The benefits of observing a model when acquiring a new motor skill are well known, however, there is little research on the influence of viewing angle of the model. The purpose of the present experiment was to assess whether a looking-glass (face on) or subjective (facing away) viewing angle would result in different acquisition and retention levels when learning a complex Zumba dance sequence. Greater cognitive effort was expected during the looking-glass condition, consequently resulting in slower acquisition but greater physical performance scores and error recognition/identification. Thirty females were evenly divided into the looking-glass or subjective group and began with the pre-test phase to assess degrees of motivation, self-efficacy, and physical performance. Participants were then lead through six acquisition dances, within which they performed the to-be-learned sequence 18 times. An assessment of cognitive effort followed, then post-test performances and error recognition/identification scores were obtained to conclude the study. While both the looking-glass and subjective conditions demonstrated equal rates of acquisition (p>.05), the looking-glass group performed significantly fewer errors during the post-test (p<.05) and were significantly better at identifying errors when a video of the dance sequence was shown from the same viewing angle as the acquisition phase (p<.05). No differences were reported between the two conditions with respect to cognitive effort (p>.05). Based on the results of this study, the looking-glass viewing angle appears to result in better learning of a dance sequence, but cannot be explained by cognitive effort.

The benefits of observing a model when acquiring a new motor skill are well known, however, there is little research on the influence of viewing angle of the model. The purpose of the present experiment was to assess whether a looking-glass (face on) or subjective (facing away) viewing angle would result in different acquisition and retention levels when learning a complex Zumba dance sequence. Greater cognitive effort was expected during the looking-glass condition, consequently resulting in slower acquisition but greater physical performance scores and error recognition/identification. Thirty females were evenly divided into the looking-glass or subjective group and began with the pre-test phase to assess degrees of motivation, self-efficacy, and physical performance. Participants were then lead through six acquisition dances, within which they performed the to-be-learned sequence 18 times. An assessment of cognitive effort followed, then post-test performances and error recognition/identification scores were obtained to conclude the study. While both the looking-glass and subjective conditions demonstrated equal rates of acquisition (p>.05), the looking-glass group performed significantly fewer errors during the post-test (p<.05) and were significantly better at identifying errors when a video of the dance sequence was shown from the same viewing angle as the acquisition phase (p<.05). No differences were reported between the two conditions with respect to cognitive effort (p>.05). Based on the results of this study, the looking-glass viewing angle appears to result in better learning of a dance sequence, but cannot be explained by cognitive effort.

The benefits of observing a model when acquiring a new motor skill are well known, however, there is little research on the influence of viewing angle of the model. The purpose of the present experiment was to assess whether a looking-glass (face on) or subjective (facing away) viewing angle would result in different acquisition and retention levels when learning a complex Zumba dance sequence. Greater cognitive effort was expected during the looking-glass condition, consequently resulting in slower acquisition but greater physical performance scores and error recognition/identification. Thirty females were evenly divided into the looking-glass or subjective group and began with the pre-test phase to assess degrees of motivation, self-efficacy, and physical performance. Participants were then lead through six acquisition dances, within which they performed the to-be-learned sequence 18 times. An assessment of cognitive effort followed, then post-test performances and error recognition/identification scores were obtained to conclude the study. While both the looking-glass and subjective conditions demonstrated equal rates of acquisition (p>.05), the looking-glass group performed significantly fewer errors during the post-test (p<.05) and were significantly better at identifying errors when a video of the dance sequence was shown from the same viewing angle as the acquisition phase (p<.05). No differences were reported between the two conditions with respect to cognitive effort (p>.05). Based on the results of this study, the looking-glass viewing angle appears to result in better learning of a dance sequence, but cannot be explained by cognitive effort.

