Co-learning companionship benefits word learning in a new language: Evidence from a dual-brain EEG examination

Abstract

Companionship refers to one’s being in the presence of another individual. For adults, acquiring a new language is a highly social activity that often involves learning in the context of companionship. However, the effects of companionship on new language learning have gone relatively underexplored, particularly with respect to word learning. Using a within-subject design, the current study employs electroencephalography to examine how two types of companionship (monitored and co-learning) affect word learning (semantic and lexical) in a new language. Dyads of Chinese speakers of English as a second language participated in a pseudo-word-learning task during which they were placed in monitored and co-learning companionship contexts. The results showed that exposure to co-learning companionship affected the early attention stage of word learning. Moreover, in this early stage, evidence of a higher representation similarity between co-learners showed additional support that co-learning companionship influenced attention. Observed increases in delta and theta interbrain synchronization further revealed that co-learning companionship facilitated semantic access. In all, the similar neural representations and interbrain synchronization between co-learners suggest that co-learning companionship offers important benefits for learning words in a new language.

Full text

Received: April 23, 2024. Revised: June 19, 2024. Accepted: June 28, 2024
© The Author(s) 2024. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permission@oup.com.
Cerebral Cortex, 2024, 34, bhae289
https://doi.org/10.1093/cercor/bhae289
Advance access publication date 16 July 2024
Original Article
Co-learning companionship benefits word learning in
a new language: Evidence from a dual-brain EEG
examination
Yujing Shen1,2, Xu Liu1,2, Yingyi Xiang1,2, John W. Schwieter3,4, Huanhuan Liu 1,2,
*
1Research Center of Brain and Cognitive Neuroscience, Liaoning Normal University, 850 Huanghe Road, Shahekou District, Dalian, Liaoning Province, Dalian
116029, China
2Key Laboratory of Brain and Cognitive Neuroscience, 850 Huanghe Road, Shahekou District, Liaoning Province, Dalian 116029, China
3Language Acquisition, Cognition, and Multilingualism Laboratory/Bilingualism Matters, Wilfrid Laurier University, Waterloo, ON N2L 3C5, Canada
4Department of Linguistics and Languages, McMaster University, Waterloo, ON N2L 3C5, Canada
*Corresponding author: Huanhuan Liu, Research Center of Brain and Cognitive Neuroscience, Liaoning Normal University, Dalian 116029, China. Email:
abcde69503@126.com
Companionship refers to one’s being in the presence of another individual. For adults, acquiring a new language is a highly social
activity that often involves learning in the context of companionship. However, the effects of companionship on new language learning
have gone relatively underexplored, particularly with respect to word learning. Using a within-subject design,the current study employs
electroencephalography to examine how two types of companionship (monitored and co-learning) affect word learning (semantic and
lexical) in a new language. Dyads of Chinese speakers of English as a second language participated in a pseudo-word-learning task
during which they were placed in monitored and co-learning companionship contexts. The results showed that exposure to co-learning
companionship affected the early attention stage of word learning. Moreover, in this early stage, evidence of a higher representation
similarity between co-learners showed additional support that co-learning companionship influenced attention. Observed increases
in delta and theta interbrain synchronization further revealed that co-learning companionship facilitated semantic access. In all, the
similar neural representations and interbrain synchronization between co-learners suggest that co-learning companionship offers
important benefits for learning words in a new language.
Key words: language word learning; companionship; representation similarity; interbrain synchronization.
Introduction
In both native and second language acquisition, social interaction
and the companionship of caregivers and peers play important
roles. Previous studies have shown a strong link between social
interaction and learning (Gass and Mackey 2014), and regarding
companionship, researchers argue that more meaningful and
engaging learning processes can be fostered (Kinginger 2000), as
companionship inherently involves mutual influences between
people. Monitored and co-learning are two common types of
companionship (Dindar and Akcay 2015; Lytle et al. 2018). For
instance, in monitored companionship, an individual, oftentimes
a parent or teacher, monitors the learning of another individual. In
co-learning, learners interact with one another in pairs or groups.
According to the Social Foreign Language Learning Theory (Li and
Jeong 2020), in traditional new language learning contexts, indi-
viduals typically learn new words through picture-word (semantic
learning) or word-word (lexical learning) associations. These two
methods rely heavily on individuals’ own efforts in their learn-
ing processes (García-Gámez and Macizo 2022). However, in the
context of these two learning methods, the possible influence of
companionship has not yet been studied. Due to this gap in the
literature, little is known about whether and how companionship
affects word learning in a new language.
Companionship refers to one’s presence in the context of being
with someone else (Bowlby 1982;Hadot 1995). In complex learning
tasks, the presence of others may share or increase an individual’s
cognitive load, depending on the role of the companion and
the nature of the interaction (Papeo and Abassi 2019; Skuballa
et al. 2019). Presence and joint attention, a social phenomenon
that is achieved when two individuals are attending to the same
things within a shared environment, occur simultaneously (Steier
2020). Joint attention is coordinated through mutual awareness
(Tomasello 1995), and the variability in the propensity to use
joint attentional abilities can be inf luenced by monitor/co-learner
behaviors, such as those observed in early companionship and
parent–child interactions (Mundy et al. 2003).
In monitored companionship, the presence of a companion
who monitors the learning process can prompt learners to choose
and use self-regulated learning strategies more effectively, thus
improving learning outcomes (Van der Stel and Veenman 2014).
However, joint attention may be impaired during monitoring if
learners may pay too much attention to the monitor’s reactions
because they may be overly focused on making a mistake. Thus,in
monitored companionship situations, it may be difficult for some
learners to concentrate on the task at hand (Feinberg and Aiello
2010).
Co-learning companionship is a state of “collective presence”
that avoids the sense of isolation that comes with solitary learning
and instead, offers accessible and concrete companionship
(Collins 2014). Emotional support from a partner may help
learners to better focus on learning and memory tasks (Zajonc
1965). Specifically, as cognitive load is shared among partners,
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

