2024 EFL Annual Meeting: Poster Session

Abstracts

Individual Differences in Leveraging Regularity in Emergent L2 Readers in Rural Côte d’Ivoire

Henry Brice
University of Toronto

Research Team:
Benjamin Zinszer, Danielle Kablan, Fabrice Tanoh, Kaja J. Jasińska

Statistical learning (SL) approaches to reading maintain that proficient reading requires assimilation of the rich statistical regularities in the writing system. Reading skills in developing first- and second-language readers in English have been shown to be substantially predicted by individual differences in sensitivity to statistical regularities in orthography and semantics, with good readers relying more on orthographic consistency, and less on semantic associations. However, the study of SL and its relation to reading has been primarily studied in English readers in WEIRD countries, limiting our ability to generalise to the majority of the world’s population. In this study, we examined individual differences in sensitivity to regularities utilising a word naming task in emergent French readers in rural communities in Côte d’Ivoire (N=134). We show that, in contrast to previous studies in WEIRD contexts, in our cohort better readers leverage semantic associations more strongly, while individual differences in sensitivity to orthographic consistency did not predict reading skill. Crucially, relatively little variance in reading skill was explained by sensitivity to these regularities, as compared to previous studies in WEIRD populations. This showcases the importance of cross-linguistic and cross-cultural research to back up universal theories of literacy, and suggests that current SL accounts of reading must be updated to account for this variance in reading skills.

Predicting Literacy in the Brain in Emergent Readers in Rural Côte d’Ivoire: A Longitudinal Study

Henry Brice
University of Toronto

Research Team:
Benjamin Zinszer, Joelle Hannon, Fabrice Tanoh, Konan Nana N’Goh Anicet, Kaja Jasińska

We will present preliminary results from a longitudinal intervention in 135 5th-graders in rural Côte d’Ivoire, utilising fNIRS to probe print and speech processing in the brain in L2 learners of Ivorian-French. Much of the study of the neural underpinnings of literacy has focused on populations from high-income countries. Côte d’Ivoire, a low-middle income country with literacy rates <50%, provides a very different perspective. Children growing up in rural Côte d’Ivoire typically speak one over 60 local languages as their L1, and only begin acquiring literacy skills in L2 French in school. Children vary greatly in the age at which they begin school, and can begin as late as 12. This presents a unique window into L2 language and literacy acquisition, both in that here, literacy is acquired relatively late and in parallel with spoken language, and without prior L1 literacy skills. Such a situation is far from rare from a global perspective, but is largely lacking from the literature. We present results showing how individual differences in activation for print processing predicts literacy skills two later, and an examination of the impact of age of exposure to literacy on the development of the reading network.

Impact of interrupted schooling on functional connectivity for reading in resettled Syrian refugee children in Canada

Kaja Jasińska
University of Toronto

Research Team:
Hassan Abdulrasul, Henry Brice, Angela Capani

A significant gap in our understanding of the development of the reading brain lies in comprehending how it is affected by the timing of literacy instruction. Current literature predominantly focuses on children who commence reading instruction concurrently with their formal schooling, leaving a gap in understanding about how delayed or interrupted literacy instruction impacts the neural underpinnings of reading.

We investigate the impact of educational interruptions on the functional connectivity within the developing reading network in the context of recently resettled refugee children. This population, often experiencing disrupted schooling and delayed literacy development, presents a unique opportunity to explore the trajectory of reading network development.

We examined the resting-state functional connectivity of the reading network using fNIRS in a cohort of 54 resettled Syrian refugee children (age 8-17), who have encountered varying durations of educational interruptions at different ages. The study correlated the observed neural connectivity with standardized reading assessment scores to examine how the age and duration of educational interruptions affect the functional connectivity of the reading network. Our preliminary results suggest that both inter- and intra-hemispheric connectivity were negatively associated with age of displacement, however connectivity in the left reading network was positively related to reading scores.

