Académique Documents
Professionnel Documents
Culture Documents
PSYCHOSHORTS 2024
Keynote Lectures / Discours liminaires
Primary Keynote / Conférencière d'honneur
1
PSYCHOSHORTS 2024
Keynote Lectures / Discours liminaires
Group comparisons are logical and economical research designs. However, narrowing complex
and dynamic experiences to binary levels may unintentionally embed inequitable and unjust
inferences. In this presentation, I will discuss why seeking bilingual (dis)advantages should not
be the only research goal in bilingualism and cognition. Drawing on my transdisciplinary
research, I will demonstrate the affordances of examining bilingualism and cognition with
equity and justice for people in mind.
La comparaison entre des groupes définis est une approche de recherche logique et
économique. Cependant, le fait de réduire des expériences complexes et dynamiques à des
niveaux binaires peut involontairement mener à des conclusions inéquitables et injustes. Dans
cette présentation, j'expliquerai pourquoi la recherche de (dés)avantages bilingues ne devrait
pas être le seul objectif de la recherche sur le bilinguisme et la cognition. En m'appuyant sur
mes recherches transdisciplinaires, je souhaite montrer comment on peut se pencher sur le
bilinguisme et la cognition en gardant en tête l'équité et la justice pour tous les types de
locuteurs. (N.B.: la présentation sera livrée en anglais.)
2
PSYCHOSHORTS 2024
Keynote Lectures / Discours liminaires
Student Keynote / Conférencière étudiante invitée
3
PSYCHOSHORTS 2024
Keynote Lectures / Discours liminaires
Are production and processing alike after all? Examining the cognitive
effects of code-switching by constructing an ecologically valid bilingual
corpus
Leah Gosselin
Student Keynote
Leah Gosselin
Conférencière étudiante invitée
Dans de nombreuses populations bilingues stables - y compris dans la région de la capitale canadienne -
l'alternance codique est un comportement courant au sein d'un groupe. Bien que l'alternance codique soit décrite
par certains comme "la façon la plus confortable" de parler (Dorleijn, 2017 : 16), la recherche empirique continue
d'associer l'alternance codique, même lorsque habituelle, à des indices comportementaux et neurophysiologiques
liés à l'effort (ex. : Gosselin & Sabourin, 2021). Parallèlement, les personnes rapportant pratiquer fréquemment
l'alternance codique ont démontré des résultats associés à des capacités accrues d'inhibition (ex. : Gosselin et
Sabourin, 2023), ce qui laisse supposer que le l'alternance codique est en effet exigeante sur le plan cognitif. Quand
ce "coût de l'alternance codique" est-il constaté, et qu'indique-t-il ? Est-il raisonnable de comparer la production
linguistique dans la vie réelle à son traitement artificiel en laboratoire ? Cet exposé détaille la manière dont ma
thèse tente de démêler le paradoxe production-traitement dans la recherche portant sur l'alternance codique. En
bref, je présenterai FEBLOC (French-English Bilingual Loved Ones Corpus), un nouveau corpus construit dans le but
d'examiner le sujet à travers une lentille naturaliste. Je présenterai des données préliminaires soutenant le
découplage de la production et du traitement bilingues. (N.B.: la présentation sera livrée en anglais, mais la période
de questions sera bilingue.)
4
PSYCHOSHORTS 2024
Panel: Going beyond good intentions: making EDI considerations concrete
Table ronde: Au-delà des bonnes intentions: rendre concrètes les considérations EDI
8
language would increase the chances of co-activation (and competition) between the two grammars of early
bilingual speakers when reading these sentences. Future steps will include investigation based on other
languages (such as French, Chinese, etc.) and EEG data (such as ERPs, oscillations, etc.).
It is well known that visual information presented to the left visual field is projected into the right
hemisphere and information presented to the right visual field is projected into the left hemisphere. Crucially,
when English endocentric bimorphemic compounds are split into visual fields, the compound head is
projected directly onto the language-dominant left hemisphere— possibly with a head-start analysis by the
visual word form area (VWFA; Cohen et al., 2000).
Considering that the transfer between hemispheres is estimated to be in the magnitude of 10-40 ms
(Brown et al., 1998; Nowicka & Tacikowski, 2011), we manipulated the spatial and temporal presentation of
compound constituents in a novel variation of a temporal order judgment (TOJ) task.
Our study involved a TOJ and a lexical decision (LD) task that included elements of the TOJ paradigm. In
the LD task participants were asked to make word-nonword judgements to disjointly presented (60 ms apart)
segments of compounds (FOOTBALL), pseudo-compounds (CARPET), and unsegmentable monomorphemic
words (JINGLE). We manipulated both the spatial location of constituents (FOOT appears in the left visual
field with BALL in the right or vice-versa) and their temporal order (FOOT presented before BALL or BALL
presented before FOOT). We also manipulated the legality of the segments (FOOT+BALL or FOO+TBALL).
See Figure 1. In the TOJ task, segments were presented in rapid succession and participants had to judge
which segment appeared first. In both tasks, segments appeared for 60 ms and backward masked for 500ms.
Participants response times and accuracy (in percentage) were input into linear regressions.
Preliminary results for the TOJ task (N=17) demonstrate a significant difference in accuracy between
temporal congruency and incongruency across word type (t = -12.411, p <.0001). A similar effect is found
between spatial incongruency and congruency (t = -3.445, p = 0.0006). No significant effects were found on
response times in the preliminary analyses. For the LD task (N=16), preliminary results indicate that congruent
conditions are associated with greater accuracy compared to incongruent conditions for compounds (t = -
4.889, p <.0001), pseudocompounds (t = - 7.621, p <.0001), and unsegmentable monomorphemics (t = -
8.987, p <.0001). Response times were faster for pseudocompounds in spatially congruent conditions (t =
2.238, p = 0.0254).
Although it is early to draw major implications from this, there is a clear similarity between compound
and pseudocompound processing, which is in line with theories that posit an early morpho-orthographic
segmentation during word recognition.
