Orthographic Sorting for Second Language Acquisition Using MindSort
Keywords
Loading...
Authors
Issue Date
2022-08-24
Language
en
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
For foreign vocabulary acquisition, similar words are sometimes found to
be confusing, even for more advanced learning. MindSort, a brain-inspired
application for learning foreign vocabulary, focuses on this challenge of similarity.
Hereto, MindSort maps more similar words closer together on the
screen. To investigate the best way to map the words on the screen, this
study investigates the question How do participants judge orthographic similarity
between unknown words in a 2D spatial sorting task? To this end,
an experiment was conducted where 20 participants were asked to spatially
sort 15 Swahili words on a screen for two vocabulary lists, based on how
(dis)similar they judged words to be. The resulting data from this experiment
consisted of lists with word pair distances, which were compared to five
different orthographic models of word pair distances: Levenshtein distance,
Damerau-Levenshtein distance, normalized Levenshtein distance, weighted
open bigram, and extended spatial coding. To get the word pair distance
per model, the following four steps were taken: (1) implement the algorithm,
(2) calculate the distances between words in a distance matrix, (3)
map the distances to a 2D map using Multidimensional Scaling (MDS), and
(4) calculate the Euclidean distances between word pairs. We calculated the
correlation between the experimental data of the word pair distances in the
sorting of the participants and the data from the models using Spearman’s
rank correlation coefficient. Comparing these correlation for the different
models using Friedman’s test and Wilcoxon signed rank test showed that
the weighted open bigram model scored significantly better than Damerau-
Levenshtein and normalized Levenshtein. Contributing to further developments
in the field of psycholinguistics we notice that participants mentioned
sound as most important word feature for sorting besides the first letter of
the word. Because of the limited scope of this study, we suggest to do further
research to see whether our results also hold with more data. Especially the
selection of the words for the word list needs further attention for which our
research provides several starting points.
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen