Department of Education

Viewing archives for Seminar

Abstract tbc

All welcome to join in person.

If you wish to join online, pre-registration is required (no need to register again if you have already done so in a previous week of Trinity Term): Register to join this event online via Zoom

Once your registration has been approved, you will receive a confirmation email with joining instructions.

Objectives: Educational resilience is the exhibition of positive educational experiences and outcomes despite exposure to risk. It is also the product of a multidimensional interaction between the child and their immediate environment. There is a growing number of unaccompanied refugee minors worldwide seeking asylum and protection. In response, education systems in host countries must deepen their knowledge and engagement with the needs and circumstances of unaccompanied refugee minors. There is still little empirical evidence on the educational resilience of unaccompanied refugee minors.

Methods: The study analyzed the reading skills outcomes of 410 Palestinian refugee minors enrolled at UNRWA (United Nations Relief and Works Agency) schools in Jordan at age 15 in 2009 (91 of whom are unaccompanied). Using stepwise multilevel regression, this study sought to identify student-level and school-level factors that function as educational resilience correlates for Palestinian refugee minors in Jordan.

Findings: Young age, female gender, high socio-economic status, positive teacher-student relations, and exposure to structuring and scaffolding strategies were associated with higher reading skills among Palestinian unaccompanied refugee minors. Educational and school-based interventions and programs need further elaboration to account for educational resilience correlates specific to this population.

All welcome to join in person.

If you wish to join online, pre-registration is required (no need to register again if you have already done so in a previous week of Trinity Term) Register to join this event via Zoom

Once your registration has been approved, you will receive a confirmation email with joining instructions.

All welcome to join in person.

If you wish to join online, pre-registration is required (no need to register again if you have already done so in a previous week of Trinity Term): Register to join this event online via Zoom

Once your registration has been approved, you will receive a confirmation email with joining instructions.

Abstract

The universalisation of school education in India has altered the playing field, changing the expectations placed on teachers and schools to provide ‘quality’ teaching in the face of increasing diversification of the student population, growing privatisation of schooling, and what is frequently termed a ‘learning crisis’. In this context, this presentation explores some of the trends in school and teacher performance and classroom practices in two states in Southern India, using multilevel analysis of quantitative school survey data from the Young Lives study.

Policymakers, conceptualized here as principals, disagree as to whether US student performance has changed over the past half century. To inform conversations, agents administered seven million psychometrically linked tests in math (m) and reading (rd) in 160 survey waves to national probability samples of cohorts born between 1954 and 2007. Estimated change in standard deviations (sd) per decade varies by agent (m: –0.10sd to 0.27sd, rd: –0.02sd to 0.12sd). Consistent with Flynn effects, median trends show larger gains in m (0.19sd) than in rd (0.04sd), though rates of progress for cohorts born since 1990 have increased in rd but slowed in m. Greater progress is shown by students tested at younger ages (m: 0.31sd, rd: 0.08sd) than when tested in middle years of schooling (m: 0.17sd, rd: 0.03sd) or toward the end of schooling (m: 0.06sd, rd: 0.02sd). Young white students progress more slowly (m: 0.28sd, rd: 0.09sd) than Asian (m: 46sd, rd: 0.28sd), black (m: 0.36sd, rd: 0.19sd), and Hispanic (m: 0.29sd, rd: 0.13sd) students. These ethnic differences generally attenuate as students age. Young students in the bottom quartile of the SES distribution show greater progress than those in the top quartile (difference in m: 0.08sd, in rd: 0.15sd), but the reverse is true for older students. Moderators likely include not only changes in families and schools but also improvements in nutrition, health care, and protection from contagious diseases and environmental risks. International data suggest that subject and age differentials may be due to moderators more general than just the United States.

Read more: Shakeel, M.D. & Peterson, P.E. (2022). A Half Century of Progress in US Student Achievement: Agency and Flynn Effects, Ethnic and SES Differences. Educational Psychology Review.

All welcome to join in person.

If you wish to join online, pre-registration is required (no need to register again if you have already done so in a previous week of Trinity Term): Register to join this event online via Zoom

Once your registration has been approved, you will receive a confirmation email with joining instructions.

All welcome to join in person. Register to join events online via Zoom

There is large interest in intensive longitudinal data analysis in the social, educational and health sciences. Datasets can include (1) self-reports or multiple-reporter data (e.g., observed on-task behaviour, self-reported situation-specific competence) collected using diaries, experience sampling, or ecological momentary assessments, (2) task-data (e.g., trace-data, executive functioning), (3) real-time ambulatory data (e.g., accelerometer, electrodermal activity, eye-tracking), or mixtures of these. In the talk I will focus on challenges researchers face when they (i) handle and aggregate data, (ii) consider the time-structure for analysis, and (iii) specify statistical models. Time-series-based Dynamic Structural Equation Models (DSEM) using the Bayesian estimator are emerging, allowing researchers to switch focus from modelling fixed and random effects, to modelling individual processes over time. In the talk, I will illustrate intensive longitudinal data handling and modelling with on-going research, in order to highlight its relevance for understanding intraindividual processes in educational research.

