Learning Analytics and Spelling Acquisition in German – The Path to Individualization in Learning

Abstract

This paper shows how Learning Analytic Methods are combined with German orthography in the IDeRBlog-project (www.iderblog.eu). After a short introduction to the core of the platform – the intelligent dictionary – we focus on the presentation and evaluation of a new training format. The aim of this format is, that pupils can train misspelled words individually in a motivating and didactic meaningful setting. As a usability test was run with twenty one third graders, we are able to present the results of this evaluation.

Full text

Draft – originally published in: Ebner M., Edtstadler K., Ebner M. (2020) Learning Analytics and Spelling
Acquisition in German – The Path to Individualization in Learning. In: Zaphiris P., Ioannou A. (eds) Learn-
ing and Collaboration Technologies. Designing, Developing and Deploying Learning Experiences. HCII
2020. Lecture Notes in Computer Science, vol 12205. Springer, Cham. pp. 317-325.
https://doi.org/10.1007/978-3-030-50513-4_24
Learning Analytics and Spelling Acquisition in German –
the Path to Individualization in Learning
Markus Ebner2[0000-0002-5445-1590] and Konstanze Edtstadler1 and Martin Ebner2[0000-0001-
5789-5296]
1 University College of Teacher Education Styria, Institute of Early Childhood and Primary
Teacher Education, Hasnerplatz 12, 8010 Graz, Austria
2 Graz University of Technology, Department Educational Technology, Münzgrabenstraße 36/I,
8020 Graz, Austria
Konstanze.Edtstadler@phst.at
Abstract.
This paper shows how Learning Analytic Methods are combined with German
orthography in the IDeRBlog-project (www.iderblog.eu). After a short introduc-
tion to the core of the platform – the intelligent dictionary – we focus on the
presentation and evaluation of a new training format. The aim of this format is,
that pupils can train misspelled words individually in a motivating and didactic
meaningful setting. As a usability test was run with twentyone third graders, we
are able to present the results of this evaluation.
Keywords: Learning Analytics; German orthography; qualitative analysis of
misspellings; Technology Enhanced Learning; Educational media; K-12 educa-
tion.
1 Introduction
During the last years IDeRBlog was used in many schools in German speaking coun-
tries as well as schools who offer German language courses in foreign countries, there-
fore we were able to gain a better insight in the acquisition of German orthography [4],
[5], [6], [7]. Due to the success of the project a follow up project called IDeRBlog ii
was started with the following goals:
1. further implementation in schools in order that more students can benefit from the
platform;
2. further individualization of the training material to foster the orthographic compe-
tence;
3. extension of the number of words, which give a specific feedback for correcting the
mistakes (see following chapters).

2
The purpose of this paper is to show, how the individualization is achieved and how the
extension of the words is accomplished.
1.1 The IDeRBlog-Platform
German orthography is known to be quite difficult to master, especially for primary-
and secondary-school pupils. Therefore, we try to make writing and spelling activities
as attractive and useful as possible. Our system is an attempt to bridge Learning Ana-
lytics (LA) research and daily writing activities for pupils aged from eight years on by
offering a web-based platform with various functions, some of them we will present
later in this paper. The core system is designed to support the acquisition of German
orthography during the text-writing process - that is focused on writing blog entries -
as well as with individualized material when focusing exclusively on problematic or-
thographic areas.
For assisting children during the text writing process, the developed web-based plat-
form for German-speaking users offers an “intelligent dictionary” [5]. This is the at-
tempt to combine spelling and LA. The intelligent dictionary works on different levels:
a) During the text-writing process the intelligent dictionary provides specific feed-
back that encourages pupils to think about the spelling and to correct it when an
orthographic mistake occurred. In contrast to a conventional auto correction sys-
tem, which only provides information that the word is (probable) wrong and
may suggest the correct word or a list of possibly words, our systems helps to
gain deeper insights in the system of German orthography and its acquisition.
b) After and besides the text-writing process the intelligent dictionary supports
teachers and students to train problematic orthographic areas specifically. To do
so, teachers and pupils are provided with a qualitative analysis based on the
occurred orthographic mistakes identified by the intelligent dictionary. Conse-
quently, teachers and students can focus on specific explanations and selected
(predefined) online and offline exercises, which are provided by the system.
c) For an even more specific training of the German orthography the new approach
is to develop new training formats, which are even more specific. In contrast to
predefined exercises, the new training formats should be generated automati-
cally based on the occurred mistakes. For example, an exercise is developed that
trains exactly the occurred misspelled words in combination with a meaningful
didactic approach.

3
2 Individualization
2.1 Extension of the Intelligent Dictionary
In order to improve the platform, we need to extend the intelligent dictionary. Selecting
the words for this extension is quite hard. Therefore, we decided on the following strat-
egies for the word selection:
The first strategy is, that the texts written between October 2016 and October 2019
are analyzed in order to select the most frequently used words either spelled correctly
or incorrectly. Consequently, they are implemented in the intelligent dictionary by de-
riving all their word forms and the mistakes corresponding to the categories.
The other strategy is that teachers could suggest words that should be implemented
in the intelligent dictionary. For this purpose, an interface was designed that should
encourage teachers to participate in order to gain a more individualized intelligent dic-
tionary (see Fig. 1). If teachers discover important words while correcting/reading blog
entries that have not yet been categorized, they can suggest them for inclusion easily.
They just need to click on the encountered word that is used by a pupil in a blog entry
and tick the box: “I think that this word is important and suggest it for inclusion in the
in the intelligent dictionary.” Furthermore, they can – but need not to – fill in the form
by mentioning the intended word for a misspelled word and suggest the category of the
mistake.
PLACE Fig. 1 HERE
Fig. 1. Interface for suggesting words to be included in the intelligent dictionary. Translation:
(1) Editing the Category of Mistakes; (2) Here you can edit the category of a selected word, (3)
Misspelled word:, (4) crazy; (5) Intended word (please fill in); (6) Feedback; (7) Possible mis-
take encountered; (8) Note for correcting (optional); (9) I think that this word is important and
suggest it for inclusion in the in the intelligent dictionary.; (10) Editing the category (optional);
(11) Please choose a category (optional)
2.2 A New Training Format
When training the German orthography, it is necessary to consider two approaches:
One is to focus on problematic orthographic areas based on the qualitative analysis of
mistakes. This was the focus of the first IDeRBlog-project (see chapter The IDeRBlog-
Platform). The other one is to train the specific words that are spelled incorrectly. This
is a more individualized way that will be implemented in the second IDeRBlog-project.
Example of a new training format: Incorrectly spelled words.
In order to train incorrectly spelled words, a specific didactic approach is necessary.
For example, it is not enough to just read the words or write them several times. Rather
the words should be trained by applying the strategy of conscious copying as described

4
in several publications on spelling acquisition [1]: In contrast to a normal copying pro-
cess that is probably based on a letter by letter approach a conscious copying process is
characterized by taking into account specific problematic positions in a word. As chil-
dren with spelling problems are often not aware of the problematic areas the following
format was developed. Although the training format is based on this general way of
training [1] the digital training format is - in contrast to the paper-and-pencil-setting -
enriched with various features.
Selection of the words to be trained:
One of the advantages of the digital training format is, that the words that need to be
trained are chosen automatically by the system based on the occurred mistakes when
writing a blog entry in the IDeRBlog-Platform. This guarantees that the children are
only confronted with the words that are important for them and releases the teachers
from selecting the words.
Basic way of training:
In the first step, the child is confronted with the correctly spelled word. The child is
encouraged to look closely at the word and to think about possible problematic posi-
tions in the word.
PLACE Fig.2 HERE
Fig. 2. Screenshot of the first step for the word ähnlich ‘similar’ [2]
As can be seen in the screenshot (see Fig. 2) the child is led trough the by the orders.
On top the orders say: “Read these words closely! Think about problematic ortho-
graphic positions. Highlight the positions that seem difficult to you!”
After highlighting the possible positions, the order says:” In case you are sure how
to spell these words, click here!”
In the second step (see Fig. 3), the child should write the word by heart without
seeing the correctly spelled word. This prevents a letter-by-letter-copying approach.
The child is encouraged to remember how the possibly problematic position of the word
are spelled correctly. As soon as the child thinks, that she/he is ready for writing she/he
can press the button “hide word” in order to get to step three.
PLACE Fig. 3 HERE
Fig. 3. Screenshot of the second step [2]
In the third step, the child gets a feedback on the spelling attempt in contrast to the
correct spelling. In case, the spelling attempt is correct, the spelled word and the target
word appear in green. In case the spelling attempt is not correct, the incorrectly spelled
word appears in red and the target word in green (see Fig. 4). Consequently, the child
is encouraged to look again closely at the correctly presented word.

