
ble research. Bilingualism: Language and Cognition, 24(5),
801–806. https://doi.org/10.1017/S1366728921000535
Bosch, L., & Sebastián-Gallés, N. (2001). Evidence of early language
discrimination abilities in infants from bilingual environments.
Infancy, 2(1), 29–49. https://doi.org/10.1207/S15327078IN0201_3,
PubMed: 33451225
Brito, N., & Barr, R.. (2012). Influence of bilingualism on memory
generalization during infancy. Developmental Science, 15(6),
812–816. https://doi.org/10.1111/j.1467-7687.2012.1184.x,
PubMed: 23106735
Brito, N., Greaves, UN., Leon-Santos, UN., Fifer, W., & Noble, K.
(2020). Associations between bilingualism and memory general-
ization during infancy: Does socioeconomic status matter? Bilin-
gualism: Language and Cognition, 24(2), 1-dix. https://est ce que je.org/10
.1017/S1366728920000334, PubMed: 33776545
Brito, N., Sebastián-Gallés, N., & Barr, R.. (2015). Differences in
language exposure and its effects on memory flexibility in
monolingual, bilingual, and trilingual infants. Bilingualism:
Language and Cognition, 18(4), 1–13. https://est ce que je.org/10.1017
/S1366728914000789
Byers-Heinlein, K., Bergmann, C., & Savalei, V. (2021). Six solu-
tions for more reliable infant research. Infant and Child Develop-
ment, Article e2296. https://doi.org/10.1002/icd.2296
Byers-Heinlein, K., Schott, E., Gonzalez-Barrero, UN. M., Brouillard,
M., Dubé, D., Jardak, UN., Laoun-Rubenstein, UN., Mastroberardino,
M., Morin-Lessard, E., Pour Iliaei, S., Salama-Siroishka, N., &
Tamayo, M.. P..
(2019). MAPLE: A Multilingual Approach to
Parent Language Estimates. Bilingualism: Language and Cogni-
tion, 23(5), 1–7. https://doi.org/10.1017/S1366728919000282
Byers-Heinlein, K., Tsui, UN., Bergmann, C., Noir, UN. K., Brun, UN.,
Carbajal, M.. J., Durrant, S., Fennell, C. T., Fiévet, A.-C., Frank,
M.. C., Gampe, UN., Gervain, J., Gonzalez-Gomez, N., Kiley Hamlin,
J., Havron, N., Hernik, M., Kerr, S., Killam, H., Klassen, K., …
Wermelinger, S. (2021). A multi-lab study of bilingual infants:
Exploring the preference for infant-directed speech. Advances
in Methods and Practices in Psychological Science. 4(1).
https://doi.org/10.1177/2515245920974622
Byers-Heinlein, K., Tsui, R.. K., van Renswoude, D., Noir, UN. K.,
Barr, R., Brun, UN., Colomer, M., Durrant, S., Gampe, UN.,
Gonzalez-Gomez, N., Hay, J.. F., Hernik, M., Jartó, M., Kovács,
Á. M., Laoun-Rubenstein, UN., Lew-Williams, C., Liszkowski, U.,
Liu, L., Noble, C., … Singh, L. (2020). The development of
gaze following in monolingual and bilingual infants: A multi-
laboratory study. Infancy, 26(1), 4–38. https://doi.org/10.1111
/infa.12360, PubMed: 33306867
Canfield, R.. L., Forgeron, E. G., Brezsnyak, M.. P., & Snow, K. L. (Éd.).
(1997). Information processing through the first year of life: UN
longitudinal study using the visual expectation paradigm. Mono-
graphs of the Society for Research in Child Development, 62(2).
https://doi.org/10.1111/j.1540-5834.1997.tb00511.x, PubMed:
9353949
Comishen, K. J., Bialystok, E., & Adler, S. UN. (2019). The impact of
bilingual environments on selective attention in infancy. Devel-
opmental Science, 22(4), Article e12797. https://doi.org/10.1111
/desc.12797, PubMed: 30600863
Costa, UN., Hernández, M., & Sebastián-Gallés, N. (2008). Bilingual-
ism aids conflict resolution: Evidence from the ANT task. Cogni-
tion, 106(1), 59–86. https://doi.org/10.1016/j.cognition.2006.12
.013, PubMed: 17275801
Dal Ben, R., Brouillard, M., Gonzalez-Barrero, UN. M., Killam, H.,
Kremin, L. V., Quirk, E., Sander-Montant, UN., Schott, E., Tsui, R..
K.-Y., Byers-Heinlein, K. (2022). How open science can benefit
bilingualism research: A lesson in six tales. Bilingualism:
Language and Cognition, First View, 1–8. https://est ce que je.org/10
.1017/S1366728922000256
DeBolt, M.. C., Rhemtulla, M., & Oakes, L. M.. (2020). Robust data
and power in infant research: A case study of the effect of
number of infants and number of trials in visual preference pro-
cedures. Infancy, 25(4), 393–419. https://doi.org/10.1111/infa
.12337, PubMed: 32744759
de Bruin, UN., Treccani, B., & Della Sala, S. (2015). Cognitive advan-
tage in bilingualism: An example of publication bias? Psycholog-
i c a l S c i e n c e , 2 6 ( 1 ) , 9 9 – 1 0 7 . h t t p s : / / d o i . o rg / 1 0 . 11 7 7
/0956797614557866, PubMed: 25475825
Diamond A. (2013). Executive functions. Annual Review of Psy-
cologie, 64, 135–168. https://doi.org/10.1146/annurev-psych
-113011-143750, PubMed: 23020641
Dixon, P.. (2008). Models of accuracy in repeated-measures
designs. Journal of Memory and Language, 59(4), 447–456.
https://doi.org/10.1016/j.jml.2007.11.004
D’Souza, D., Brady, D., Haensel, J.. X., & D’Souza, H. (2020). Is
mere exposure enough? The effects of bilingual environments
on infant cognitive development. Royal Society Open Science,
7(2), Article 180191. https://doi.org/10.1098/rsos.180191,
PubMed: 32257297
Edwards, D., & Christophersen, H. (1988). Bilingualism, literacy
and meta-linguistic awareness in preschool children. British Jour-
nal of Developmental Psychology, 6(3), 235–244. https://doi.org
/10.1111/j.2044-835X.1988.tb01097.x
OPEN MIND: Discoveries in Cognitive Science
115
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
Epstein, R.. (1905). La pensee et Iapoligloise. Libraire Payot.
Fenson, L., Dale, P.. S., Reznick, J.. S., Thal, D., Bates, E., Hartung,
J.. P., Pethick, S., & Reilly, J.. S. (1993). The MacArthur communi-
cative development inventories: Users guide and technical
manual. Singular Publishing Group.
Fernández-López, M., & Perea, M.. (2019). The bilingualism wars: Is
the bilingual advantage out of (executive) control? Psicológica,
40(1), 26–33. https://doi.org/10.2478/psicolj-2019-0002
Frank, M.. C., Bergelson, E., Bergmann, C., Cristia, UN., Floccia, C.,
Gervain, J., Kiley Hamlin, J., Hannon, E. E., Kline, M., Levelt, C.,
Lew-Williams, C., Nazzi, T., Panneton, R., Rabagliati, H.,
Soderstrom, M., Sullivan, J., Waxman, S., & Yurovsky, D.
(2017). A collaborative approach to infant research: Promoting
reproducibility, best practices, and theory-building. Infancy, 22(4),
421–435. https://doi.org/10.1111/infa.12182, PubMed:
31772509
Gold, B. T., Kim, C., Johnson, N. F., Kryscio, R.. J., & Forgeron, C. D.
(2013). Lifelong bilingualism maintains neural efficiency for
cognitive control in aging. The Journal of Neuroscience,
33(2), 387–396. https://doi.org/10.1523/ JNEUROSCI.3837-12
.2013, PubMed: 23303919
Gonzalez-Barrero, UN. M., Schott, E., & Byers-Heinlein, K. (2020).
Bilingual adjusted vocabulary: A developmentally-informed
bilingual vocabulary measure (Manuscript submitted for publica-
tion). Département de psychologie, Concordia University, Mon-
treal, Canada. https://doi.org/10.31234/osf.io/x7s4u
Vert, D. W. (1998). Mental control of the bilingual lexico-
semantic system. Bilingualism: Language and Cognition, 1(2),
67–81. https://doi.org/10.1017/S1366728998000133
Hilchey, M.. D., & Klein, R.. M.. (2011). Are there bilingual advan-
tages on non-linguistic interference tasks? Implications for the
plasticity of executive control processes. Psychonomic Bulletin
and Review, 18, 625–658. https://doi.org/10.3758/s13423-011
-0116-7, PubMed: 21674283
Hoff, E., Core, C., Place, S., Rumiche, R., Señor, M., & Parra, M..