All welcome to join in person. Register to join events online via Zoom

Over the past five years, the broader field of Natural Language Processing (NLP) has undergone a renaissance, driven largely by the emergence of pre-trained, word-embedding-based language models such as BERT and GPT-3, resulting in significant improvement in a variety of core NLP challenges such as sentiment analysis, machine translation, and transcription, the latter two of which have reached human-level performance. While educational applications of NLP have been a topic of research for decades, the limitations of previous NLP techniques had meant that most successful applications had been restricted to narrow domains. However, recent advances in NLP mean that challenges that had been considered prohibitively complex such as interactive chatbots, speech recognition, or automatic grading of complex open-ended responses, may now be tractable.

My specific focus of research is on the potential of NLP to assist in the formative assessment of basic literacy in low-and-middle-income countries (LMICs). In many LMICs, it is challenging to conduct high-quality, formative assessments of children’s literacy due to a variety of factors. As a result, large-scale standardized assessments, which typically consist of silently reading passages and then answering multiple-choice questions is become the de-facto method for nations to assess students’ literacy. This is a problem both because the assessment format is poorly suited for assessing basic literacy, and because the assessments are conducted infrequently and on a small sample of students, meaning the results cannot be used at the classroom level to improve instruction. In the past, more effective approaches to formative literacy assessment (e.g., oral reading, story-retell, short-answer questions), were rarely used because they were substantially more difficult and time-consuming to administer and grade.

However, given the recent advances in NLP and the proliferation of publicly available pre-trained language models, it appears feasible to partially automate the administration and scoring of formative literacy assessment. To test this, I am collaborating with a school network in Ghana to conduct a series of literacy assessments with approximately 500 of their primary school students. Students’ responses will be graded by a mix of experts and crowd workers and will be used to train language models to score student responses similar to how would human raters. The results can be used in conjunction with the school network’s pre-existing reading achievement and student demographic data to investigate both the predictive and convergent validity of open-ended questions compared to traditional measures of reading ability, as well the models’ performance relative to human raters.

Anticipated Agenda

Recent advances in Natural Language Processing (20 min)
Implications for and potential applications in Education (20 min)
NLP and formative literacy assessment: current research and initial findings (30 min)
Questions/discussion (20 min)

 

Join the event on the day, via Teams link

Foundation programs are typically a one-year preparation course to meet the language requirement or academic requirement for bachelor or master studies in British universities. This population is a relatively under-researched learner cohort in the study-abroad literature. Phrasal verbs are “two-part verbs consisting of a lexical verb followed by a contiguous (adjacent) or non-contiguous adverbial particle” (Gardner & Davies, 2007, p. 341). Phrasal verbs are omnipresent in daily natural discourse, while they pose extraordinary difficulty for L2 learners (Gardner & Davies, 2007). This longitudinal mixed-methods study collected data from over 200 international foundation year students in the UK for one academic year, to shed light on the relationship between phrasal verb development and informal language contact. Data collection took place once per term and three times in total. Phrasal verb data were collected through standardized tests and speech samples, while L2 contact was measured via language contact questionnaire, social network surveys, and semi-structured interviews. Other confounders such as initial vocabulary knowledge and personal backgrounds were also considered. Mixed-effects modelling results showed that the overall English proficiency, the corpus frequency of phrasal verbs, and the language contact of the participants were significant predictors of their phrasal verb knowledge. This study revealed the lack of interactive L2 contact and limited phrasal verb knowledge of the Chinese foundation students in the UK and proposed some implications.

About the Speaker

Siyang Zhou is a PhD candidate at the Department of Education. Siyang has a passion for languages and is interested in how study abroad benefits the vocabulary gains of international students in the UK. She obtained a BA from Sun Yat-sen University in China and masters from the University of Sydney and the University of Cambridge, specializing in language education. She has taught English in a higher vocational college for a few years in China and has accumulated rich work experiences in the education sector. She has presented her research in a few international conferences such as BAAL, AAAL, and AILA. She was awarded the Richard Pemberton Prize in the annual conference of BAAL 2019 and she is working on publications based on her PhD thesis.