5
PLACE Fig. 4 HERE
Fig. 4. Screenshot of the third step in case the word is spelled incorrectly
Possible ways of modification:
In order to create more attractive exercises that also fit better the needs of different
children, we offer several ways to modify the format. The settings can be changed by
the teacher for each pupil individually. The possibilities for changing the settings are:
Number of presented words: Up to five words can be presented at the same time in
the first step. We suggest the following settings: Easy level: one word; medium level:
three words; most difficult level: five words;
Marking of problematic positions:
In order to lead the children to a conscious approach, they can mark specific posi-
tions of the word by clicking on them in the first step. Consequently, they appear high-
lighted in red. In case, children do not want or need not to highlight anything, this func-
tion can be deactivated.
Time limit:
It is possible to choose a time limit as well as for the presentation of the correct word
in step 1 as well as for typing the word by heart in step 2. For both, the time limit can
be freely defined in number of seconds. Zero seconds means that no time limit should
be applied. It is suggested that for the easiest level no time limit is defined.
Presentation of misspelled words:
In step 3 the children get a feedback on their spelling attempt by presenting the
spelled word and the target word. As some teachers do not like to confront the children
with misspelled words, it is possible to only show the correctly spelled word. Concern-
ing the presentation of the spelled and the intended word, it is also possible to define a
time limit. Our suggestion is to not give a time limit.
3 Evaluation of the Training Format
According to [9] the interaction with a game by the user requires a constant cycle of
hypothesis formulations, testing and reworking during the play. This requires instant
feedback while the game is running. E.g. to easy games are not engaging the learning
for a long time. The game must adjust itself to the skillset of the player.

6
To ensure a good quality of the game we conducted a field test to detect possible
problems or bad game design before the official release, which we will describe in the
following section.
Further, using the Training Formats on PC or iPads comes also with some points to
keep in mind as a teacher [10]:
• check if the existing infrastructure is able to support you in a meaningful way,
• make sure that the login procedure is understood by everyone,
• let them play with the app so that they can get familiar with it,
• take your time to assist the children
• children are open minded – assist them in their creativity
• provide them with immediate feedback when they have done the exercises
• competition is not the priority; collaboration should be the goal.
During our field test wo could further train the teachers to make them become better in
using digital exercises as support in teaching.
3.1 Method
In a first step the training format was designed in an interdisciplinary team consisting
of professionals of the area of language education (Konstanze Edtstadler and Elisabeth
Herunter from University College of Teacher Education Styria, Graz, Austria) and me-
dia informatics (Markus Ebner, Marko Burazer and Markus Friedl from Graz Univer-
sity of Technology, Graz, Austria). This interdisciplinary work is crucial in order to
offer training formats that are not only database-based and online, but also didactically
valuable formats following the research on the acquisition of German orthography. In
a second step, the training format was programmed in a pre-version. Before this pre-
version can get online, it is necessary to conduct a usability test.
This usability test was conducted in June 2019 with twentyone 3rd graders (eleven
girls and ten boys) of the school that is part of the Teacher Training College Styria. The
children were supervised by their class teacher and the group of researchers assisted the
children during the whole usability test (90 minutes).
As more training formats than the described one were tested on this day, this format
was run on four computers in two different difficulty levels. The easy level was pre-
pared with one word without time limitation for reading and writing and the possibility
to highlight problematic positions. The harder level was prepared with two words at the
same time, 20 seconds time for reading and 40 seconds time for writing. Of course, the
time limit could be exceeded [2].
The children received for each training format a questionnaire with four statements
in German [2]:
1. I knew what the task was. (Ich wusste, was die Aufgabenstellung war.)

7
2. I could solve the task myself. (Ich konnte die Aufgabe selber lösen.)
3. It was fun to practice. (Es hat mir Spaß gemacht zu üben.)
4. I also want to do the exercise at home. (Ich möchte die Übung auch zuhause ma-
chen.)
For these statements one of five smileys needs to be chosen, ranging from 1 (very
happy) to 5 (very sad). Therefore, the lower the lower the average level the better the
feedback.
4 Results
All twentyone pupils gave feedback, although the number of questionnaires is vary-
ing because sometimes two children filled out one questionnaire together or decided on
the same feedback on two different feedback sheets [2].
Table 1. Results of the feedback on the two different difficulty levels [8]
Statement
Easy level
Hard level
I knew what the task was.
1.41
1.74
I could solve the task myself.
1.41
1.53
It was fun to practice.
1.35
1.89
I also want to do the exercise at home.
1.71
2.47
4.1 Interpretation
It is clear that the easier and the harder level differ from each other. In fact, the harder
level seems to be more exhausting and more error-prone. This is also derivable from
the log-data. Concerning the easy mode out of 17 sessions 7 were conducted without
any mistake whereas in the hard mode out of 24 sessions no one was error-free.
4.2 Further findings
This training formats offers a number of possibilities for analyzing the way children
encounter spelling problems. As children can highlight problematic positions of the
word to spelled, we can analyze the most problematic positions of a word from the
learner’s perspective. For example, in the word “ähnlich” (which means similar) the
<h> was marked in five sessions, the <ch> in three sessions and the <ä> in two sessions.
Looking at the spelling attempts after presenting the word for memorizing we can
see the following mistakes in the easy mode: ahndlieh, ahnlicheres, anhlich (2x), anlich
and the following mistaktes in the harder mode nothing (7), ählich, ähnlichere, Än,
änhlich, Änlhich, Änli, änlich (4), Blege (cf. Burazer Masterarbeit). In contrast to easy
mode children do not only make more mistakes in total but also show a greater variety
of spelling mistakes. The “nothing” answers point in the direction that either the work-
ing memory was overwhelmed or that the time for typing was running out.