(2012). Dual language exposure and early bilingual develop-
ment. Journal of Child Language, 39(1), 1–27. https://est ce que je.org/10
.1017/S0305000910000759, PubMed: 21418730
Humphrey, C., & Swingley, D. (2018). Regression analysis of pro-
portion outcomes with random effects. ArXiv. https://arxiv.org
/abs/1805.08670. https://doi.org/10.48550/arXiv.1805.08670
Ibánez-Lillo, UN., Pons, F., Costa, UN., & Sebastián-Gallés, N. (2010).
Inhibitory control in 8-month-old monolingual and bilingual
infants: Evidence from an anticipatory eye movement task. Poster
presented at the 22nd Biennial International Conference on
Infant Studies, March 11–14, Baltimore, MARYLAND.
Jaeger, T. F. (2008). Categorical data analysis: Away from ANOVAs
(transformation or not) and towards logit mixed models. Journal
of Memory and Language, 59(4), 434–446. https://est ce que je.org/10
.1016/j.jml.2007.11.007, PubMed: 19884961
Kalashnikova, M., Pejovic, J., & Carreiras, M.. (2020). The effects of
bilingualism on attentional processes in the first year of life.
Developmental Science, 24(2), Article e13011. https://doi.org
/10.1111/desc.13011, PubMed: 32603543
Kalashnikova, M., Pejovic, J., & Carreiras, M.. (2021). Corrigendum
to Kalashnikova, M., Pejovic, J., & Carreiras, M.. (2020). Le
effects of bilingualism on attentional processes in the first year
de la vie. Developmental Science, 25(1), Article e13011. https://
doi.org/10.1111/desc.13139
Kavé, G., Eyal, N., Shorek, UN., & Cohen-Mansfield, J.. (2008).
Multilingualism and cognitive state in the oldest old. Psychologie
and Aging, 23(1), 70–78. https://doi.org/10.1037/0882-7974.23
.1.70, PubMed: 18361656
Kidd, C., Piantadosi, S. T. & Aslin, R.. N. (2012). The Goldilocks
Effect: Human infants allocate attention to visual sequences that
are neither too simple nor too complex. PLoS ONE, 7(5), Article
e36399. https://doi.org/10.1371/journal.pone.0036399,
PubMed: 22649492
Kousaie, S., Chaï, X. J., Sander, K. M., & Klein, D. (2017). Simulta-
neous learning of two languages from birth positively impacts
intrinsic functional connectivity and cognitive control. Cerveau
et cognitif, 117, 49–56. https://doi.org/10.1016/j.bandc
.2017.06.003, PubMed: 28648285
Ková cs, Á. M., & Mehler, J.. (2009un). Cognitive gains in
seven-month-old bilingual infants. Actes de la Nationale
Académie des Sciences, 106(16), 6556–6560. https://est ce que je.org/10
.1073/pnas.0811323106, PubMed: 19365071
Kovács, Á. M., & Mehler, J.. (2009b). Flexible learning of multiple
speech structures in bilingual infants. Science, 325(5940),
611–612. https://doi.org/10.1126/science.1173947, PubMed:
19589962
Leivada, E., Westergaard, M., Duñabeitia, J.. UN., & Rothman, J..
(2020). On the phantom-like appearance of bilingualism effects
on neurocognition: (Comment) should we proceed? Bilingualism:
Language and Cognition, 24(1), 1–14. https://est ce que je.org/10.1017
/S1366728920000358
Macnamara, J.. (1966). Bilingualism and primary education.
Edinburgh University Press.
ManyBabies Consortium. (2020). Quantifying sources of variability
in infancy research using the infant-directed-speech preference.
Advances in Methods and Practices in Psychological Science,
3(1), 24–52. https://doi.org/10.1177/2515245919900809
McMurray, B., & Aslin, R.. N. (2004). Anticipatory eye movements
reveal infants’ auditory and visual categories. Infancy, 6(2),
203–229. https://doi.org/10.1207/s15327078in0602_4,
PubMed: 33430530
Molnar, M., Pejovic, J., Yee, E., & Carreiras, M.. (2014). Cognitive
gains in Basque-Spanish bilingual infants? Poster Presented at
the International Conference on Infant Studies, July 3–5, Berlin,
Allemagne.