An examination of social emotional wellbeing and reading development in Syrian refugee children

Angela Capani
University of Toronto

Research Team:
Sara Qadoumi, Sherry Wu (BA), Angela Capani (MA), Brooke Wortsman (MEd), Raabya Rashad (BSc), Mandana Jafarian (MASc), Hassan Abdulrasul (BSc), Kaja Jasińska (PhD)

Refugee children exhibit persistent challenges in reading and language, which are crucial for academic success and host-country integration. Reading skills, and learning outcomes broadly, are influenced by social emotional wellbeing (SEWB); those with SE concerns are less able to focus in the classroom, including during direct instruction required to develop reading skills. SEWB may be especially relevant for refugee children, who face pre- and post-arrival challenges that increase their SE vulnerability. However, research examining the connection between SEWB and learning in refugee children is sparse and limited to short-term outcomes. We examined the association between SEWB and reading in Syrian refugee children (n=49; Mage=13.74, SD=2.30, range=8-17), 1.6-7.8 years (Myears=6.7, SD=1.2) post-arrival in Canada. We measured SEWB using the Strengths and Difficulties Questionnaire. Reading and language skills (phonological awareness (PA), vocabulary (V), decoding (D), comprehension (C)) were assessed using standardized measures. Preliminary data show that 43% of children displayed reduced SEWB, and lower reading and language scores (~≥1 SD below M) compared to same-aged North American children (D: 88.05; PA: 86.29; V: 73.39; C: 65.25). Aspects of SEWB (i.e. prosocial behaviour) and reading (i.e. decoding) were moderately correlated (r(32)=0.49, p=0.004). Results suggest that learning interventions should include supports targeting SE functioning of refugee children.

Investigating fNIRS Test-Retest Reliability during Lexical Decision

Laura Elliott
Dalhousie University

Research Team:
Laura Elliott, Saisha Rankaduwa, Devon Bode, Marilla Hulls, Clare Reynolds, Cindy Hamon-Hill, Aaron Newman

View a digital copy of the poster here.

Functional near infrared spectroscopy (fNIRS) is a relatively new technique in the field of neuroimaging, and the reliability of its signals has not yet been established during tasks of reading. Prior to using fNIRS to investigate reading and its development, it is necessary to establish the technique provides a reliable measure of an individual’s current task-related neural activity. Therefore, the objective of the present study is to assess the test-retest reliability (TTR) of the signals recorded by fNIRS, during a standard reading task, at the group and subject level. We recruited adults to complete a lexical decision task during fNIRS recording twice, with the sessions occurring one week apart. We hypothesized we would observe significant print, sublexical, and lexical tuning effects at the group level, as these have been demonstrated in adult readers using electroencephalography and functional magnetic resonance imaging. However, our preliminary results have demonstrated no significant print, sublexical, or lexical tuning effects at the group level. Additionally, we hypothesized that TTR, as measured by the intraclass correlation coefficient (ICC), would be good (> 0.59) at the group level, and fair (> 0.40) to good at the single subject level. Our results demonstrate significant variability in TTR at the single subject level, for all contrasts of interest. At the group level, we observed fair to good TTR for print and lexical tuning contrasts in specific regions of interest, but not for sublexical tuning. Finally, we hypothesized we would observe better TTR when analyzing data at the level of regions of interest, relative to individual channels, and this was supported by our preliminary results at the group level. Identifying the reliability of fNIRS during a standard reading task will help to guide recommendations and improve confidence in its use in future studies of reading development.