2|Cerebral Cortex, 2024, Vol. 34, No. 7
co-learning can lessen a learner’s cognitive load and enhance
allocation of cognitive resources (Kirschner and Erkens 2013).
In other words, co-learning companionship may facilitate joint
attention among co-learners in social learning interactions.
Moreover, the use of joint attention by individuals contributes
to their cognitive and language development (Baldwin 2014).
Research has shown that the event-related potential (ERP)
component P2, a typical indicator of attention in social interaction
(Luyster et al. 2008), increases during joint attention (Bolt and
Loehr 2023). Such joint attention also elicits theta oscillations
(4–7 Hz) which are closely related to attentional adjustment
(Grubbs et al. 2020). As theta oscillations increase during
sustained joint attention, so does learners’ engagement in social
interactions (Michel et al. 2015; Hernandez et al. 2021).
For adults, the acquisition of words in a new language relies
on meaningful exposure to the target language (Elgort et al. 2015).
The Revised Hierarchical Model (Kroll and Stewart 1994)proposes
that novice language learners first rely on concepts linked to
words in their first language (L1) as a bridge to access and com-
prehend meaning in their second language (L2). As L2 proficiency
increases, learners develop the ability to directly access concepts
in their L2 without the need to associate them with their L1
translation equivalents. Under these assumptions, new language
learning involves both semantic and lexical learning, with both
types of learning having differential developmental trends (Smith
and Magee 1980; Hauk et al. 2006). The N400 component is a
typical electrophysiological index of lexical-semantic access of
words (Liu and van Hell 2020) which peaks ∼400 ms following
stimulus onset (Kutas and Federmeier 2011; Kummerer et al.
2013). Delta oscillations (1–3 Hz) and theta oscillations also ref lect
semantic processing. Delta oscillatory activity is thought to be
related to the integration and storage of semantic processing
(Voytek et al. 2015) in which increased oscillations are associated
with improved semantic processing (Bastiaansen and Hagoort
2015). Theta oscillations may also contribute to semantic process-
ing (Kahana 1996) during which different brain regions effectively
communicate information (Major et al. 2004). Moreover, both
semantic and lexical learning inevitably involve working memory,
which can be seen in the late positive component (LPC) in which
amplitude increases when individuals retain or process phonolog-
ical information during a memory task (McLaughlin et al. 2004).
The current study utilizes electroencephalography (EEG) hyper-
scanning to examine how learners acquire new language seman-
tic and lexical information in monitored and co-learning condi-
tions. To eliminate the inf luence of previous language learning
experiences, we created pseudowords which were presented with
pictures or word translations of their meaning in participants’ L1.
As shown in Fig. 1, in the monitored companionship condition,
the monitor simply watched the learner while the latter learned
pseudowords.In the co-learning companionship condition, the co-
learners participated in the learning together.
Based on previous research, co-learning companionship may
be a more effective approach to learning words in a new language
because its collaborative atmosphere may enhance joint attention
among dyads of learners. If this is the case, we expect to see
larger early attention (reflected by P2), improved semantic access
(pronounced by N400), and sustained attention (reflected by LPC).
However, the inf luence of companionship may be due to the
similarity of representation processing and the synchronization
of neural activity in monitored companionship (mo-learners and
monitors) or in co-learning companionship (co-learners and co-
learners). When peers perform similar cognitive tasks, their brain
activity patterns will show a certain degree of similarity which
may stem from the neural networks and cognitive processes that
are activated when jointly processing task information (Skuballa
et al. 2019). This may be valid neuroimaging evidence for pro-
moting increased joint attention. To examine this possibility, we
conducted a representation similarity analysis (RSA) to reveal the
temporal overlap of neural responses (McLaughlin et al. 2004;
Midgley et al. 2009) for individuals with different social roles
(mo-learner and monitor vs. co-learner and co-learner), with the
specific purpose of exploring the difference between representa-
tion similarity in the monitored condition (mo-learner and mon-
itor) and in the co-learning condition (co-learner and co-learner).
In addition, we analyzed inter-brain synchronization (calculated
based on phase-locking values, PLV) as an indicator of delta and
theta oscillations to capture joint neural activity during the two
learning situations.
Methods
Participants
We recruited 60 healthy adults whose L1 was Chinese and L2 was
English. These individuals were then randomly placed in dyads
of the same sex. Data from four dyads (n=8) were excluded from
the analyses due to excessive errors (> 20%) or EEG artifacts
(> 50%). Thus, the final sample included 26 dyads of par-
ticipants (42 females, 10 males), aged 19 to 26 years old
(M =21.25 ±2.63 years). All participants were undergraduate
students at Liaoning Normal University and were compensated
financially for ∼1 h of participation. Participants were right-
handed with normal or corrected-to-normal vision and reported
having no language disabilities or neurological issues. All
participants provided their informed consent before taking part
in the study.
Language proficiency was estimated using a language history
questionnaire (Liu et al. 2020) and the Oxford Placement Test
(maximum score 50 points; Allan 2004). In the questionnaire,
participants provided self-ratings of their language skills and
reported the age at which they started learning their L2. Self-
ratings were based on a 6-point scale, with “6” indicating perfect
fluency and “1” indicating no knowledge. Information about the
participants’ L1 and L2 proficiency and L2 age of acquisition
can be seen in Table 1. Paired sample t-tests showed that L1
proficiency was higher than L2 across all language skills. The
participants are estimated to be at an intermediate level of L2
proficiency (see also Liu et al. 2020 for a similar sample).
Stimuli
The stimuli consisted of 112 bisyllabic pseudo-words (e.g. djawi,
sbaru) taken from Chomsky and Halle (1968). Pseudo-words were
chosen to examine responses to new words without influence
of prior word knowledge. All stimuli belonged to the following
semantic categories: fruits, animal, food, kitchen, vehicles, sta-
tionery, body, and clothing. Ninety-six of these pseudo-words were
used in the formal experiment, and 16 words used as practice
trials (see Supplementary Table 1).
Procedure and task design
To avoid ambiguity of the pictures’ names in the experiment,
participants were first asked to familiarize themselves with the
pictures and their Chinese names. The experiment simulated
social situations of monitored and co-learning companionship.
The participants were randomly assigned to be part of a mo-
learner/co-learner or monitor/co-learner dyad, such that in the
monitored companionship condition, there was one mo-learner
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

Shen et al. |3
Fig. 1. Flow chart of the experiment. a1 and a2 represent the monitored and co-learning companionship conditions; b1 and b2 represent lexical and
semantic learning; c1 and c2 display the letter judgment and category judgment tasks.
Tab l e 1 . Information about participants’ language proficiency and age of acquisition.
L1 L2 t p
Self-ratings of proficiency
Listening 5.18 ±0.85 3.15 ± 0.76 16.176 < 0.001
Speaking 4.87 ±0.79 2.95 ± 0.72 15.435 < 0.001
Reading 4.32 ±1.14 2.48 ± 0.93 12.485 < 0.001
Writing 4.65 ±1.07 2.50 ± 1.29 11.412 < 0.001
Age of acquisition — 8.44 ±1.98 — —
Oxford placement test — 32.75 ± 5.19 — —
and one monitor,and in the co-learning companionship condition,
there were two co-learners. In the monitored companionship
condition, the mo-learner was required to learn new words, and
the monitor was asked to simply observe the mo-learner (see
Fig. 1a1). In the co-learning companionship condition, both co-
learners were required to learn new words together, and tri-
als of words could only be completed once the two co-learners
pressed a button. During the testing phase, an opaque foam board
(1 m×0.5 m) was placed between the two co-learners to obstruct
their view of each other (see Fig. 1a2).
Word-learning task
The stimuli were presented in eight blocks: four blocks of 12
pseudo-words accompanied by L1 translation words, and four
blocks of 12 pseudo-words accompanied by pictures. Among the
eight blocks, four were placed in a monitored companionship
condition and the other four in a co-learning companionship
condition. The two companionship conditions and word-learning
types were counterbalanced across dyads of participants. This
manipulation resulted in a 2×2 factorial design examining com-
panionship type (monitored companionship, co-learning com-
panionship) ×learning type (semantic learning, lexical learning).
The task consisted of a learning phase and testing phase as
described next.
Learning phase
The lexical learning condition involved four blocks of word-word
association trials (see Fig. 1b1). After a fixation point appeared
in the center of the screen for 500 ms, a target pseudo-word
and its Chinese translation were shown. Once the participants
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