Join the event on the day, via Teams link

This study investigated the association between the Manchu writing system and novice learners’ uju symbol block decoding performance. The Manchu writing system was examined on two parameters: visual complexity of uju and mapping complexity between symbol and sound. Although previous studies have shown that visual and mapping complexities were associated with word decoding, most studies were conducted on alphabetic writing systems. Thus, it remains unknown whether similar results would be found in Manchu writing system. Designed as a mixed methods study, this study randomly selected undergraduate students (n = 196) in a Chinese university, who were given Manchu uju naming tasks in Test 1 on visual complexity and Test 2 on symbol-sound mapping complexity. Moreover, the instruction on Manchu orthographic knowledge was also observed. Results of Test 1 revealed that visual complexity of uju is significantly and positively associated with participants’ decoding performance measured by error rate and reaction time. Furthermore, connected points were identified by multiple linear regression as the unique predictor for decoding accuracy, and connected points and disconnected components were identified as predictors for decoding speed. The results of Test 1 suggested that readers may use connected points to identify features in visual processing of an uju. Results of Test 2 showed that the participants’ decoding performance was not significantly different when the symbols were provided in the manipulated “mixed” and “blocked” condition, but their performance was more accurate and rapid in naming the symbol blocks with phoneme markers in the consistent one–to–one mapping than those in one-to-multiple mapping with sound. The results suggested that Manchu phoneme markers rather than uju were used for symbol block processing, which was further corroborated by the observation on Manchu orthographic knowledge instruction. Finally, this study posits that reading Manchu uju symbol blocks requires intra-symbol processing, which involves (a) at the feature level, identification and configuration of visual features into phoneme markers, and (b) at the symbol level, conversion of phoneme markers into sounds.

About the Speaker

Mark Bai Li is a final year DPhil student in Applied Linguistics and Second Language Acquisition course at the Department of Education, University of Oxford. His doctoral project investigates the cognitive process of reading the “critically endangered” Manchu language in China. Mark holds a BA in English from Dalian University of Foreign Languages, and an MSEd in TESOL from the Graduate School of Education, University of Pennsylvania. He also spent one and half years doing MPhil level research on Manchu education at King’s College London. Prior to Oxford, Mark taught as a Lecturer in English in China for seven years and as a volunteer ESOL teacher in the School District of Philadelphia for one year.

 

Join the event on the day, via Teams link

In this talk, Dr Murakami will illustrate the utility of multifactorial learner corpus research with two of their studies on the accuracy of L2 English grammatical morphemes. Drawing data from the Cambridge Learner Corpus, Murakami and Alexopoulou (2016) examined how the accuracy varies depending on a number of factors such as learners’ native language (L1), proficiency, and the specific morpheme. The study demonstrated that the acquisition order of L2 English grammatical morphemes varies across the learners’ L1 backgrounds and that L1 influence is morpheme-specific, in that some morphemes are more strongly affected by L1 than others.

While Murakami and Alexopoulou (2016) focused on accuracy differences between morphemes, it is well-known that the accuracy of morpheme use systematically varies within each morpheme as well. Murakami and Ellis (2022) investigated such within-morpheme accuracy differences. Specifically, they analysed whether morpheme accuracy in L2 writing depends on (i) availability (i.e., surface-form accuracy; e.g., asked vs requested), (ii) contingency between a surface-form and the lemma (e.g., decided vs liked), and (iii) formulaicity of the context in which the surface form occurs (e.g., since I graduated from college vs wanted a lot of). Contrary to the previous study that identified positive associations between the three distributional factors and morpheme accuracy (Guo & Ellis, 2021), our study found that only contingency is meaningfully associated with accuracy. The difference in the findings is likely to be due to the difference between the elicited imitation task used in Guo and Ellis (2021), which required the online processing of pre-determined stimuli, and free writing tasks in Murakami and Ellis (2022), in which learners decided the linguistic forms they would need.

References

  • Guo, R. & Ellis, N. C. (2021). Language usage and second language morphosyntax: Effects of availability, reliability, and formulaicity. Frontiers in Psychology, 12. 582259.
  • Murakami, A., & Alexopoulou, T. (2016). L1 influence on the acquisition order of English grammatical morphemes: A learner corpus study. Studies in Second Language Acquisition, 38(3), 365–401.
  • Murakami, A., & Ellis, N. C. (2022). Effects of availability, contingency, and formulaicity on the accuracy of English grammatical morphemes in second language writing. Language Learning.

About the Speaker

Akira Murakami a Birmingham Fellow at the Department of English Language and Linguistics, University of Birmingham, as well as a Visiting Scientist at the Natural Language Understanding Team at the Center for Advanced Intelligence Project, RIKEN. His primary research interests include second language acquisition, corpus linguistics, and quantitative data analysis for applied linguistics research. Prior to joining Birmingham in 2018, he was a post-doctoral researcher at the Universities of Birmingham, Cambridge, and Tübingen.