Far, Car, War, Boar: Mechanisms of Automatic Word Recognition

Niki Sinha
Western University

Research Team:
Niki Sinha & Marc Joanisse

Word recognition is integral to fast and fluent reading. This automatic process retrieves orthographic (word form), and phonological (sound) information immediately, but is dependent on previous experience and expectations. Neuroimaging studies have shown differences in expected vs. perceived word information elicit EEG responses sensitive to distinct types of word information. To assess how the brain evaluates and responds to orthographic and phonological information individually during reading, this study created sound-spelling conflict by word association. Twenty-nine English speaking adults completed a rhyming task while EEG data was recorded. Participants visually presented with word-pairs were instructed to judge if the words rhymed via button press as quickly and accurately as possible. Word-pairs were visually presented in four conditions: congruent trials which shared both orthography and phonology (cool/pool) or shared neither (boat/fair), and incongruent trials with conflicting orthography and phonology (cane-rain or most/cost). Participants had slower reaction time and lower accuracy in their judgement of incongruent trials. Comparison of congruent trials to their incongruent counterparts revealed N400 negativity occurred independently during recognition of orthographic and phonological mismatch. Additionally, a more positive P600 was found in incongruent trials than congruent, suggesting the late component was evoked by sound-spelling conflict itself. Findings indicate that both orthographic and phonological differences between word-pairs are automatically recognized at the N400 window (400-600ms), however the processing of conflict does not fully occur until the P600 interval (600-900ms). The late positivity may reflect additional error monitoring given unexpected sound-spelling mismatch, or moderation of prior expectations following conflict.

Evaluation of Urdu Text Input Methods

Yalmaz Ali Abdullah
University of Alberta

Research Team:
Yalmaz Ali Abdullah, Carrie Demmans Epp

View a digital copy of the poster here.

Urdu is one of the most widely spoken languages in the world, with over 77 million native speakers in Pakistan and India alone. Despite this, it has a scant digital presence. What presence exists, is primarily in the form of a non-standard Romanized Urdu. This greatly limits accessibility as familiarity with the Latin alphabet becomes a prerequisite for accessing most digital Urdu content. Part of the reason that romanized Urdu is used is because of the difficulty associated with entering text in the more traditional Nastaliq script. Text entry in Nastaliq requires the use of Urdu layout keyboards which arrange characters according to frequency or following a phonetic mapping to QWERTY. The phonetic keyboards present a gentler learning curve at the potential cost of ergonomics; the frequency based layouts are the opposite. Nastaliq also has a larger number of characters than Latin so both these layout types hide some characters behind the shift key (ex: the ذ character requires shift+z on phonetic keyboards). This places a greater burden on both the ring and little finger of the left hand which is ergonomically undesirable. This project quantitatively examines the usability of both types of layouts and compares them with a potential third option, a standard QWERTY keyboard with the addition of an Input Method Editor (IME). IME keyboards allow users to input text using the Latin alphabet. These keyboards then transliterate the Romanized input to the associated script. An IME keyboard has the potential for the gentlest learning curve because of a greater degree of knowledge transfer (from the commonly used QWERTY layout) and better ergonomics. Access to text input would still be limited to those familiar with the Latin alphabet but by reducing the input burden we can potentially encourage the creation of digital Nastaliq Urdu content. More accessible content can increase the audience for digital Urdu and help stimulate further development in other areas.

Decoding social attribution: Using computational models to analyze non-social and social signal components in neuroimaging data of autistic and non-autistic viewers

Brea Chouinard
University of Alberta

Research Team:

Chouinard, B., Keskin, M., Lahert, N., O’Connaill, R., Gallagher, L., Kelly,C., & Fyshe, A.

Researchers use animations of moving geometric shapes to evaluate whether viewers attribute human characteristics to the shapes and movements. Compared to non-autistic individuals, autistic individuals identify fewer social elements and use fewer terms inferring what the characters were thinking or feeling. We used computer models of language and encoding models to investigate the types of information being processed while autistic and non-autistic adolescents viewed a 7-minute social attribution animation. We found that variance in autistic participant neuroimaging responses was strongly correlated with the presence of the three main characters, while for non-autistic participants the strongest correlation was with social interaction/emotion. Using a language model we analyzed encoding model fit for representations that did or did not have access to the social interpretation of the animation. We found that encoding models without access to social language (e.g., “small triangle”) outperformed those with access to social language (e.g. “boy”) in the autistic group alone, with no difference in the non-autistic group. Altogether, we offer characterization of the semantic information that was encoded by the brain during viewing of the movies in each group, suggesting that representations of social information differs between autistic and non-autistic participants.