Figure 1: Stimuli Manipulations
9
3. Generating and Revising Moral Decisions During First- and Second-Language Reading
Esteban Hernández-Rivera1, Karla Tarín Murillo1, Debra Titone1
1
Center for Research on Brain, Language and Music (CRBKM), Department of Psychology, McGill University
Reasoning and decision-making about real-world concerns are core features of human experience.
These decisions routinely involve judgements about the morality of people and their actions. For example, if
Person A does something objectively bad to Person B, we might decide that Person A behaved immorally.
However, if we later learn that Person B previously did something bad to Person A, we may be emotionally
inclined to revise our initial decision to partially exonerate Person A’s bad action. This example is framed
using negative valence; however, similar effects can ensue for positively valenced scenarios. For example, if
Person A does something objectively and unexpectedly good/kind for Person B, we might decide that Person
A behaved morally. However, if we later learn that Person B had previously done something objectively good
for Person A, we may be emotionally inclined to revise our initial decision to partially discount Person A’s
action as merely transactional.
Indeed, recent research on moral decision-making suggests that such effects emerge for people who
engage with such scenarios in their first language (Malle, 2021; Kim et al., 2022). Still, a crucial open question
is how different sources of information (i.e., objective facts about specific actions vs. extenuating
circumstances that appeal to our emotions) modulate people’s moral decisions, and of particular importance
here, how the languages in which we encode this information further modulates these phenomena (Caldwell-
Harris & Ayçiçeği-Dinn, 2021; Guglielmo, 2015). On the latter point, a growing body of work on the Foreign
Language Effect suggests that people who speak multiple languages are more objective and less emotional
when making moral decisions when reading in a second vs. a first language (L2 vs. L1; Costa et al., 2019;
Hayakawa et al., 2017). However, there are numerous gaps in this literature. For example, existing studies
mostly address negative scenarios, use small numbers of stock moral scenarios that are not
psycholinguistically controlled (e.g., variations of the Trolley Dilemma; Christensen et al., 2012), and have yet
to sort through whether such effects arise from individual differences among bilinguals vs. monolinguals, or
basic differences in comprehension ability between L1 and L2 language processing (e.g., Kirova & Camacho,
2020; Maschio et al., 2022).
To begin to address these issues, we investigate how L1 and L2 adult readers make use of contextual
and linguistic cues when generating and updating moral decisions about other people’s actions. We have
thus far recruited 71 English-French bilinguals who read and made a sequence of moral decisions for 122
linguistically controlled sentence pairs describing negative or positive transitive actions (e.g., A shamed B
after B conned the store customers; A admired B / because B valued the store customers). Participants’ word-
by-word reading times and two-alternative forced-decisions were recorded via button presses during a self-
paced reading and decision-making task.
A preliminary analysis using Bayesian logistic mixed effects models of participants’ initial and revised
moral decisions revealed three key results. First, irrespective of language background, all readers judged
negative moral actions to be less moral than positive actions. Second, given later additional context for each
action, all readers showed a tendency to revise their decisions to morally exonerate negative actions and, to
a lesser degree, morally discount positive actions (see also Kim et al., 2022). Third, and crucially, L2 readers
differed from L1 readers in their moral revisions, though not in a manner straightforwardly predicted by past
work on the Foreign Language Effect. Rather, L2 readers were less likely to exonerate negative actions given
additional context AND more likely to discount positive actions given past work. This suggests an overall
tendency to evaluate other people’s actions in a negative manner rather than in an unemotional utilitarian
manner, a finding that is different in spirit from past work on the foreign language effect. Whether this pattern
extends to on-line reading processes that precede decision outcomes, or persists after controlling for
10
people’s inherent moral tendencies, are among the many research questions we continue to pursue with
these data.
Antonio Iniesta1, Michelle Yang1, Anne L. Beatty-Martínez2, Inbal Itzhak3, Jason Gullifer4, Debra Titone1
1
Center for Research on Brain, Language and Music (CRBKM), Department of Psychology, McGill University
2
University of Southern California San Diego
3
Canadian Consortium on Neurodegeneration in Aging
4
Marianopolis College
Individual people use language in highly diverse ways. Multilingualism research suggests that language
diversity is central for regulating how languages are represented, accessed, and controlled, over and above
measures such as L2 age of language acquisition, and self-reported ability (e.g., Gullifer & Titone, 2020; Tiv
et al., 2021; Titone & Tiv, 2022). Inspired by information theory, language entropy (Gullifer & Titone, 2018;
2019) is a way to characterize how bilinguals use their languages in daily life. This measure indexes balance,
or diversity, in the use of two or more languages and can be computed from questionnaire data (e.g., Li,
Zhang, Yu, & Zhao, 2019). Language entropy can track compartmentalized versus integrated language use in
both general (i.e., using global language usage proportions) and socially based manners (i.e., by averaging
language use across different social spheres). However, a gap in this literature is how language entropy
(derived from basic questionnaires) related to more ecologically valid measures that quantify socially realistic
language use, such as people’s unique social networks.
Thus, we use social network data, which has high ecological validity but is lengthy to ascertain, to ground
questionnaire-based language entropy. In doing so, we specifically examined the links between
questionnaire-based language entropy (both general and socially-specific), and social network entropy (both
general and socially-specific) across 95 multilingual adults living in Montreal. We used detailed social network
measures that had ego and ego-alter measures to calculate general social network entropy and specific social
network entropy based on the information provided for each alter (i.e., specific people in each individual
social network). Exploratory factor analysis can provide an initial answer to the question, exploring how the
questionnaire-based language entropy and social network entropy loaded onto latent factors (see also
Gullifer et al., 2021). The results showed that questionnaire-based language entropy and social network
entropy grouped together for general and socially-specific manners to compute language use. These results
are an external validation of entropy as a tool to explore language diversity. Moreover, in a complementary
work, we validated entropy by exploring how different bilingual groups vary in their language usage and how
the composition of their social network plays a role (see Yang et al., 2023).
11
2. Virtual Reality Artificial-Intelligence Companion for Persons with Dementia in Long-
term Care
Background: Individuals with mild to moderate dementia living in long-term care (LTC) often enjoy reliving
memories and sharing stories. This project aims to reduce social isolation and increase engagement in LTC
residents by introducing them to a virtual companion presented using virtual reality (VR), to encourage
conversation and nurture storytelling.