Open Science Collaboration. (2015). Estimating the reproducibility
of psychological science. Science, 349(6251). https://est ce que je.org/10
.1126/science.aac4716, PubMed: 26315443
Paap, K. R., Anders-Jefferson, R., Mason, L., Alvarado, K., & Zimiga,
B. (2018). Bilingual advantages in inhibition or selective atten-
tion: More challenges. Frontiers in Psychology, 9. https://doi.org
/10.3389/fpsyg.2018.01409, PubMed: 30158887
Paap, K. R., & Greenberg, Z. je. (2013). There is no coherent evi-
dence for a bilingual advantage in executive processing. Cogni-
tive Psychology, 66(2), 232–258. https://est ce que je.org/10.1016/j
.cogpsych.2012.12.002, PubMed: 23370226
Paap, K. R., Johnson, H. UN., & Sawi, Ô. (2015). Bilingual advan-
tages in executive functioning either do not exist or are restricted
to very specific and undetermined circumstances. Cortex, 69,
265–278. https://doi.org/10.1016/j.cortex.2015.04.014,
PubMed: 26048659
Pearson, B. Z., Fernández, S. C., & Oller, D. K. (1993). Lexical
development in bilingual infants and toddlers: Comparison to
monolingual norms. Language Learning, 43(1), 93–120. https://
doi.org/10.1111/j.1467-1770.1993.tb00174.x
Philipp, UN. M., Gade, M., & Koch, je. (2007) Inhibitory processes in
language switching: Evidence from switching language-defined
response sets. European Journal of Cognitive Psychology, 19(3),
395–416. https://doi.org/10.1080/09541440600758812
Poulin-Dubois, D., Blaye, UN., Coutya, J., & Bialystok, E. (2011). Le
effects of bilingualism on toddlers’ executive functioning. Journal
OPEN MIND: Discoveries in Cognitive Science
116
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
/
/
.
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Effects of Bilingualism on Cognition Dal Ben et al.
of Experimental Child Psychology, 108(3), 567–579. https://est ce que je
.org/10.1016/j.jecp.2010.10.009, PubMed: 21122877
Reuter, T., Emberson, L., Romberg, UN., & Lew-Williams, C. (2018).
Individual differences in nonverbal prediction and vocabulary
size in infancy. Cognition, 176, 215–219. https://est ce que je.org/10
.1016/j.cognition.2018.03.006, PubMed: 29604470
Simmons, J.. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive
psychologie: Undisclosed flexibility in data collection and analysis
allows presenting anything as significant. Sciences psychologiques,
22(11), 1359–1366. https://doi.org/10.1177/0956797611417632,
PubMed: 22006061
Singh, L., Fu, C. S. L., Rahman, UN. UN., Hameed, W. S., Sanmugam,
S., Agarwal, P., Jiang, B., Seng Chong, Y., Meaney, M.. J., Rifkin-
Graboi, UN., & GUSTO Research Team. (2014). Back to the basics:
A bilingual advantage in infant visual habituation. Child Devel-
opération, 86(1), 294–304. https://doi.org/10.1111/cdev.12271,
PubMed: 25074016
Struys, E., Duyck, W., & Woumans, E. (2018). The role of cognitive
development and strategic task tendencies in the bilingual
advantage controversy. Frontiers in Psychology, 9. https://est ce que je
.org/10.3389/fpsyg.2018.01790, PubMed: 30319495
Trudeau, N., Frank, JE., & Poulin-Dubois, D. (1999). Une adaptation
en franç ais qué bécois du MacArthur Communicative Develop-
ment Inventory [A Quebec-French Adaptation of the MacArthur
Communicative Development Inventory]. Revue d’orthophonie
et d’audiologie, 23, 61–73.
Tsui, UN. S. M., & Fennell, C. T. (2019). Do bilingual infants possess
enhanced cognitive skills? PsyArXiv. https://doi.org/10.31234/osf
.io/dcn7y
Wood, S. N. (2017). Generalized additive models: An introduction
with R (2nd ed.). CRC Press/Taylor & Francis Group. https://est ce que je
.org/10.1201/9781315370279
Yelland, G. W., Pollard, J., & Mercuri, UN. (1993). The metalinguistic ben-
efits of limited contact with a second language. Applied Psycholinguis-
tics, 14(4), 423–444. https://doi.org/10.1017/S0142716400010687
Yoshioka, J.. G. (1929). A study of bilingualism. Journal of Genetic
Psychologie, 36, 473–479. https://doi.org/10.1080/08856559
.1929.10532205
je
D
o
w
n
o
un
d
e
d
F
r
o
m
h
t
t
p
:
/
/
d
je
r
e
c
t
.
m
je
t
.
/
e
d
toi
o
p
m
je
/
je
un
r
t
je
c
e
–
p
d
F
/
d
o
je
/
je
.
/
/
1
0
1
1
6
2
o
p
m
_
un
_
0
0
0
5
7
2
0
3
3
4
9
8
o
p
m
_
un
_
0
0
0
5
7
p
d
/
.
je
F
b
oui
g
toi
e
s
t
t
o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
OPEN MIND: Discoveries in Cognitive Science
117