4|Cerebral Cortex, 2024, Vol. 34, No. 7
pressed the space bar, a blank screen appeared on the screen for
1000 ms before the next trial appeared. The semantic learning
condition involved four blocks of word-picture association trials
(see Fig. 1b2) and followed the same presentation as the word-
word association condition. In both word-word and word-picture
conditions, participants were asked to learn 12 new words in each
block without a time limit.
Testing phase
Each experimental block was immediately followed by a testing
block, such that after word-word learning, participants made
letter judgments (i.e. participants determined whether the newly
learned pseudo-word contained the presented letter, see Fig. 1c1)
or category judgments (i.e. participants determined whether the
pseudo-word belonged to the presented specific semantic cate-
gory, see Fig. 1c2). After word-picture learning, the participants
also made letter and semantic judgments.
In all testing blocks, a 500 ms fixation point was first presented
followed by either a letter or semantic category. Participants were
then asked to judge whether the newly learned pseudo-word con-
tained the presented letter or was part of the presented semantic
category by pressing the "F key (yes) or "J" key (no). A blank screen
of 1000 ms was presented before the next trial appeared.Response
keys were counterbalanced across participants.
Behavioral data analyses
Accuracy scores and reaction times (RTs) (transformed to log
scale) were analyzed with a generalized logistic mixed-effects
model and a generalized linear mixed-effects model, respectively.
In both models, companionship type (monitored companionship,
co-learning companionship) and learning type (semantic learn-
ing, lexical learning) were added as fixed effects. Participants,
the two different testing types, and block order were added as
random effects. In addition, we added RTs of the learning phase
as a random variable for the mixed-effects model analysis (See
the Supplementary Tables 2, 3, 4 for detailed results). Analyses
were carried out using the lmer package (Bates et al. 2014), and
we performed multiple comparisons for the mixed effects model
using the multcomp package in R version 4.2.2 (R Development
Core Team, Kelley et al. 2018). An FDR correction for multiple
comparisons was applied to the obtained p values.
Representation similarity analysis
We used a RSA (Kriegeskorte et al. 2008) to examine the 26 dyads’
EEG data. The analysis of the similarity of dual-brain representa-
tions is to compare the representation patterns of neural process-
ing in both brains. The calculated RDM matrix can then be used
to compare the dyads’ representation similarity in the monitored
condition (mo-learner and monitor) and the co-learning condi-
tion (co-learner and co-learner). For each participant, trials were
divided into conditions based on the word-learning task (semantic
learning vs. lexical learning), as well as the two social situations
companionship (monitored vs. co-learning). The EEG data of the
two different social roles were analyzed separately, then aver-
aged in each condition to obtain two arrays with channels ×
subjects × time points × conditions. To measure neural dissim-
ilarity between conditions, we used the Python tool of NeuroRA
(neural representational analysis) with custom-written scripts.
NeuroRA includes a permutation test to establish any correlation
between the two RDMs. By comparing the neural responses asso-
ciated with each pair of experimental conditions, using correla-
tion distance (1-Spearman correlation) as the dissimilarity mea-
sure (Lu and Ku 2020), we obtained the social roles’ (including mo-
learner, monitor, and co-learner) representational dissimilarity
matrix (RDM) for each time point of the neural processing stages
(Kriegeskorte et al. 2008; Hall-McMaster et al. 2019). Finally, we
calculated the similarity (Spearman correlation) between the mo-
learner and monitor’s RDMS, as well as between the co-learner
and co-learner’s RDMS. Initially, we shuffled the values of both
RDMs and then recalculated the similarity matrix between them.
By replicating this procedure 5000 times, we obtained the final
p-values from the aforementioned permutation distribution (Lu
and Ku 2020).
Electrophysiological data analyses
Electrophysiological data were recorded using a set of 64 elec-
trodes placed according to the extended 10–20 positioning system.
The signal was recorded from ANT Neuro at a rate of 500 Hz in
reference to the CPz electrode site.The electrodes M1 and M2 were
separately placed on the left and right mastoids. Impedances were
kept below 5 kΩ. Offline processing was referenced to the aver-
age of M1 and M2. Electroencephalographic activity was filtered
online within a bandpass between 0.1 and 100 Hz and refiltered
offline with a high-pass filter of 0.01 Hz and a low-pass filter
of 30 Hz. Ocular artifact reduction was performed through Inde-
pendent Component Analysis component rejection using EEGLAB
(Delorme and Makeig 2004). Continuous recordings were analyzed
in word-locked −200 to 800 ms epochs. Signals exceeding ±80 μV
in any given epoch were automatically discarded.
In the learning phase, the ERP analyses focused on three pre-
defined time windows represented by the following components:
P2, N400, and LPC, with the onset being the presentation of new
words. We performed a peak detection that time-locked at 200–
300 ms and found that the amplitude reached a peak at 216 ms on
the grand-average ERP (e.g. see Salillas et al. 2008 for determining
a specific window of interest by peak detection). Therefore, we
defined the P2 window as 200–250 ms. For the N400, we also ran a
peak detection that was time-locked at 300–500 ms and found that
the peak occurred at 310 ms. As such, we identified the 250–350
window as the N400. In the same way, the LPC time window was
defined in 500–630 ms. Moreover, we conducted a Mass Univariate
Analysis (Groppe et al. 2011) to objectively identify the time win-
dow of significant effects and their corresponding topography. The
results confirmed that the N400 window ranged from 200–400 ms
and the P2 window ranged from 200–600 ms. The N400 is known
to be implicated in semantic processing and L2 lexical learning
and memory (Chen et al. 2014; Kutas and Federmeier 2011; Wu
and Thierry 2012), whereas the P2 is reflective of attentional
recognition. The LPC was analyzed between 400 and 800 ms, and
is typically associated with emotional stimuli processing (Kissler
et al. 2009; Schacht and Sommer 2009) and other processes such
as attentional capture, evaluation, and memory encoding (Lang
and Bradley 2007; Ye et al. 2019; Chou et al. 2020).
We focused our analyzes of the P2, N400, and LPC effects over
the central-parietal regions (sensors: F1, F2, Fz, F3, F4, FC1, FC2,
FCz, FC3, FC4, C1, C2, Cz, C3, C4, CP1, CP2, CPz, CP3, CP4 electrode
sites) because the central-parietal region has been implicated as
a reliable source in previous research (Kissler et al. 2009; Herbert
et al. 2011b, Herbert et al. 2011a; Granena and Long 2013).
Data from the first two trials of each block, incorrect responses,
and any trials contaminated by artifacts were removed from
the analyses. This resulted in the exclusion of 0.4% of the data.
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

Shen et al. |5
For each time window, we conducted a generalized linear mixed
model using companionship types and learning types as fixed
effects, and participants as a random effect. If a main effect
or interaction reached significance at P< 0.05 (FDR corrected),
follow-up analyses were performed.
Interbrain synchronization data
We used PLV to estimate the synchronization of the participants’
brain signals (Tognoli et al. 2007; Pérez et al. 2017; Liu et al. 2019)
for two frequency bands (1–3 Hz delta and 4–7 Hz theta) at each
time point between each pair of participants. Given that each
participant in the dyads had 40 effective electrodes (40 ×40), there
were 1600 possible combinations of electrode pairs. For example,
in the Cz-F2 pair, the PLV was calculated using the Cz from the co-
learner and the F2 from the co-learner/monitor. The calculation
formula of the PLV is as:
PLVi,k = 1
N





N

t=1
expj(φ(t)−φk(t))