Methods: The second-generation companion was developed through an iterative process by integrating
artificial-intelligence into a friendly virtual character. Feasibility was tested with up to 10 participants with
mild-to-moderate cognitive impairment living in LTC. Participants wear the VR headset for 1-5 sessions and
engage with the virtual companion for 5-20 minutes.
Participants provide feedback on their experience, interaction, contentment, and thoughts on the
companion’s personality. Interaction details (e.g. time, topics), participants' reactions, and discomfort are also
recorded. Qualitative data are analyzed using thematic analysis.
Results: Ten participants completed one session, seven completed a second session and four completed a
third. Sessions ranged from 0-54 minutes. Seven out of 10 participants actively engaged with the virtual
companion, and five had prolonged discussions of topics of personal significance. Participants who conversed
for longer periods expressed appreciation for the companion’s intelligence and thoroughly enjoyed their
interaction. Individuals with more social personalities particularly benefited. Technical issues prevented fluid
conversations with some participants. Adjustment of the software and testing with participants is ongoing.
Conclusion: Artificial-intelligence-enabled virtual reality companions have potential to create
enjoyable conversation and decrease social isolation in persons with dementia in long-term care. Future work
includes integrating greater diversity into the companion and educating healthcare professionals on its use.
Esteban Hernández-Rivera¹, Karla Tarín Murillo¹, Dan Chen¹, Michelle Jang², Gigi Luk², Debra Titone¹
¹ Department of Psychology, McGill University
² Department of Educational and Counselling Psychology, McGill University
The fields of numerical cognition, word problem solving, and multilingualism have remained relatively
separate, despite the shared connections between language and mathematics. Past work within mathematical
cognition has investigated numerical factors that impact word problem solving; however, this work has not
emphasized the linguistic context (Daroczy et al., 2015). Conversely, past work within bilingualism
investigating word problem solving usually focuses on problem comprehension; however, this work has not
emphasized the calculation component (Peng et al., 2020).
Bridging this gap, we investigated how bilingual adults read and process numerical information
embedded in word problems in their first or second language. We specifically addressed how differences in
math-specific language use impacted: 1) reading strategies for word problems containing exact (e.g., addition)
and inexact (e.g., comparison) operations; and 2) people’s decision times for word problems when accepting
12
or rejecting proposed answers that were correct or incorrect. We asked 57 English-French bilingual adults to
read a series of two-sentence English word problems in a self-paced manner (e.g., “Maria catches 3 fish, Karen
catches 8 fish, and Shawn catches 1 fish”), with the goal of either adding (e.g. “how many fish do Maria and
Karen catch”) or comparing quantities (e.g. “Does Maria or Karen catch more fish?”) across two
counterbalanced blocks.
There were four key results. First, task accuracy was uniformly high for all participants and conditions.
Second, people’s decision times were slower for exact operations (i.e., addition) than for inexact operations
(i.e., comparisons). Third, overall, acceptance of proposed correct answers was faster than rejection of
proposed incorrect answers. Finally, individual differences in math-specific language experience modulated
both people’s reading and decision speed. Specifically, people who frequently use English arithmetically took
longer to read the same word problem when followed by an inexact comparison operation compared to
problems followed by an exact addition operation. For decision speed, high percentage use of English
arithmetically resulted in faster decision times for inexact comparison operations compared to exact addition
operations.
Taken together, results suggest that bilingual experience impacts the time taken to encode and make
mathematical decisions, particularly for inexact comparison operations that are inherently more language
dependent. However, bilingual experience does not impact the outcome of those mathematical processes
(i.e., decision accuracy). Whether this extends to mathematical situations where reading/processing time is
fixed rather than at the reader’s discretion is a question that we are now pursuing.
The ability to recognize voices is paramount in language development. Voice recognition occurs by
discriminating auditory and visual information provided by the speaker. Personal differences create unique
voice characteristics and affect the sounds produced in speech, which helps listeners recognize a speaker as
different from others. Discerning voices as uniquely different is an important process, beginning prenatally
and developing in the first stages of language acquisition. Infants recognize maternal voice at birth (Kisilevsky
et al., 2003), along with tracking number of speakers (‘one’ vs. ‘more than one’; Kulshreshtha & Ramkumar,
2012). This is important in language detection and social development, as infants must recognize the origin
of sounds as threat, ally, or noise, along with its message. Little is known about infant discrimination between
numbers of voices. Our eventual project goal is to test bilingual infants on their recognition of a within-
speaker language change (i.e., code-switching between two languages). We first need to test whether
bilingual infants can recognize a speaker number change and a language change. The former is unknown,
but the latter is expected based on past research (e.g., Bosch & Sebastian-Gallesm, 2001).
This study will use as a guide the procedure from Johnson et al. (2011), where they found
monolinguals of 7-months could detect voice changes in native, but not non-native, languages. Our goal is
to test 18 bilingual infants of 7-months. Currently, half of that sample (n = 8) has been tested and we pre-
emptively begun testing infants of 11 months (n = 3) in case the younger infants do not succeed. Participants
habituate to three speakers in their native language of English or French. At test, looking time is measured
for familiar trials (e.g., A-English, B- English C-English), novel voice trials (e.g., A-English, B-English, D-English),
and novel language trials (e.g., A-English, B-English, E-French). Infants direct visual attention to new and
exciting stimuli, thus looking time provides a behavioural proxy for detecting perceptual changes (i.e., novel
voice and language).
13
Data analysis will include a within-subjects ANOVA, with alpha-controlled post-hoc comparisons of
familiar versus novel voice and familiar versus novel language. If the 11-month group is added, we will adapt
the analysis to a mixed 3 (test) by 2 (age) ANOVA. Data collection is ongoing, and results will further our
understanding of bilingual infant ability to detect number of voices and track speaker identity in complex
language environments.
In the context of Internet’s ever-growing influence, social media platforms, especially TikTok, have seen
the creation of new lexical items called algospeak, which denote the replacement of words to bypass
algorithms’ flagged terms (Steen et al., 2023). And although, linguists have noted algospeak’s place in
language change and cognitive linguistics (Lorenz, 2022; Moskal and Supernak-Polnik, 2023; Zhang, 2023),
there has been no research interest in how these specific substitutions are processed. As algospeak seems to
fulfil the same needs as euphemisms, “to attenuate […] a certain forbidden concept or reality” (Casas Gómez,
2009), we would assume it behaves similarly to other types of figurative speech with processing moderated
by different factors (Michl, 2019; Beck and Weber, 2020; Morid et al., 2021).
The proposed study looks to answer two research questions. First, how are algospeak words processed
compared to their non-algospeak counterparts? Second, do familiarity, literal plausibility (LP), and
orthographic/phonetic similarity (OPS) influence the processing of algospeak words? 40 adult speakers of
English that will self-report spending a minimum of 5 hours on TikTok per week will be recruited to complete
a Rapid Serial Visual Presentation (RSVP) paradigm while their EEGs are recorded. The stimuli will consist of
“she/he [verb] + [target word]” clauses embedded after a context-providing main clause (e.g. “Mary has an
OnlyFans; she is an accountant.”), with the target being either an algospeak (n=30) or non-algospeak (n=30)
word. The algospeak words will be divided across 6 conditions: high/low familiarity, high/low LP, and high/low
OPS. Analysis will be time-locked to the target word to measure early, late, and post N400 (following Morid
et al., 2021).
The first hypothesis is that algospeak is in fact processed differently from non-algospeak, as would be
seen in the presence of an N400 in the algospeak condition. The second hypothesis is that high familiarity,
LP, and OPS would induce a smaller signal amplitude, and this, at different time windows (familiarity (and
OPS) during early N400, and LP during late N400, as seen in Morid et al., 2021).
Establishing these potential processing differences would, first, help determine the kind of impact
loaded Internet words can have on our language processing, continuing the quest to study Internet language
more formally. Second, if factors are seen to influence algospeak processing, we would be able to insert the
results in the context of psycholinguistic research done on idioms and euphemisms.
14
6. Vowel Articulation: Do Simultaneous Bilinguals or Late Bilinguals Produce More
Nativelikeness?
Maria Falconi1
1
ERPLing Lab, Department of Linguistics, University of Ottawa
The hypothesis of biologically based critical periods for acquiring a second language has sparked
extensive debate within both cognitive psychology and linguistics. Much research has attempted to
understand the critical period hypothesis further and determine whether this is a factor in first and second
language acquisition (Steinhauer, 2014; Leśniewska & Singleton, 2021).
Considerable research has also investigated the cross-linguistic influence during vowel
acquisition/articulation of early and late bilingual speakers (Mack, 1989; Petersen, 2015). The proposed
research further investigates CPH and cross-linguistic influence, examining whether late or simultaneous
bilinguals can exhibit more significant native-like tendencies. Specifically, the study proposes researching the
articulation of rounded French vowels by addressing the following research question: Do simultaneous or
late bilinguals produce more native-like articulation of French rounded vowels?
Defining Bilingualism:
The proposed research targets two types of bilingualism: simultaneous bilinguals and late bilinguals. A
simultaneous bilingual can be defined as someone who learned a second language earlier in life/childhood
but utilizes this L2 sporadically. A late bilingual can be defined as an individual who learns an L2 later in life,
perhaps during teenage years or adulthood.
Target Language(s) and Inventory:
The proposed study aims to record the tongue and lip position of the articulation of French rounded
Vowels. Rounded vowels of French have been specifically targeted for this study as previous models, such as
Daniel Jones’ (1940) cardinal vowel system, describe these vowels as having identical degrees in tongue
height, as well as tongue advancement. This placement is hypothesized to account for non-native speakers’
lack of discrimination. Additionally, Levy & Strange explain that French speakers typically articulate rounded
vowels with the tongue body in a very forward position and very rounded lips. Specifically, the proposed
research will target rounded French vowels: /y, œ, u, o/.
Methodology:
The control group will consist of native French speakers. Experimental group 1 will consist of
simultaneous bilinguals with English as L1 and French as L2, while experimental group 2 will consist of late
bilinguals with English as L1 and French as L2. All Subjects will be presented with the same list of stimuli and
will complete a discrimination task. Recording of articulation will be done using ultrasound with a lip camera.
More native-like tendencies are expected to be seen within experimental group 2: the late bilingual group.
7. What’s up with this and that? Demonstrative processing and comprehension in English
speakers
The present research paper aims to investigate how pronominal demonstratives this and that are
processed by native English speakers. No experiment to date has examined instances where demonstratives
have dual interpretation in the same context. For example, in (1) below, this or that in the third sentence can
refer to either the noun phrase (NP) ‘a marathon’ or the idea unit (IU) of ‘running a marathon’. Therefore, the
15
following research question is posed: How are the demonstratives this and that with dual interpretation
processed in adult native English speakers?
It is evident that these types of dual-interpretation antecedents exist. Preliminary norming data was
collected from nine native English undergraduate students. Participants read 26 mini-stories – an example is
shown in (1) below. After reading each mini-story, participants were asked to circle the antecedent that either
this or that was referring to (either an NP or an IU). A total of 234 data points were collected (9 participants
with 26 stimuli per task). A significant difference was found between the variables, with 58% associating
demonstratives this or that with an IU antecedent, and 42% with a NP antecedent, X² (1, N=9) = 6.8, p= .0089.
(1) Kate was an athlete. She was running a marathon. This/that was her biggest challenge.
(2) The man started a new job. He bought a briefcase. This/that was his best purchase.
(3) Sarah always enjoyed learning. She was writing a book. This/that was very difficult.
Previous research has analyzed online processing of demonstratives this and that in discourse through
both self-paced reading (Crossley et al., 2017;Wittenberg et al., 2021) and eye-tracking (Çokal et al., 2014,
2018). Research has investigated distance (antecedents being either proximal or distal to the demonstrative),
referent type (this or that), and antecedent type (IU or NP), with antecedent type clearly defined – as shown
in examples (2) and (3) above. However, what happens when antecedents appear to be innately ambiguous?
