Machine learning for spelling acquisition - How accurate is the prediction of specific spelling errors in German primary school students - Article title

Aug 3, 2026

Upload by Ivy

Germany shows substantial spelling difficulties by the end of primary school, creating a need for digital, machine-learning-supported individualisation for teachers. The study examines how accurately specific spelling errors can be predicted across different student skill levels and why incorrect predictions occur. A web application collected N=685 first- and second-graders’ spelling attempts in Bavaria, yielding 18,133 misspellings. Six machine-learning models were trained and compared, with Random Forest achieving best average prediction.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages20

Reading Time50 min

AccessFree to download

Introduction

Spelling acquisition challenges and student diversity

Teacher diagnostic needs and limitations

Study objectives and approach

Web application and dataset collection

Model training and performance comparison

Results and implications

Best-performing model and error-level findings

Linguistic ambiguity and error entanglement

Computers and Education: Artificial Intelligence 6 (2024) 100233 Contents lists available at ScienceDirect Computers and Education: Artificial Intelligence [journal homepage: www.sciencedirect.com/journal/computers-and-education-artificial-intelligence](journal homepage: www.sciencedirect.com/journal/computers-and-education-artificial-intelligence) | Machine learning for spelling acquisition: How accurate is the prediction of specific spelling errors in German primary school students?
Richard Boehmea, *, 1, Stefan Coorsa, b, 1, Patrick Ostera, Meike Munser-Kiefer a, Sven Hilbert a
a University of Regensburg, Department of Human Sciences, Sedanstrasse 1, 93055, Regensburg, Germany b Ludwig Maximilians University, Department of Statistics, Ludwigstrasse 33, 80539, Munich, Germany | | | | --- | --- | --- | | A R T I C L E I N F O | | A B S T R A C T | | Keywords:
Machine learning Deep learning
Multi-label classification Spelling
Literacy acquisition Primary school Elementary school | | In Germany (similar to other countries), 30 % of students demonstrate insufficient spelling skills at the end of primary school – partly owing to the challenge for teachers to manage a variety of students’ learning needs. Digital tools using Machine Learning can enable teachers to individualise students’ learning. However, there are still no suitable approaches for demographics of students who are not yet proficient in spelling.
With an aim to adapt Machine Learning for students of all proficiencies, we investigate how accurately specific spelling errors can be predicted across different skill levels, and what the content-related reasons for incorrect predictions are.
To that end, we developed a web application to record the spelling efforts of N = 685 first-and second-gradersin Bavaria, Germany. A total of 18,133 different misspellings were recorded. Using this dataset, we trained six Machine Learning models and compared their performances to predict misspellings.
Comparing all Machine Learning models employed in this work, the Random Forest performed best on average as a predictor of spelling errors. Errors at the syllable- and morpheme-levels were predicted best, and errors atthe basic phoneme-grapheme-level were predicted slightly less accurately. Confusions often concerned cases that are considered linguistically ambiguous or occurred in complex error entanglements. The implications of these results are discussed. | 1. Introduction Approximately 130 million people speak German as their first or second language, making it the most-spoken native language in the European Union and one of the most-spoken languages worldwide (German Federal Statistical Office, 2022). The acquisition of skills in reading and writing in this language (just as in most other languages) is critical for educational success and successful participation in modern society (G¨opferich & Neumann, 2016). In particular, the acquisition of spelling is a very demanding task, where a large proportion of children lag behind the skill level that would be expected for a given age. For example, a 2016 study from the Institute for Educational Quality Improvement (IQB) examining trends in student achievement shows that on average in Germany, more than 22 % of children (including 6 % of children with officially stated special needs) achieve neither the normal standard nor the minimum standard by the end of fourth grade (Stanat, Schipolowski, Rjosk, Weirich, & Haag, 2017). Results from the more recent 2021 IQB study (Stanat et al., 2022) show that this proportion increased to over 30 %. According to the authors, it is likely that the worsening is not exclusively the result of the COVID-19 pandemic. One reason for the poor performance in Germany is seen in the great diversity of the students’ prerequisites for learning, which can result in a variety of different learning needs (Stanat, Schipolowski, Rjosk, Weirich, & Haag, 2017). Primary school te

FAQs

What problem does the study address in German primary spelling acquisition?

A significant share of German primary students end the primary phase with insufficient spelling skills, making it difficult for teachers to manage diverse learning needs. The study targets the lack of suitable approaches to model spelling errors for students not yet fully proficient.

How was the dataset for spelling-error prediction collected?

A web application recorded spelling efforts from N=685 first- and second-graders in Bavaria. In total, 18,133 different misspellings were captured for model training and evaluation.

Which machine-learning model performed best, and at what linguistic levels were predictions strongest?

Across the models tested, Random Forest performed best on average. Predictions were most accurate for syllable- and morpheme-level errors, while basic phoneme-grapheme-level errors were predicted slightly less accurately.

Aug 3, 2026

Upload by Ivy