In the formula, i and k represent the EEG signals of Cz (co-
learner) and F2 (monitor or co-learner), N represents the sampling
points of time window,and ϕ is the phase of the trial starting from
time t at channel i. When the PLV value is equal to 1, the EEG
signals of the two channels are perfectly phase-locked with each
other. Conversely, if the value is 0, they are not synchronized at
all. A 200 ms overlapping moving time window was used in a time
series of 1000 ms.
We conducted a linear mixed-effects (lmer) model to com-
pare the PLV of experimental data with companionship types
(monitored vs. co-learning companionship) and learning types
(lexical learning vs. semantic learning) as fixed effects. The
PLV values from the combination of the electrode pairs in each
condition (e.g. monitored companionship-lexical learning: FCz-
FC2, monitored companionship-semantic learning: C1-Cz, co-
learning companionship-lexical learning: C1-C2, co-learning
companionship-semantic learning: Cz-F1) were entered into the
lme model. When the models did not converge, we removed
the slope that explained the least amount of variance until
they converged. Parameters were estimated with the restricted
maximum likelihood approach, and the results from the best-
fitting model representing the most significant times of electrode
pair combinations are reported.
Results
Behavioral results of monitored vs. co-learning
companionship: There is higher accuracy of
semantic learning relative to lexical learning in
co-learning companionship
Table 2 shows the behavioral results of the testing phase. As for
the RTs, the results of the mixed-effect models of learning type
(semantic learning, lexical learning) ×companionship type (mon-
itored companionship, co-learning companionship) showed a sig-
nificant main fixed effect of companionship type,such that faster
RTs were elicited in the co-learning companionship condition
(M=1343 ±129 ms) compared to the monitored companionship
condition (M=2994 ±524 ms). There was no other main effect or
significant interaction found in the RT analyses.
With respect to accuracy, there was a significant interaction
between learning type and companionship type, demonstrating
higher accuracy of semantic learning (M=0.68 ±0.49) compared
Tab l e 2 . Results of the mixed-effects model analysis performed
on RT and accuracy data in the testing phase.
Predictor bSE t/z p
RT
Learning types .02 .04 .49 .626
Companionship types −.94 .07 −12.80 < 0.001
Learning types ×
Companionship types
.07 .08 .88 .382
Random effects Participants 7.49 .07
Accuracy
Learning types −.04 .11 −.37 .708
Companionship types .10 .11 .91 .361
Learning types ×
Companionship types
−.32 .16 −1.99 .047
Random effects Participants .66 .10
Note. Significant results (P <0.05, FDR corrected) are bolded.
to lexical learning (M=0.60 ±0.47) in the co-learning companion-
ship condition, b=0.3573, SE =0.113, z=3.167, P=0.0015. These
differences were not observed in the monitored companionship
condition (lexical: M=0.65 ±0.48, semantic: M=0.66 ±0.48,
b=0.0418, SE =0.112, z=0.374, P=0.708) (see Fig. 2a).
Electrophysiological results of the learning
phases: Lexical learning in a new language
elicits more attention compared to semantic
learning in co-learning companionship
We report the electrophysiological results of the learners
here (see Fig. 2 for results of monitors and co-learners). As
shown in Table 3, the mixed-effects model on locked ERP
components of learning type (lexical learning, semantic learn-
ing) ×companionship type (monitored companionship, co-
learning companionship) showed a significant main fixed
effect of learning type on P2, showing a more pronounced
P2 effect during lexical learning (M=3.62 ±8.73 μV) than in
semantic learning (M=1.11 ±9.36 μV). However, there was
a reversed learning type effect on N400 (semantic learning:
M= −1.56 ±9.28 μV > lexical learning: M= −.37 ±8.53 μV) and
LPC (semantic learning: M=3.15 ±8.83 μV>lexical learning:
M=0.84 ±8.30 μV). A main fixed effect of companionship type
was only found on LPC, indicating a greater LPC effect in the co-
learning companionship condition (M=2.42 ±8.57 μV) relative to
the monitored companionship condition (M=1.58 ±8.70 μV).
There was a significant interaction between learning type and
companionship type on P2, which revealed a larger P2 effect
for lexical learning (M=4.49 ±8.73 μV) relative to semantic
learning (M=1.25 ±9.83 μV) in both co-learning companionship
(b= −3.26, SE =0.621, z= −5.249, P<0.0001) and monitored
companionship (lexical learning: M=2.78 ±8.64 μV > semantic
learning: M=0.98 ±8.89 μV, b= −1.77, SE =0.614, z= −2.89,
P=0.0056) (see Fig. 2b, c). An independent samples t-test revealed
that lexical learning elicited a larger P2 effect than semantic
learning in the co-learning companionship condition compared
to the monitored companionship condition (co-learning com-
panionship: M= −3.55 ±3.08 μV > monitored companionship:
M=1.82 ±2.80 μV, t=2.12, SE =0.816, F=0.075, P=0.039).
Moreover, the P2 effect showed a significant interaction of
learning type × companionship type only for lexical learning:
there was a greater P2 effect in co-learning companion-
ship (M=4.49 ±8.73 μV) relative to monitored companionship
(M=2.78 ±8.64 μV) for lexical learning (b= −1.71, SE =0.538,
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

6|Cerebral Cortex, 2024, Vol. 34, No. 7
Fig. 2. The influence of companionship on word learning in a new language. Behavioralaccuracy of learners in monitored and co-learning companionship
(a), along with waveforms (b) and violin plots of amplitude (c) for learners. Notes: The dots represent mean accuracy or mean amplitude in violin f igures.
The boxes represent the quartiles (75% and 25%). The appearance of ∗∗ in the figure indicates a significant difference between their corresponding
conditions.
Tab l e 3 . Results of the mixed-effects model analysis performed on ERP data in the learning phase.
Predictor bSE t p
P2
Learning types 2.04 .38 6.68 < 0.001
Companionship types 1.00 .48 2.08 .052
Learning types × Companionship types 1.52 .76 2.00 .042
Random effects Participants 2.35 .67
N400
Learning types 1.22 .38 3.22 .001
Companionship types .40 .54 .75 .465
Learning types × Companionship types 1.17 .76 1.54 .123
Random effects Participants −.92 .58
LPC
Learning types −2.29 .49 −4.66 < 0.001
Companionship types .82 .37 2.21 .028
Learning types × Companionship types .86 .75 1.16 .247
Random effects Participants 1.98 .41
Note. Significant results (P <0.05, FDR corrected) are bolded.
z=
−3.183, P=0.0015), but not for semantic learning (co-
learning companionship: M=1.25 ±9.83 μV, m o n i t o r e d c o m -
panionship: M=0.98 ±8.89 μV, b= −.229, SE =0.536, z= −.427,
P=0.6698).
Results of the representation similarity analysis
in learning phases: Co-learners in co-learning
companionship exhibit a significantly higher
representation similarity in the P2 time window
As shown in Fig. 3, there was a significant representation
similarity between co-learners and co-learners in the P2 time
window (200–250 ms, r > 0.6, P=0.004), yet there was low
representation similarity between mo-learners and monitors
(200–250 ms, 0<r <0.5, P=0.156). An independent samples
t-test revealed that the representation similarity values of the co-
learning companionship were higher than those of the monitored
companionship in the 200–250 ms time window (co-learning
companionship: M=0.72 ±0.03 >monitored companionship:
M=0.31 ±0.07, t= − 5.10, SE =0.416, F=4.60, P < 0.001). The P2
time window showed a higher similarity representation between
co-learners (i.e. co-learning companionship) for both semantic
and lexical learning compared to monitored companionship.
These findings provide robust evidence for the inter-brain
synchronization of learning words in a new language across
different companionships. However, there was low representa-
tion similarity of monitored companionship (mo-learners and
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

Shen et al. |7
Fig. 3. Temporal based representation dissimilarity matrix and RSA of monitored companionship (mo-learner, monitor) and co-learning companionship
(co-learner, co-learner). Notes:Weobtained4×4 RDMs for different social roles in P2 time window. In these RDMs, every two cells represent a condition,
with rows and columns indexing the semantic learning and lexical learning conditions. The RDMs of mo-learner (a), monitor (b), co-learner (c, d). The
mo-learner, monitor, and co-learner dissimilarity values are from 0 to 0.8, with values approaching 0 indicating high similar neural patterns, and a
value of 0.8 indicating low similarity. Representational dissimilarity matrices of the monitored condition (mo-learner and monitor) and the co-learning
condition (co-learner and co-learner) used to calculate the similarities neural representations (e). Shading around solid lines indicates SE. Cluster-
corrected p values are shown below each time course.
monitors: 0<r < 0.48, P=0.622) and co-learning companionship
(co-learners and co-learners: 0 < r< 0.54, P=0.208) in the N400
time window (280–340 ms). No significant similarity was found
in the LPC time window (500–630 ms, mo-learners and monitors:
0 <r< 0.54, P=0.622; co-learners and co-learners: 0<r<0.58,
P=0.265) (see Supplementary Fig. 2).
Interbrain synchronization results of the
learning phase: Semantic learning elicits higher
delta and theta synchronization than lexical
learning in co-learning companionship, which
was concentrated in the right frontal and right
central neural regions of co-learners.
Delta synchronization (1–3 Hz).
There was a significant main effect for companionship
in which delta synchronization was higher in co-learning
companionship (M=0.70 ±0.11) relative to monitored compan-
ionship (M=0.60 ±0.15) in the FC3-F2 electrode pairs (b=0.09,
SE =0.02, t=4.08, P<0.001) (see Table 6). Similar results were
obtained for the following electrode pairs: Cz-F2 (co-learning
companionship: M=0.70 ±0.12 >monitored companionship:
M=0.65 ±0.13; b=0.05, SE =0.02, t=2.22, P=0.029), F2-F2
(co-learning companionship: M=0.66 ±0.12 > monitored com-
panionship: M=0.61 ±0.14; b=0.05, SE =0.02, t=2.08, P=0.043),
F2-F4 (co-learning companionship: M=0.64 ±0.13 >monitored
companionship: M=0.59 ±0.14; b=0.05, SE =0.02, t=2.17,
P=0.035), Fz-F2 (co-learning companionship: M= 0.68 ±0.12 >
monitored companionship: M=0.62 ±0.13; b=0.05, SE =0.02,
t=2.23, P=0.029), FC1-F2 (co-learning companionship: M=0.68 ±
0.12 > monitored companionship: M=0.62 ±0.13; b=0.06,
SE =0.02, t=2.59, P=0.011), FC3-C4 (co-learning companionship:
M=0.68 ±0.13 >monitored companionship: M=0.60 ±0.15;
b=0.08, SE =0.02, t= 2.94, P=0.006), FC3-FC2 (co-learning
companionship: M=0.67 ±0.15 > monitored companionship:
M=0.60 ±0.16; b=0.07, SE =0.03, t=2.67, P=0.009), FCz-F2
(co-learning companionship: M=0.68 ±0.12 >monitored com-
panionship: M=0.62 ± 0.14; b=0.06, SE =0.02, t=2.33, P=0.023).
A significant interaction between learning type × com-
panionship type revealed that in the C4-F1 electrode pairs,
delta synchronization in the semantic learning (M=0.69 ±0.12)
was higher than lexical learning (M=0.62 ±0.11) in the co-
learning companionship (b=0.07, SE =0.04, t=2.04, P=0.046).
This difference did not emerge in the monitored companionship
condition (semantic learning: M= −1.56 ±9.28; lexical learning:
M= −.37 ±8.53, b= −.05, SE =0.04, t= −1.30, P=0.200) (see
Fig. 4a, b, c). The same results were obtained for the following
electrode pairs: C4-F2 (semantic learning: M=0.69±0.10 >lexical
learning: M=0.64 ±0.12; b=0.05, SE =0.03, t=1.27, P=0.008),
F1-C2 (semantic learning: M=0.68 ±0.14 >lexical learning:
M=0.58 ±0.13; b=0.10, SE =0.03, t=2.97, P=0.004), F1-F4
(semantic learning: M=0.71 ±0.14 >lexical learning: M=0.61 ±
0.12; b=0.10, SE =0.03, t=3.04, P=0.003), F2-C2 (semantic learn-
ing: M=0.67 ±0.14 > lexical learning: M=0.56 ±0.12; b=0.11,
SE =0.04, t=3.07, P=0.003), F4-C2 (semantic learning: M=0.65 ±
0.14 >lexical learning: M=0.56 ±0.14; b=0.09, SE =0.04, t=2.24,
P=0.028), FC4-C2 (semantic learning: M=0.63 ±0.15 >lexical
learning: M=0.56 ±0.14; b=0.07, SE =0.04, t= 2.02, P= 0.047)
(see Supplementary Table 6).
Theta synchronization (4–7 Hz)
Theta synchronization was higher in co-learning compan-
ionship (M=0.59 ±0.11) relative to monitored companionship
(M=0.55 ±0.11) in the C4-Cz electrode pairs (b=0.04, SE =0.02,
t=2.09, P=0.040). The same results were obtained for the
following electrode pairs: Cz-Cz (co-learning companionship:
M=0.59 ±0.11 > monitored companionship: M=0.55 ±0.10;
b=0.04, SE =0.02, t=2.0, P=0.049), F1-Cz (co-learning compan-
ionship: M=0.60 ±0.12 >monitored companionship: M=0.54 ±
0.12; b=0.06, SE =0.03, t=2.28, P=0.031), F2-Cz (co-learning
companionship: M=0.59 ±0.12 >monitored companionship:
M=0.54 ±0.11; b=0.05, SE =0.02, t=2.24, P=0.028), F3-Cz
(co-learning companionship: M=0.58 ±0.11 > monitored com-
panionship: M=0.53 ±0.11; b=0.06, SE =0.02, t=2.68, P=0.008),
F4-Cz (co-learning companionship: M=0.59 ±0.12 >monitored
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

8|Cerebral Cortex, 2024, Vol. 34, No. 7
Fig. 4. Interbrain delta and theta synchronization and coupled electrode pairs during the learning phase. Note: (a, d) Interbrain delta (a) and theta
(d) synchronization obtained using a 200 ms sliding window during learning phase. (b, e) Average interbrain delta (b) and theta (e) synchronization
experimental data during word learning provided for each condition. (c, f) Significantly coupled electrode pairs, determined by the PLV analysis with
FDR correction (P <0.05). (g, h) The same results are indicated in the matrix. Interbrain coupling which increased for the co-learning condition (co-
learner and co-learner) (g) and for the monitored condition (mo-learner and monitor) (h). In the violin plot, the dots represent mean amplitude. The
boxes represent the quartiles (75% and 25%). The ∗∗ in the figure indicates a significant difference between their corresponding conditions.
companionship: M=0.52 ±0.10; b=0.07, SE =0.02, t=3.33,
P=0.001), Fz-C2 (co-learning companionship: M=0.60 ±0.11 >
monitored companionship: M=0.56 ±0.13; b=0.04, SE =0.02,
t=2.00, P=0.049), FC1-Cz (co-learning companionship: M=0.58 ±
0.11 > monitored companionship: M=0.54 ±0.12; b=0.04,
SE =0.02, t=2.07, P=0.042), FC4-C2 (co-learning companionship:
M=0.59 ± 0.12 > monitored companionship: M=0.55 ±0.12;
b=0.05, SE =0.02, t=2.18, P=0.032) (see Supplementary Table 7).
Figure 4(d, e, f) shows the interaction between learning type
× companionship in the FC1-F1 electrode pairs as reflected by
higher theta synchronization in semantic learning (M=0.60 ±0.12)
compared to lexical learning (M=0.54 ±0.12) in the monitored
companionship condition (b=0.06, SE =0.03, t=2.04, P=0.0462).
This difference did not emerge in the co-learning compan-
ionship (semantic learning: M=0.55 ±0.12, lexical learning:
M=0.60 ±0.11; b= −.05, SE =0.03, t=−1.69, P=0.09).
Discussion
The present study examined how monitored and co-learning com-
panionship affect word learning in a new language. As reflected
by a larger P2 effect, co-learning companionship had a greater
influence on lexical learning than on semantic learning. More-
over, the RSA revealed a significant influence of co-learning com-
panionship on word learning, such that among the dyads, there
was a higher representation similarity compared to monitored
companionship. These findings imply that there is a difference in
the neural representation between monitored companionship
(mo-learner/monitor) and co-learning companionship (co-
learner/co-learner). More important, all dyads showed higher
delta and theta synchronization in semantic learning relative
to lexical learning while in the co-learning companionship,
which also corresponded to higher accuracy in semantic learning
than lexical learning. Overall, these findings suggest that co-
learning companionship affects early stages of new language
word learning and that representation similarity and interbrain
synchronization analyses can serve as an underlying neural
indicator of the influence of companionship on new language
learning. We elaborate on these findings in the next sections.
Co-learning companionship affects early stages
of word learning
We found that lexical learning elicited larger P2 effects than
semantic learning in the co-learning companionship, but not in
monitored companionship. This finding supports our hypothesis
that in lexical learning, it is more difficult to integrate semantics,
which can be reflected by attenuated N400, LPC, and lower accu-
racy in lexical learning relative to semantic learning.1 Therefore, it
may be the case that the presence of co-learners fosters empathy
and a collaborative atmosphere, consequently improving co-
learners’ attention. Greater P2 amplitude is associated with early
recognition of a target stimulus (Luck and Hillyard 1994)(Luyster
et al. 2008; Lu et al. 2018). In the current study, co-learners allo-
cated more attention to learning words through association with
meanings in their native language during early stages of word
learning, showing an increased P2 effect with lower accuracy for
lexical learning compared to semantic learning. Taken together,
these findings suggest that co-learning companionship may
influence the early stages of word learning through attention.
In addition, it should be noted that our testing phase was
based on a comparison of learners’ performance according to
the type of learning (semantic and lexical). Therefore, the pos-
sible differences between the category judgment task and the
1 The lexical and semantic tests were presented in different task forms
(letter vs. category judgment task). However, the results of testing phase
showed that there was no difference between the two learning types in the
monitored condition, indicating that the testing form per se did not result
in accuracy difference. Higher accuracy in the semantic learning relative to
lexical learning in co-learning companionship was indeed derived from social
interaction companionship. Future research could match the testing task form
to validate these results.
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