Challenges faced when investigating these dual interpretation constructions are that (1) within both
online and offline processing measures, it is difficult to capture which antecedent type participants are
interpreting and (2) despite the antecedent or referent type, sentences remain grammatical. However,
although sentences remain grammatical, what are the processing consequences of interpreting one referent
or antecedent type over another? In what way can we develop an experimental design which captures the
fine-grained nuance that exists within dual- interpretation demonstrative processing?
Listening to someone speak is a complex task even under optimal circumstances. Ideal listening
conditions are rare, and speech is often heard in noisy environments. The speech signal is often imperfect,
and a noisy background may mask this already suboptimal signal (Lecumberri et al., 2010). Fortunately,
compensatory strategies exist.
While adult listeners can disambiguate noisy speech by relying on pre-existing language knowledge,
such as semantic (Golestani et al., 2009) or lexical context (Kaandorp et al., 2016), background noise seems
to differentially impact adults and children. A ‘developmental listening disadvantage’ has been demonstrated
in children under 15 years and is thought to be linked to cognitive and perceptual abilities that are not fully
developed in younger listeners, along with language proficiency when two or more languages are involved
(Johnson, 2000; Nelson et al., 2005; Neuman & Hochberg 1983).
Developmental research on infants’ language-related perceptual and lexical development rarely
examines the impact of background noise on speech perception, and when it does, only monolinguals with
homogeneous language exposure experiences are tested (e.g., Leibold et al., 2016; Newman, 2005, 2009).
The underlying cognitive processes and language exposure experiences involved in speech perception could
differ significantly between monolinguals and bilinguals in this most rapid language acquisition phase of the
lifespan (e.g., Werker et al., 2009). This may imply that the larger cognitive load associated with perceiving
words in noise may negatively impact bilinguals’ performance compared to monolinguals. Further, they may
have weaker word recognition in the target language due to reduced exposure (Morini & Newman, 2020).
16
In this study, we investigated how infants with various language exposures, aged 12 to 14 months,
listen to and recognize familiar words when hearing overlapping conversations, as it often happens at home,
at day care, etc. We assessed recognition accuracy in a group of English monolinguals and French-English
bilinguals on a preferential looking task involving hearing targets (i.e., English labels) for familiar objects
within different levels of overlapping babble (i.e., 2- and 9-talker babble). We also assessed the participants’
English proficiency and English exposure with parent-reported measures. Using mixed ANOVA and
correlational analyses, we will compare task performance with levels of noise and language exposure. We
expect poorer performance from the bilingual group. Indeed, larger cognitive load associated with word
perception in noise and a weaker language knowledge may negatively impact them compared to
monolinguals (e.g., Newman, 2009; Polka et al., 2008; Werker et al., 2009).
9. Does Better Objective Second Language Proficiency Predict Better Executive Function?
Madeline Smith1
1
Cognitive Neuroscience of Bilingualism Lab, University of Ottawa
Research on the potential relationship between being bilingual and executive functioning abilities often
produces different results on whether there is a bilingual advantage. The cause for these discrepancies may
be due to different methodologies across research studies, for example, the use of self-reported vs. objective
language measures and lack of consideration of individual variations in the bilingual experience. This project
aims to examine the use of speech samples as an objective measure of proficiency and the potential
relationship between objective proficiency and executive function, specifically inhibition. Participants are
young French-English bilingual adults between the ages of 18-35. Participants performed a series of
behavioral tasks in both languages, including a language background questionnaire, verbal fluency, speech
samples, Wechsler Adult Intelligence Scale subtests (digit span, letter-number sequencing, and matrix
reasoning), as well as two computerized executive function tasks (the Go/NoGo and the AX- CPT). It is
predicted that individuals who are more proficient in their second language, as evidenced by the objective
measure, will score better on the executive function tasks, demonstrating that bilingualism does indeed
interact with executive function. Data analysis is ongoing. Developing an objective measure of proficiency
that can be easily replicated across labs will help create more replicable data collection procedures and
findings. The results from this study will have implication for the development of such a measure.
The study of Second Language Acquisition (SLA) has long focused on understanding how learners
process sentences in real-time and whether their parsing mechanisms fundamentally align with or differ from
those of native speakers. Despite extensive research, no consensus has been reached, with two main theories
in the literature: 1) native-like processing is possible (Sabourin & Stowe, 2008); 2) L2 processing is
fundamentally distinct (e.g., Pakulak & Neville, 2011; Hawkins & Chan, 1997).
17
Limited research compares sentence processing between native English speakers and Mandarin
Chinese speakers who learned English as their second language (L2) (Dallas, 2013; Jessen, 2017; Dong, 2022).
This study used self-paced reading to investigate the processing of filler-gap dependencies by native English
speakers and Mandarin Chinese speakers with English as their L2. It examined how the semantic plausibility
of the filler and the length of the dependency influence the integration of the filler into the gap.
During ongoing data collection, 8 participants (5 native English speakers, 3 Mandarin Chinese speakers
learning English as L2) were included in a preliminary analysis. All were recruited from the Integrated System
of Participation in Research (ISPR) Student Pool. In the L1 group, while participants are anticipated to have
an L2 such as French, we only included those who learned L2 after L1 was fully acquired to avoid L2
interference as much as possible, and none of the L1 speaker reported proficiency in a second language. L2
participants' first language had to be Mandarin, with English as their only second language. They completed
a self-paced reading task on Gorilla, reading sentences and pressing the space bar to reveal the next word
while reaction times were recorded. A sample stimulus is provided.