8
5 Outlook
Currently we are working to develop additional training formats and to extend the
intelligent dictionary by analyzing the texts written so far. To speak frankly, our system
will not be able to cover all the words pupils use when writing blog-entries, but we are
able to show that the combination of Learning-Analytic-Methods and German orthog-
raphy is possible and fruitful. Furthermore, we offer a lot of trainings for teachers – in
class and online – in order to disseminate the project.
Further, we are working on the extension of the intelligent dictionary by analyzing
the texts written so far. Concerning the possibility to suggest words for the intelligent
dictionary the results so far are quite disappointing as teachers do not make use of this
possibility, although the feature has been promoted via newsletters and webinars. Since
the activation of the feature in October 2016 only 83 suggestions have been made.
Acknowledgements
The IDeRBlog-project and the IDeRBlog-ii project are funded by the European
Commission in the framework of Erasmus+ (IDeRBlog: VG-SPS-SL-14-001616-3,
2014-2017; IDeRBlog-ii: VG-IN-SL-18-36-047317, 2018-2021).
Project-team of IDeRBlog-ii: Germany: LPM Saarland (M. Gros = coordinator, N.
Steinhauer); Gebundene Ganztagsschule Dellengarten (S. Pfeifer, J. Gregori); Austria:
PH Steiermark (K. Edtstadler, E. Herunter); TU Graz (M. Ebner, M. Ebner); Belgium:
Gemeindeschule Raeren (A. Huppertz, V. Kistemann);
References
1. Sommer-Stumpenhorst, N.: Lese-und Rechtschreibschwierigkeiten: vorbeugen und über-
winden. In: Cornelsen Scriptor (1991).
2. Burazer, M., Ebner, M., Ebner, M.: Implementation of Interactive Learning Objects for Ger-
man Language Acquisition in Primary School based on Learning Analytics Measurements.
In Proceedings of EdMedia: World Conference on Educational Media and Technology, Am-
sterdam (in print, 2020).
3. Friedl, M., Ebner, M., Ebner, M.: Mobile Learning Applications for Android und iOS for
German Language Acquisition based on Learning Analytics Measurements. In: International
Journal of Learning Analytics and Artificial Intelligence for Education (in print, 2020).
4. Ebner, M., Edtstadler, K., Ebner, M.: Learning Analytics and Spelling Acquisition in Ger-
man - Proof of Concept. in P. Zaphiris, & A. Ioannou (Hrsg.), Learning and Collaboration
Technologies. Technology in Education: 4th International Conference, LCT 2017, Held as
Part of HCI International 2017, Vancouver, BC, Canada, July 9-14, 2017, Proceedings, Part
II (S. 257-268). Cham: Springer International Publishing AG . (2017).
https://doi.org/10.1007/978-3-319-58515-4_20
5. Edtstadler, K., Ebner, M., Ebner, M.: Improved German Spelling Acquisition through
Learning Analytics. eLearning papers , 45, 17-28. (2015).

9
6. Edtstadler, K., Ebner, M., Ebner, M., Gros, M., Steinhauer, N., Peifer, S., Gregori, J., Her-
unter, E., Huppertz, A., Kistemann, V.: Analysis of Misspellings in German Orthography
in Grade 3 to 6. Postersitzung präsentiert bei European Dyslexia Autumn Seminar, Mün-
chen, Deutschland. (2018).
7. Ebner, M., Edtstadler, K., Ebner, M.: Tutoring writing spelling skills within a web-based
platform for children. Universal Access in the Information Society. (2017).
https://doi.org/10.1007/s10209-017-0564-6
8. Burazer, M.: Implementation of Interactive Learning Objects for German Language Acqui-
sition in Primary School based on Learning Analytics Measurements, Master Thesis at Graz
University of Technology, Graz, Austria, (2019).
9. Van Eck, R.: Digital game-based learning: It's not just the digital natives who are restless.
EDUCAUSE review, 41(2), 16. (2006).
10. Ebner, M., Schönhart, J., & Schön, S.: Experiences With iPads in Primary School. Profeso-
rado, revista de currículum y formación del profesorado, 18(3), 161-173. (2014).