Shen et al. |9
letter judgment task in the word learning tests merits further
exploration.
Representation similarity and interbrain
synchronization demonstrate neural effects of
companionship on word learning
To reveal how companionship affects word learning in a new
language, we conducted representation similarity and interbrain
synchronization analyses. First, we explored the representation
similarity in monitored companionship (mo-learner and monitor)
and co-learning companionship (co-learners and co-learners).
The RSA showed a higher similarity of neural representations
between co-learners in co-learning companionship for the P2 time
window (i.e. the attentional processing phase) compared to mon-
itored companionship. This finding indicates that the representa-
tion of neural responses in the initial attention stage of learning
is similar. However, these similarities did not emerge in the N400
and LPC time windows. The possible reason for this result is that
monitored companionship may distract and hamper mo-learners’
attention when monitored by another individual (Kirschner and
Erkens 2013). We found a high representation similarity between
co-learners in co-learning companionship, indicating that word
learning is affected by co-learning companionship in the initial
attention stage. Our research has shown that co-learning com-
panionship elicits more empathy than when in the presence of
others. Co-learners can provide more emotional support to each
other (Zajonc 1965) and can reduce the stress and anxiety of
learners during learning and memory (Dillenbourg 1999). The
high representation similarity between co-learners in the early
attention stage provides valid neural evidence that co-learning
companionship promotes the formation of joint attention in the
early learning stage among co-learners.
Next, we used PLVs to further examine the effects of com-
panionship on word learning. In line with our hypothesis that co-
learning companionship is an effective approach which facilitates
language learning, we found that semantic learning elicited
higher delta and theta synchronization than lexical learning.
Increased delta oscillations are linked to the improvement of inte-
gration and storage of semantics (Bastiaansen and Hagoort 2003;
Voytek et al. 2015), and increased theta oscillations represent
strengthened semantic processing (Major et al. 2004). Our results
indicated that each dyad had deeper semantic processing in co-
learning companionship. We also found that in the monitored
companionship condition, theta synchronization was higher for
semantic learning than for lexical learning. Theta synchroniza-
tion plays an important role in semantic access and attention
(Kahana 1996; Klimesch 1999). Thus, such synchronization may
have been enhanced. Theseresults suggest that semantic learning
may be independent of co-learning or monitored companionship.
However, the delta and theta synchronization observed during
co-learning companionship were higher than in monitored
companionship. Learners also had higher accuracy of seman-
tic learning compared to lexical learning during co-learning
companionship.
Our results can be interpreted through Social Cognitive The-
ory (Nregueuela-Azarola 2012). That is, even the smallest social
connection with another person increases social arousal (Walton
et al. 2012) that may consequently enhance dyads’ joint attention
in co-learning companionship (Luyster et al. 2008; Grubbs et al.
2020). In addition, the dual brain map showed that electrode pairs
with higher theta synchronization in the co-learning condition
were concentrated in the right frontal and right central neural
regions of co-learners. These results are consistent with previ-
ous studies reporting that the degree of individual engagement
and empathic behavior correlates with right frontal–frontal and
central-to-central synchronization (Kruppa et al. 2021), and that
frontal–frontal synchronization is linked to perceived similarity
between partners (Hu et al. 2017). Empathy is a higher-order
ability known to sustain inter-brain synchrony (Schwartz et al.
2022). We believe that in the co-learning companionship con-
dition, the high representation similarity of dual-brain learning
in co-learners supports the notion that co-learners have more
similar psychological perceptions and empathy, which promotes
the synchronization of the right frontal and central brain regions
in co-learners. Altogether, our findings echo previous views high-
lighting the importance of companionship in learning (Philp et al.
2014). However, our results uniquely offer a neural basis of how
types of companionship differentially influence word learning in
a new language.
Conclusion
This study has revealed an influence of social companionship
on learning words in a new language. The findings from the
ERP analyses showed that co-learning companionship influences
early attention in lexical learning, and the results of the RSA
provide robust evidence for the inter-brain synchronization of
learning words in a new language during co-learning compan-
ionship. Interbrain synchronization analyses between co-learners
indicated that co-learning companionship significantly improved
semantic learning. In all, our findings offer important implica-
tions for language researchers and educators in understanding
how individuals learn words in a new language and how different
types of companionship can affect learning outcomes.
Author contributions
Yujing Shen (Writing—original draft, Methodology, Investigation,
Formal Analysis, Visualization), Xu Liu (Methodology, Formal
Analysis), Yingyi Xiang (Investigation, Formal Analysis), John W.
Schwieter (Writing—review & editing), Huanhuan Liu (Concep-
tualization, Methodology, Writing—review & editing, Supervision,
Visualization).
Supplementary material
Supplementary material is available at Cerebral Cortex online.
Funding
This work was supported by a Grant from STI 2030—Major
Projects 2021ZD0200500, General Program of National Natural
Science Foundation of China (32371089), Liaoning Social Science
Planning Fund of China (L20AYY001), Dalian Science and Technol-
ogy Star Fund of China (2020RQ055), and Youth Project of Liaoning
Provincial Department of Education (LJKQZ2021089), Research
and Cooperation Projects on Social and Economic Development
of Liaoning Province (2024lslybhzkt-17), and Liaoning Educational
Science Planning Project (JG21DB306).
Conflict of interest statement: None declared.
Data availability
Data will be made available on request.
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