Reaction times at the gap location in sentences were analyzed using Linear Mixed Models. The results
indicated that there were no significant main effects or interactions for the factors of filler plausibility,
dependency length, or language group on reaction times. Specifically, filler plausibility (F(1, 8.35) = 2.32, p =
0.14), dependency length (F(1, 0.65) = 0.002, p = 0.97), and language group (F(1, 1.49) = 0.69, p = 0.52) did
not significantly influence reaction times. In addition, no significant interaction effects were observed
between the two way and three-way interactions. Interestingly, from the mean values in our descriptive
statistics, we observed that both the L1 and L2 groups dedicated more time to reading plausible sentences
than to implausible ones, for both long and short dependencies. This observation contrasts with previous
research suggesting that parsers slow down their reading time at gaps that cannot be plausibly filled (Wagers
& Phillips, 2014). However, the divergence could be due to the parsers’ sensitivity to implausible fillers,
leading them give up on integrating it into the gap. If so, this sensitivity to the semantic anomaly is consistent
with the N400 observed at the gap location when the filler cannot be integrated plausibly (Dallas et al., 2013).
The lack of significance may also be attributed to the limited sample size in the preliminary analysis.
We aim to recruit more participants and perform further analysis before the conference.
This study aims to determine the potential effects of adult masking during the COVID-19 Pandemic on
the speech production of children. Increased literacy success in children has been shown to correlate with
more exposure to multisensory input from the ages of three to five, as well as better phonological awareness
18
and frequent mouth gazing (Choi et al., 2018; Tenenbaum et al., 2014). Limited/restricted access to visual
articulation cues have revealed blind children developing differently (Gordon-Pershey et al., 2018; Merrit,
2015), which appears to be replicated by masking (Gilger, 2022; Lewkowicz, 2021). Masks have been found
to reduce word recognition and degrade speech quality (Bottalico et al., 2020; Sfakianaki et al., 2021; Toscano
& Toscano, 2021). This study aimed to evaluate if children learning to speak through the pandemic,
particularly when masking was common, have lower speech accuracy than those whose speech development
happened before the pandemic. To do this, we examined speech accuracy in children 3-5 years old and
compared a group of children developing their language during the pandemic (i.e., “intra- pandemic group”)
with a “pre-pandemic” group. For the pre-pandemic group, we used pre-existing data from the CHILDES
database of speech, from typically developing children aged four to five in 2018 (Judd, 2018). We examined
children’s accuracy in producing consonant phonemes that are expected “early”, “mid”, and “late” in
development. We then replicated the corpus measure and analyses which included a narrative task – a picture
book without words, “Frog Where Are You?” by Mercer Mayer (1969) – and a picture naming task. The pre-
pandemic children performed as expected, with the most accuracy in early phonemes and the least in late
phonemes. To date, one interview has been conducted for the intra-pandemic group, but this interview did
not to elucidate any results, due to resistance to the interview process. An intra-pandemic group is expected
to have the most difficulty with late phonemes - more so than a pre-pandemic group because early and mid
sounds would be expected to have already been acquired before the Pandemic, but not late phonemes.
Therefore, if a child experienced an extended period in masked environments, they could face linguistic
“delays” due to decreased visuo-linguistic information. However, any issues with speech are expected to be
temporary setbacks that will self-correct once visual speech cues are available.
Theories of metaphor and simile processing often emphasize the influence of context during
interpretation. However, other variables, such as aptness (i.e., how well a set of properties of the vehicle apply
to the topic), familiarity (i.e., the frequency of previous exposure to a given topic-vehicle pairing), and
conventionality (i.e., the vehicle term expressing a stable metaphoric meaning across various contexts of use),
play important roles in this process (Bowdle & Gentner, 2005). Although evidence suggests that the effort
associated with metaphor interpretation is influenced by some of those underlying variables (e.g., Lai et al.,
2009), less is known about these variables’ influence on the speed at which metaphors and similes are
recognized. Consider the conventional metaphor ‘time is money’. The vehicle (i.e., ‘money’) may be quickly
recognized as metaphorical because processing the topic (i.e., ‘time’) may prime the metaphorical meaning
of the vehicle. The goal of the present study (ongoing) is to explore how such variables influence the
recognition of the figurative vehicle. This study uses a gating task in which participants listened to a sentence
that was cut 50ms following the onset of the vehicle. Their task was to guess the word that was cut short and
provide a confidence rating. Over the trials, the duration of the vehicle was progressively increased in 50ms
increments until its offset. After each audio segment, the participant guessed the word and provided a
confidence rating. The vehicles were included in one of three conditions: (1) metaphor (e.g., ‘time is money’);
(2) simile (e.g., ‘time is like money’); and (3) neutral contexts (i.e., “The next word is: X”). Both metaphors and
similes were embedded in discourse contexts. The design allows for the identification of the recognition point
of the vehicle, defined as the point at which the participant first guessed the word correctly and did not
19
subsequently change their guess. Aptness, familiarity, and conventionality norms will be regressed on the
vehicle recognition time, among all three conditions. We predict that high aptness, familiarity, and
conventionality will facilitate the recognition of metaphor and simile vehicles, which will be reflected in earlier
vehicle recognition than expressions low in aptness, familiarity, and conventionality. No effect from these
variables should be observed in the neutral context condition. This study may improve our understanding of
how different contextual variables are involved in the interpretation of similes and metaphors.
Michelle Yang1, Antonio Iniesta1, Anne Beatty-Martinez2, Inbal Ithzak3, Jason Guillifer4, Debra Titone1
1
Center for Research on Brain, Language and Music (CRBKM), Department of Psychology, McGill University
2
University of Southern California San Diego
3
Canadian Consortium on Neurodegeneration in Aging
4
Marianopolis College
In the linguistically diverse setting of Montréal, each speaker has a unique language background that
is influenced by the interplay of historical, political, and sociocultural factors (Kircher, 2014; Leimgruber, 2020).
For example, 59.9% of residents in the metropolitan area of Montreal reported French as their mother tongue,
compared to 11.2% of residents who reported English as their mother tongue (Statistics Canada, 2021).
However, 56.4% of residents report that they are able to have a conversation in both English and French,
compared to just 18.0% in the rest of Canada, and 46.4% in Quebec as a whole (Statistics Canada, 2021). As
such, each and every speaker in Montreal has a unique experience using French and English in their day-to-
day activities. One way to quantify this type of experience is using language entropy (Gullifer & Titone, 2018;
Gullifer & Titone, 2019). Language entropy measures the diversity of multilingual language usage. While there
has been some research indicating that language entropy relates straightforwardly to other markers of
bilingual experience (such as exposure and abilities), predicts cognitive control abilities (and corresponding
neural correlates), and varies depending on the communicative context of language usage (Gullifer et al.,
2018; Gullifer & Titone, 2020; Gullifer & Titone, 2021), it is not well understood what the impact of language
background and social context is on language entropy.