Accent intelligibility: Understanding the effects of race-based expectations and bilingualism

Vanessa Ritsema
Dalhousie University

Research Team:
Vanessa Ritsema, Rebeka Workye, and Drew Weatherhead

Race-based expectations and linguistic background can both influence accent intelligibility. However, past research on accent intelligibility has yielded conflicting results, and the underlying mechanisms of accented-speech processing remain unclear. Furthermore, the effects of race-based expectations and linguistic background have been studied largely in isolation, with past research focusing on monolingual listeners. The current study seeks to better understand the combined effects of race-based expectations and linguistic background on accent intelligibility by replicating the experimental design of McGowen (2015) with bilingual participants. The current study recruited 127 American participants fluent in English and Norwegian who listened to and transcribed 60 Mandarin-accented English sentences in noise. Sentences were presented with a White face or an East Asian face, with each condition designed to match or violate listeners’ race-based expectations. Following sentence transcription, participants were asked to rate how strong they found the accent to be from the experiment on a Likert scale. Participants also completed a short questionnaire that assessed their language usage, proficiency, familiarity, and attitudes. For analysis, a generalized linear mixed effects model was fitted to the data with condition, sex, accentedness rating, and participant language use as predictor variables. Results suggested that speaker race is not a salient speaker cue for bilingual listeners, and that bilingual listeners may therefore have different associations between ethnicity and accent compared to monolingual listeners.

Are You Wearing Your Clothes or are Your Clothes Wearing You? The Roots of Children’s Clothing-Based Inferences about Group-Level Knowledge

Alexandra Dyack
Dalhousie University

Research Team:
Cognition and language learning research group, Rebeka Workye, Marise Barsoum, Emma Reeves, Victoria Aragon, Drew Weatherhead

Clothing is a powerful social marker that allows adults and children to implicitly gather information about others such as group membership, occupation, and social status. Recent research shows that children make inferences about people’s knowledge based on their clothing, meaning they can deduce a group of people wearing the same clothing are likely to share similar knowledge. However, this research does not speak to where children are attributing group-level knowledge. It is possible that they believe knowledge comes within an individual or that it is attached to clothing, such that a person gains knowledge when they put on certain clothes and loses it when the clothes are removed. The current study assessed 5-8-year-old children’s beliefs about knowledge when clothing was manipulated, by asking whether an individual retained previous knowledge and gained new knowledge upon changing her clothes. On average, participants believed an individual retained her previous knowledge after changing her clothing, but were unsure if she gained knowledge associated with her new clothes. These results support person-specific knowledge attribution, where children believe knowledge is an internal property that persists across external changes. Early on, children make complex inferences about people’s internal properties based on group membership; education about within- and between-group diversity as it relates to clothing is crucial to ensure the prevention of overgeneralization, stereotyping and prejudice.

Representational Similarity Analysis of the Neural Representations of Visual, Orthographic, Phonologic, and Semantic Processing During Silent Word Reading

Deanne Tak On Wah
Western University

Research Team:
Deanne Tak On Wah¹, Marc Joanisse^1,2

¹Department of Psychology, University of Western Ontario, London, Canada

²Haskins Laboratories, New Haven, CT, USA

Contemporary reading models propose two word recognition processes: decoding from orthography to phonology, and whole-word reading from orthography to semantics. Prior neuroimaging studies localized regions involved in reading, while emerging techniques enable the comparison of activation patterns with cognitive models outside the brain. Our study uses Representational Similarity Analysis to investigate patterns of activity in sub-regions of the reading network and evaluates their similarity to visual, orthographic, phonologic, and semantic models during word reading. Fifty neurologically healthy, monolingual English-speaking adults participated in two sessions. The first included standardized reading and intelligence measures, demographics questionnaires, and a word naming task. In the second, participants silently read monosyllabic words while detecting person names in a 3.0T fMRI scanner. Word-by-word theoretical representational dissimilarity matrices (RDM) were constructed, with the visual model using the proportion of non-overlapping pixels, the orthographic model using Levenshtein Distance, the phonological model using Phonological Edit Distance, and the semantic model using cosine distance of Global Vectors. Empirical RDMs were calculated using Pearson Correlation Distance. Spearman Rank Correlations were used to correlate the theoretical and empirical RDMs. One-sample t-tests determined whether the correlations were distinct from 0. Results indicate distributed patterns of activation in the reading network associated with semantic processing, suggesting a task-related bias towards semantic processes.