10 |Cerebral Cortex, 2024, Vol. 34, No. 7
Declaration of competing interest
No potential conflict of interest was reported by the authors.
References
Allan D. Oxford placement test 2: test pack, Oxford, England: Oxford
University Press; 2004.
Baldwin DA. Understanding the link between joint attention and
language. In: Moore C, Dunham PJ, Dunham P, (eds), Joint attention.
New York: Psychology Press; 2014. pp. 131–158.
Bastiaansen M, Hagoort P. Event-induced theta responses as a win-
dow on the dynamics of memory. Trends Cogn Sci. 2003:7(12):
481–488.
Bastiaansen M, Hagoort P. Frequency-based segregation of syntactic
and semantic unification during online sentence level language
comprehension. J Cogn Neurosci. 2015:27(11):2095–2107. https://
doi.org/10.1162/jocn_a_00829.
Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects
models using lme4. J Stat Softw. 2014:67(1):133–199.
Bolt NK, Loehr JD. The auditory P2 differentiates self-from
partner-produced sounds during joint action: contributions
of self-specific attenuation and temporal orienting of atten-
tion. Neuropsychologia. 2023:182:108526. https://doi.org/10.1016/j.
neuropsychologia.2023.108526.
Bowlby J. Attachment and loss: retrospect and prospect. Am J Orthop.
1982:52(4):664–678. https://doi.org/10.1111/j.1939-0025.1982.
tb01456.x.
Chen Y, Zhong Y, Zhou H, Zhang S, Tan Q, Fan W. Evidence
for implicit self-positivity bias: an event-related brain potential
study. Exp Brain Res. 2014:232(3):985–994. https://doi.org/10.1007/
s00221-013- 3810-z.
Chomsky N, Halle M. The sound pattern of English; Cambridge, Mass:
Noam Chomsky and Morris Halle, 1968.
Chou LC, Pan YL, Lee CL. Emotion anticipation induces emo-
tion effects in neutral words during sentence reading: evi-
dence from event-related potentials. Cognitive, Affective, & Behav-
ioral Neuroscience. 2020:20(6):1294–1308. https://doi.org/10.3758/
s13415-020- 00835-z.
Collins R. (ed), Interaction ritual chains. In: Interaction Ritual Chains.
Princeton: Princeton University Press; 2014.
Delorme A, Makeig S. EEGLAB: an open source toolbox for anal-
ysis of single-trial EEG dynamics including independent com-
ponent analysis. J Neurosci Methods. 2004:134(1):9–21. https://doi.
org/10.1016/j.jneumeth.2003.10.009.
Dillenbourg P. What do you mean by collaborative learning?
In: Collaborative-learning: cognitive and computational approaches.
Oxford, United Kingdom: Pierre Dillenbourg, 1999. pp. 1–19.
Dindar M, Akcay B. The effect of peer presence on students’ task
involvement and self-regulation strategies. Journal of Educational
Research. 2015:108(5):423–433.
Elgort I, Perfetti CA, Rickles B, Stafura JZ. Contextual learning
of L2 word meanings: second language proficiency modulates
behavioural and event-related brain potential (ERP) indicators of
learning. Language, Cognition and Neuroscience. 2015:30(5):506–528.
https://doi.org/10.1080/23273798.2014.942673.
Feinberg JM, Aiello JR. The effect of challenge and threat appraisals
under evaluative presence. J Appl Soc Psychol. 2010:40(8):
2071–2104. https://doi.org/10.1111/j.1559-1816.2010.00651.x.
García-Gámez AB,Macizo P. Lexical and semantic training to acquire
words in a foreign language: an electrophysiological study. Bilin-
gualism: Language and Cognition. 2022:25(5):768–785. https://doi.
org/10.1017/S1366728921000456.
Gass SM, Mackey A. Input, interaction, and output in second
language acquisition. In: VanPatten B, Williams J, (eds), Theo-
ries in second language acquisition. New York: Routledge; 2014.
pp. 194–220.
Granena G, Long MH. Age of onset, length of residence, lan-
guage aptitude, and ultimate L2 attainment in three linguis-
tic domains. Second Lang Res. 2013:29(3):311–343. https://doi.
org/10.1177/0267658312461497.
Groppe DM, Urbach TP, Kutas M. Mass univariate analysis of
event-related brain potentials/fields I: a critical tutorial review.
Psychophysiology. 2011:48(12):1711–1725. https://doi.org/10.1111/
j.1469-8986.2011.01273.x.
Grubbs V, Fortier-Gauthier U, Shultz S. Oscillatory dynamics of
joint attention networks: the role of theta rhythm. Front Psychol.
2020:11:2410.
Hadot P. Philosophy as a way of life: spiritual exercises from Socrates to
Foucault; Hoboken: Pierre Hadot, 1995.
Hall-McMaster S, Muhle-Karbe PS, Myers NE, Stokes MG.Reward
boosts neural coding of task rules to optimize cognitive flex-
ibility. J Neurosci. 2019:39(43):8549–8561. https://doi.org/10.1523/
JNEUROSCI.0631-19.2019.
Hauk O, Davis MH, Ford M, Pulvermüller F, Marslen-Wilson WD.
The time course of visual word recognition as revealed by
linear regression analysis of ERP data. NeuroImage. 2006:30(4):
1383–1400. https://doi.org/10.1016/j.neuroimage.2005.11.048.
Herbert C, Herbert BM, Ethofer T, Pauli P. His or mine?
The time course of self–other discrimination in emotion
processing. Soc Neurosci. 2011a:6(3):277–288. https://doi.
org/10.1080/17470919.2010.523543.
Herbert C, Pauli P, Herbert BM. Self-reference modulates the pro-
cessing of emotionalstimuli in the absence of explicitselfrefer-
ential appraisal instructions. Soc Cogn Affect Neurosci. 2011b:6(5):
653–661. https://doi.org/10.1093/scan/nsq082.
Hernandez LM, Kim Y, Bowman LC. Longitudinal changes in infants’
theta and alpha power during sustained joint attention predicts
emerging social abilities. Developmental Cognitive Neuroscience.
2021:48:100951. https://doi.org/10.1016/j.dcn.2021.100951.
Hu Y, Hu Y, Li X, Pan Y, Cheng X. Brain-to-brain synchroniza-
tion across two persons predicts mutual prosociality. Soc Cogn
Affect Neurosci. 2017:12(12):1835–1844. https://doi.org/10.1093/
scan/nsx118.
Kahana MJ. Theta synchrony and REM sleep in the human hip-
pocampus: a local field potential study. J Neurosci. 1996:16(12):
3966–3972.
Kelley K, Kelley MK,& Imports MASS (2018). The MBESS R
package. Computer software. MBESS. Retrieved from: https://
CRAN.R-project.org/package=MBESS.
Kinginger C. Learning the pragmatics of solidarity in the networked
foreign language classroom. In: Hall JK, Verplaetse LS, (eds), Sec-
ond and foreign language learning through classroom interaction.New
York,2000:23–46.
Kirschner PA, Erkens G. Toward dynamic and constructive
alignment between teachers’ and students’ activities in
computer-based learning environments. Instr Sci. 2013:41(1):
159–173.
Kissler J, Herbert C, Winkler I, Junghofer M. Emotion and attention
in visual word processing—an ERP study. Biol Psychol. 2009:80(1):
75–83. https://doi.org/10.1016/j.biopsycho.2008.03.004.
Klimesch W. EEG alpha and theta oscillations reflect cognitive
and memory performance: a review and analysis. Brain Res
Rev. 1999:29(2–3):169–195. https://doi.org/10.1016/S0165-0173(98
)00056-3.
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