Grading assignments is a time-consuming part of teaching translation. Automatic tools that facilitate this task would allow teachers of professional translation to focus more on other aspects of their job. Within Natural Language Processing, error recognitionhas not been studied for human translation in particular. This thesis is a first attempt at both error recognition and classification with both mono- and bilingual models. BERT– a pre-trained monolingual language model – and NuQE – a model adapted from the field of Quality Estimation for Machine Translation – are trained on a relatively small hand annotated corpus of student translations. Due to the nature of the task, errors are quite rare in relation to correctly translated tokens in the corpus. To account for this,we train the models with both under- and oversampled data. While both models detect errors with moderate success, the NuQE model adapts very poorly to the classification setting. Overall, scores are quite low, which can be attributed to class imbalance and the small amount of training data, as well as some general concerns about the corpus annotations. However, we show that powerful monolingual language models can detect formal, lexical and translational errors with some success and that, depending on the model, simple under- and oversampling approaches can already help a great deal to avoid pure majority class prediction.

Gegenstand der Arbeit ist zunächst eine Analyse der didaktischen Anforderungen an Sprachlernsysteme, die sich zum Teil aus dem Fremdsprachenunterricht ergeben. Daraus ergibt sich ein Übungstyp, der vom Lerner eine frei gestaltete Eingabe erfordert und damit insbesondere die kommunikative Kompetenz fördert, der aber auch mit Hilfe computerlinguistischer Methoden realisiert werden kann. Anschließend wird zur Auswahl einer geeigneten Grammatiktheorie insbesondere die Lexical Functional Grammar (LFG) näher betrachtet. Die Theorie muss sich aus computerlinguistischer Sicht für eine Implementierung im Rahmen eines Sprachlernprogramms eignen und es ist von zusätzlichem Vorteil, wenn die verwendeten Konzepte denen in Lernergrammatiken ähneln, um so die Generierung von Rückmeldungen zu vereinfachen. Im darauf folgenden Abschnitt wird kurz das eigentliche Programm PromisD (Projekt mediengestütztes interaktives Sprachenlernen - Deutsch) vorgestellt, wie es sich auch dem Nutzer präsentiert. Schließlich wird ein so genanntes antizipationsfreies Verfahren entwickelt, bei dem weder in der Grammatik noch im Lexikon Informationen zur Fehleridentifizierung enthalten sind. Die Fehlererkennung wird dabei auf die Bereiche eingeschränkt, in denen sich in einem Lernerkorpus häufig Fehler zeigen, um einerseits wesentliche Fehlertypen abzudecken und andererseits eine größere Effizienz bei der Analyse von realen Eingaben zu erreichen. Die Vorstellung des Verfahrens unterteilt sich entsprechend den grundlegenden Struktureinheiten der LFG in zwei Bereiche: die Konstituentenstruktur mit einer modifizierten Form des Earley-Algorithmus zur Integration von Fehlerhypothesen in die Chart und die Feature-Struktur mit einer veränderten Unifikationstrategie zur Behandlung und Speicherung von sich widersprechenden Werten in F-Strukturen. Zum Abschluss erfolgt die Evaluation und es werden die Möglichkeiten zur Gestaltung einer Rückmeldung an den Lerner diskutiert. / The dissertation starts with an analysis of the requirements for Intelligent Computer-Assisted Language Learning systems (ICALL), which partially depend on didactic aspects of foreign language teaching. Based on this a type of exercise can be identified, that on the one hand allows the learner to enter free formed input supporting the so called communicative competence as a major didactic goal and on the other hand may be realised with advanced computational linguistics'' methods. In the following chapter a look at grammar theories and especially Lexical Functional Grammar (LFG) is taken. The grammar theory needs to be tractable in an implementation and it is of a further advantage if the concepts of the theory are similar to the concepts in learner grammars in order to simplify the generation of feedback. Subsequently the user interface of the actual program is presented with a focus on error messages. The implementation is named PromisD, which stands for "Projekt mediengestütztes interaktives Sprachenlernen - Deutsch". Finally an anticipation-free parsing method is developed using neither information from the lexicon nor the grammar in order to identify grammar errors. The recognition is restricted to those areas where errors occur frequently in a learner corpus in order to allow for a greater efficiency parsing authentic data. Along the two structural levels in LFG the presentation of the algorithm follows: the constituent-structure with a modified Early-algorithm integrating error hypotheses into the chart and the feature-structure with a new unification-strategie storing information about clashing values in the f-structure. The dissertation closes with an evaluation and an outlook on the generation of error messages.