To this end, we build on a complementary work (see Iniesta et al., in prep) to analyze how three different
bilingual groups vary in their language usage and how the composition of their social network plays a role.
We tested French-English, English-French, and simultaneous bilinguals from the McGill University community.
We thus examined the links between general language entropy (both general and context-specific), and
entropy based on more detailed social network measures (both general and context-specific). We use
Exploratory Factor Analysis to explore how the social context of multilingualism affects language entropy. We
found that English-French bilinguals had higher factor scores for L1 Homophily (a measure of how many
people in your network with whom you speak your L1) compared to French-English and simultaneous
bilinguals, which means that English-French bilinguals in our sample are more likely to have social
connections with people who can speak English. In addition, we found that this remained consistent across
social contexts. We also found that general language entropy and social network-based entropy loaded into
the same factor, alongside L2 Homophily (a measure of how many people in your network with whom you
speak your L2) both generally and across social contexts, with no significant differences between our
language groups. This suggests that general language entropy taps into the same underlying construct as
the more fine-grained social network entropy and is consistent across various sociolinguistic groups. This
20
work highlights the importance of taking specific communicative contexts and sociolinguistic background
into account when interpreting language entropy.
5. How does a Person’s L1 Writing Script Impact L2 Reading in a Same or Different Script?
Naima Mansuri1, Antonio Iniesta1, Esteban Hernández-Rivera1, Pauline Palma1, Debra Titone1
1
McGill University
Roughly half of the world’s population speaks more than one language (Grosjean, 2010) and a large
percentage of those people read in their second language (Lallier et al., 2021; Siegelman et al., 2023). Thus,
an open question is whether and how people’s multilingual knowledge impacts their second language
reading processes. Here, we investigate whether competing influences from people’s first language (L1)
writing system or orthography (i.e., alphabetic, logographic or alphasyllabic) can impact reading
comprehension and fluency when English (which has an alphabetic writing system) is a second language (L2).
Based on models of L1 and L2 reading (e.g., Ziegler & Goswami, 2005), our general hypothesis is that matches
or mismatches in people’s L1 and L2 writing scripts will modulate the expected relationship between L2
English reading proficiency and how often people use their L2 in daily life (i.e., L2 usage). Using a subsample
of 1073 healthy adult participants, from Siegelman et al. (2023), we found a main effect of L1 writing scripts
for both L2 English reading proficiency and reaction time on L2 English reading proficiency tasks. Both L1
alphabetic speakers and L1 alphasyllabic speakers scored higher and had faster reaction times than L1
logographic speakers. Additionally, we found a significant interaction between L1 Writing Script and L2
English Reading Usage on reaction time on L2 reading proficiency tasks, as alphabetic and alphasyllabic
writers both had faster reaction time with increased usage, but this effect did not hold true for the L1
logographic readers. These results suggest that orthographic features of an L1 can impact L2 reading. In
doing so, these findings clarify how people’s multilingual knowledge and experience mutually constrain each
other in the context of L2 reading. They also suggest further avenues of theoretical and empirical
advancement.
6. Code Switching and You: The Influence of Language Production on the Integration of
the Bilingual Lexicon
Over the last decade, the majority of linguistics research examining the bilingual mental lexicon has
evaluated the influence of participant level factors such as age of acquisition (Sabourin, Brien, et al., 2014),
manner of acquisition (Sabourin, Leclerc, et al., 2014), and language proficiency (van Hell & Tanner, 2012), on
the development and organisation of the bilingual lexicon. Recently, the field has seen a growing interest in
the study of code-switching, the mixing of two or more languages in a given discourse (Poplack, 1980), and
its use as a marker of language co-activation (Kootstra et al. 2020; Hofweber, & al., 2023). While the literature
agrees that language switching incurs cognitive costs (Hofweber, & al., 2023), a number of ERP studies
(Blackburn & Wicha, 2022) and behavioral studies (Hui et al., 2022) have proposed that these costs are
lessened for bilinguals who regularly engage in this behavior. To my knowledge, only one study (Gosselin &
Sabourin, under review) has previously examined the influence of code-switching on lexical decisions,
reporting that individuals who deliberately code-switched performed better on mixed language lexical
decision task than those who unintentionally code-switched.
21
This project, which starts recruiting participants in mid February, will evaluate the influence of frequent
code-switching on the integration of the bilingual lexicon. French-English simultaneous and early bilinguals
will be asked to complete two lexical decision tasks, one with english targets and one with french targets.
These lexical decision tasks will be primed cross linguistically, using two priming conditions of interest,
semantic, using related words, and translation, using direct translations. Assessed using our labs language
backgrounds questionnaire, each participant will be given a code-switching score which reflects their
language-switching profile. We hypothesize that participants with larger code-switching scores will present
with faster reaction times compared to their non-switching peers, in the translation and cross-linguistic
semantic conditions. At its most basic, code-switching is the insertion of a word from a different language
based on its meaning, with repeated use this may strengthen connections between language nodes. If
significant, these results would suggest that language production profiles have a larger influence on the
development of the bilingual lexicon than was previously believed. Overall, these findings would underline
the importance of detailed language use profiles, and suggest the need to collect more comprehensive
language use information when investigating the bilingual mental lexicon.
Previous work has linked bilingualism with executive functioning, suggesting that bilingualism confers
a cognitive advantage. Results have been mixed, due in part to methodologies largely comparing bilinguals
and monolinguals on behavioral performance alone, while overlooking individual differences within bilinguals
(e.g., language proficiency, frequency of use, or age of acquisition of a second language). The current
multimodal study aims to identify specific aspect(s) of bilingualism (e.g., proficiency, fluency, age of
acquisition of a second language) that may confer neurocognitive and behavioural advantages in executive
functioning to start to address these issues.