Shen et al. |11
Kriegeskorte N, Mur M, Bandettini PA. Representational similar-
ity analysis-connecting the branches of systems neuroscience.
Front Syst Neurosci. 2008:2:1662–5137. https://doi.org/10.3389/
neuro.06.004.2008.
Kroll JF, Stewart E. Category interference in translation and picture
naming: evidence for asymmetric connections between bilingual
memory representations. J Mem Lang. 1994:33(2):149–174. https://
doi.org/10.1006/jmla.1994.1008.
Kruppa JA, Reindl V, Gerloff C, Oberwelland Weiss E, Prinz J, Her-
pertz-Dahlmann B,Konrad K,Schulte-Rüther M. Brain and motor
synchrony in children and adolescents with ASD—a fNIRS hyper-
scanning study. Soc Cogn Affect Neurosci. 2021:16(1–2):103–116.
https://doi.org/10.1093/scan/nsaa092.
Kummerer D, Hartwigsen G, Kellmeyer P, Glauche V, Mader I,
Kloppel S, Weiller C. Damage to ventral and dorsal language
pathways in acute aphasia. Brain. 2013:136(2):619–629. https://
doi.org/10.1093/brain/aws354.
Kutas M,Federmeier KD.Thirty years and counting: finding meaning
in the N400 component of the event-related brain potential (ERP).
Annu Rev Psychol. 2011:62(1):621–647. https://doi.org/10.1146/
annurev.psych.093008.131123.
Lang P, Bradley MM. The international affective picture system
(IAPS) in the study of emotion and attention. Handbook of emotion
elicitation and assessment. 2007:29:70–73.
Li P, Jeong H. The social brain of language: grounding second lan-
guage learning in social interaction. Science of Learning. 2020:5(1):8.
https://doi.org/10.1038/s41539-020- 0068-7.
Liu Y, van Hell JG. Learning novel word meanings: an ERP study
on lexical consolidation in monolingual, inexperienced foreign
language learners. Lang Learn. 2020:70(S2):45–74. https://doi.
org/10.1111/lang.12403.
Liu H, Kong C, de Bruin A, Wu J, He Y. Interactive influence
of self and other language behaviors: Evidence fromswitching
between bilingual production and comprehension. Hum Brain
Mapp. 2020:41(13):3720–3736. https://doi.org/10.1002/hbm.25044.
Liu H, Zhang M, Pérez A, Xie N, Li B, Liu Q. Role of language con-
trol during interbrain phase synchronization of cross-language
communication. Neuropsychologia. 2019:131:316–324. https://doi.
org/10.1016/j.neuropsychologia.2019.05.014.
Lu Z, Ku Y. NeuroRA: a python toolbox of representational anal-
ysis from multi-modal neural data. Frontiers in neuroinformatics.
2020:14:1662–5196. https://doi.org/10.3389/fninf.2020.563669.
Lu H,Guo T,Guo C,Huang X. EEG-based concomitants of joint atten-
tion: a systematic review and meta-analysis. Neurosci Biobehav
Rev. 2018:90:117–130.
Luck SJ, Hillyard SA. Spatial filtering during visual search: evidence
from human electrophysiology. J Exp Psychol Hum Percept Perform.
1994:20(5):1000–1014.
Luyster RJ, Kadlec MB, Carter A, Tager-Flusberg H. Language assess-
ment and development in toddlers with autism spectrum dis-
orders. J Autism Dev Disord. 2008:38(8):1426–1438. https://doi.
org/10.1007/s10803-007-0510-1.
Lyt le SR ,Garcia-Sierra A,Kuhl PK. Two are better than one: infant lan-
guage learning from video improves in the presence of peers. Proc
Natl Acad Sci. 2018:115(40):9859–9866. https://doi.org/10.1073/
pnas.1611621115.
Major G,Larkum ME,Schiller J. Theta oscillations during active infor-
mation maintenance in a cortical network model. PLoS Comput
Biol. 2004:1(3):e100.
McLaughlin J, Osterhout L, Kim A. Neural correlates of second-
language word learning: minimal instruction produces rapid
change. Nat Neurosci. 2004:7(7):703–704. https://doi.org/10.1038/
nn1264.
Michel C, Stets M, Parise E, Reid VM, Striano T, Hoehl S. Theta-and
alpha-band EEG activity in response to eye gaze cues in early
infancy. NeuroImage. 2015:118:576–583. https://doi.org/10.1016/j.
neuroimage.2015.06.042.
Midgley KJ, Holcomb PJ, Grainger J. Language effects in second lan-
guage learners and proficient bilinguals investigated with event-
related potentials. J Neurolinguistics. 2009:22(3):281–300. https://
doi.org/10.1016/j.jneuroling.2008.08.001.
Mundy P, Fox N, Card J. EEG coherence, joint attention and language
development in the second year. Dev Sci. 2003:6(1):48–54. https://
doi.org/10.1111/1467-7687.00253.
Nregueuela-Azarola E. Sociocognitive perspectives on language
use and language learning (review). Can Mod Lang Rev. 2012:
68(2):239240.
Papeo L, Abassi E. Seeing social events: the visual specialization
for dyadic human–human interactions. J Exp Psychol Hum Per-
cept Perform. 2019:45(7):877–888. https://doi.org/10.1037/xhp000
0646.
Pérez A, Carreiras M, Duñabeitia J. Brain-to-brain entrainment:
EEG interbrain synchronization while speaking and listen-
ing. SciRep. 2017:7(1):1–12. https://doi.org/10.1038/s41598-017-
04464-4.
Philp J, Adams R, Iwashita N (eds). Peer interaction and sec-
ond language learning. New York: Rouledge, 2014. https://doi.
org/10.4324/9780203551349.
Salillas E, El Yagoubi R, Semenza C. Sensory and cognitive pro-
cesses of shifts of spatial attention induced by numbers: an
ERP study. Cortex. 2008:44(4):406–413. https://doi.org/10.1016/j.
cortex.2007.08.006.
Schacht A, Sommer W. Emotions in word and face processing:
early and late cortical responses. Brain Cogn. 2009:69(3):538–550.
https://doi.org/10.1016/j.bandc.2008.11.005.
Schwartz L, Levy J, Endevelt-Shapira Y, Djalovski A, Hayut O,
Dumas G, Feldman R. Technologically-assisted communica-
tion attenuates inter-brain synchrony. NeuroImage. 2022:
264:1053–8119. https://doi.org/10.1016/j.neuroimage.2022.119
677.
Skuballa IT, Xu KM, Jarodzka H. The impact of co-actors on cog-
nitive load: when the mere presence of others makes learning
more difficult. Comput Hum Behav. 2019:101:30–41. https://doi.
org/10.1016/j.chb.2019.06.016.
Smith MC, Magee LE. Tracing the time course of picture–word
processing. J Exp Psychol Gen. 1980:109(4):373–392. https://doi.
org/10.1037/0096-3445.109.4.373.
Steier R. Designing for joint attention and co-presence across parallel
realities. Manassas: Rolf Steier, 2020.
Tognoli E, Lagarde J, DeGuzman G, Kelso J. The phi com-
plex as a neuromarker of human social coordination. Proc
Natl Acad Sci. 2007:104(19):8190–8195. https://doi.org/10.1073/
pnas.0611453104.
Tomasello M. Joint attention as social cognition. In: Moore C,
Dunham PJ, editors. Joint attention: its origins and role in development.
Hillsdale, NJ: Erlbaum; 1995. pp. 103–130.
Van der Stel M, Veenman MV. Effects of peer inf luence on the
selection and use of self-regulated learning strategies. Br J Educ
Psychol. 2014:84(2):290–310.
Voyte k B, Kayser AS, Badre D, Fegen D, Chang EF, Crone NE,
Knight RT. Oscillatory dynamics coordinating human frontal net-
works in support of goal maintenance. Nat Neurosci. 2015:18(9):
1318–1324. https://doi.org/10.1038/nn.4071.
Walton GM, Cohen GL, Cwir D, Spencer SJ. Mere belonging: the
power of social connections. J Pers Soc Psychol. 2012:102(3):513–532.
https://doi.org/10.1037/a0025731.
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024

12 |Cerebral Cortex, 2024, Vol. 34, No. 7
Wu YJ , Thierry G. How reading in a second language protects your
heart. J Neurosci. 2012:32(19):6485–6489. https://doi.org/10.1523/
JNEUROSCI.6119-11.2012.
Ye C, Sun HJ, Xu Q, Liang T, Zhang Y, Liu Q. Working memory
capacity affects trade-off between quality and quantity only
when stimulus exposure duration is sufficient: evidence for the
two-phase model. Sci Rep. 2019:9(1):8727. https://doi.org/10.1038/
s41598-019- 44998-3.
Zajonc RB. Social facilitation. Science. 1965:149(3681):269–274.
https://doi.org/10.1126/science.149.3681.269.
Downloaded from https://academic.oup.com/cercor/article/34/7/bhae289/7714270 by Wilfrid Laurier University user on 18 August 2024