Data collection is currently ongoing with a goal of recruiting a total of 100 young adult English-French
bilingual participants (current n=10). Executive functioning and bilingual language experience will be included
as continuous predictors in regression-based analyses with measures of neural coherence across functionally
relevant frequency bands (measured using resting state electroencephalography; rs-EEG) as outcomes. We
hypothesize that language experience confers a neurocognitive advantage; for example, that greater bilingual
language experience (e.g., proficiency, fluency, and age of acquisition of a second language) will be linked to
superior executive functioning and increased neural coherence in the alpha frequency band across fronto-
parietal brain regions (e.g., dorsolateral prefrontal regions, posterior parietal regions).
Maïsha Morneau1, Christie Brien1, Theresa Rabideau1, Cristina Atance1, Chris Fennell1, Tania Zamuner1
1
UOttawa Living Lab, Department of Linguistics, University of Ottawa
22
mobilisation des connaissances. Ces activités ont pour but de démontrer les résultats récents de la recherche
sur le développement de l’enfant.
Le but de cette étude est donc d’évaluer le succès de ces activités. Afin d’y parvenir, les familles ont
répondu à un questionnaire suite à leur participation à l’activité du Singe Pirate, un jeu visant à démontrer le
développement des fonctions exécutives des 3-4 ans, comme le contrôle des impulsions et la mémoire active
(Carlson, Davis, & Leach, 2005). Le jeu consiste à avoir plus de pierres précieuses que le Singe Pirate, une
peluche portant un cache-œil. Les comportements des enfants lors de cette activité démontrent que les
enfants en plus bas âge ont tendance à choisir le plateau comportant le plus de pierres précieuses, étant
attirés par ce qu’ils voyaient et oubliant les règles du jeu.
À la fin du jeu, les participants étaient approchés, puis invités à répondre à un questionnaire avec le
logiciel Qualtrics (Qualtrics, Provo, UT, 2020). Le questionnaire comporte des questions de contenu comme «
L'activité m'a permis de mieux comprendre le développement de l’enfant », ainsi que des questions d’ordre
démographique. Les enfants avaient également la possibilité de répondre à trois énoncés du type « J’ai aimé
l’activité » suivi d’une échelle Likert/bonhomme sourire à 5 échelons, ainsi que deux questions ou- vertes.
Des 62 adultes ayant répondu au questionnaire, 82% de ceux ayant répondu vouloir en apprendre davantage
sur le développement de l’enfant ont mentionné avoir appris une nouvelle connaissance sur le sujet lors de
l’activité. Des 23 enfants ayant répondu, 93% d’entre eux ont affirmé avoir apprécié l’activité.
Les résultats démontrent donc que les activités de mobilisation des connaissances du Laboratoire en
Action ont un impact positif et sont en mesure de communiquer de nouvelles informations aux familles sur
le développement de l’enfant de manière ludique, répondant ainsi à leurs objectifs premiers. Ainsi, les activités
de recherche offrent l’occasion de participer de manière efficace à la transmission du savoir. Elles pourraient
être adaptées à plusieurs sujets de recherche différents, afin de les intégrées à la mobilisation des
connaissances d’autres intérêts.
OBJECTIFS: Des expérimentations chez les adolescents suggèrent que ceux-ci développent la maîtrise
de l’accord de l’adjectif en genre ou l’accord sujet-verbe, différemment des adultes. Toutefois, ces travaux ne
tiennent pas compte des facteurs influençant la maitrise du langage : environnementaux (p. ex., la fréquence
d’utilisation et d’exposition à la langue), ou propres au locuteur (p. ex., l’âge d’apprentissage, les compétences
cognitives). De plus, ils font uniquement appel à des adolescents monolingues, qui ne parlent qu’une langue,
et font donc l’impasse sur la manière dont l’apprentissage du langage se déroule chez les adolescents
apprenant une seconde langue (L2). Enfin, ils focalisent peu sur l’apprentissage de l’accord en genre, aspect
important de la morphosyntaxe du français. Mon projet doctoral a donc pour but d’identifier comment se
développe et quels facteurs sont déterminants pour l’apprentissage de la morphosyntaxe en L2 chez les
adolescents anglophones apprenant le français L2. Il permettra de faire le lien entre ces facteurs et leur impact
sur le développement linguistique en L2 chez l'adolescent.
MÉTHODOLOGIE: Je mènerai des expérimentations fondées sur la compréhension de phrases en
français : Ces expérimentations en laboratoire ciblant deux conditions qui ont déjà montré leurs preuves avec
des adolescents monolingues : 1. L'accord en genre : des phrases dont l'article ou l’adjectif comporte une
erreur d'accord de genre avec le nom (p. ex., La chaise *vert) seront utilisées. 2. L'accord sujet-verbe : des
incohérences d’accord sujet-verbe seront créées en présentant aux participants des phrases avec des sujets
23
et des verbes grammaticaux avec des images représentant un nombre différent de sujets (p. ex. “Ils rugissent”,
avec un seul lion sur l'image, ce qui crée une incohérence grammaticale). Une électroencéphalographique
(EEG) aura lieu pendant la présentation des stimuli. Ces signaux permettront d’identifier les réponses
neurocognitives des adolescents face aux erreurs d’accord. De potentiels évoqués (PÉs) reflétant le traitement
d’erreurs d’accord (connaissances grammaticales) ou d’erreur lexicales (connaissances des mots) seront
extraits des signaux EEG.
Des tâches linguistiques et cognitives, incluant des exercices de jugement de grammaticalité et de
production écrite serviront à mesurer les connaissances linguistiques des adolescents participants et seront
utilisés pour prédire la réponse neurocognitive aux erreurs. Nous évaluerons aussi le rôle de facteurs
extralinguistiques telles les sources d’exposition au langage (famille, amis, media) et la fréquence d’usage,
ainsi que ceux propres aux participants tel l'âge du début d’apprentissage et la mémoire de travail, afin de
savoir lesquels influencent l’acquisition de la L2.
PSYCHOSHORTS 2024
24