• Judith Gottschalk
4. semester, Informationsvidenskab (cand.it.), Kandidat (Kandidatuddannelse)
Winther-Nielsen (ms) lists vocabulary acquisition as one aspect of the development of Bible Online Learner which can be improved. The problem I address in this master’s thesis is the following: When learning Biblical Hebrew, what causes most learners serious problems is remembering and correctly applying the huge amounts of vocabulary. Many lexemes have several different meanings and, when translating from the Hebrew Bible, it is nearly impossible for the students to choose the meaning which correctly fits in the context. How can this problem be solved with Bible Online Learner and Learning Journey?

Learning Journey Online is, like Bible Online Learner, a web application written in PHP and JavaScript, and it can measure the learning progress of students by using Item Response Theory [IRT] within the persuasive architecture of Bible Online Learner. With IRT it is possible to statistically measure the performance of a learner in order to predict the probability of her response by establishing the position of the individual learner along a line of some latent dimension. The use of IRT is quite prominent in educational environments and in this context the latent trait is often called ability. Moreover, it is also possible to measure how difficult it is to acquire an item which is to be learned with IRT (cf. Parchev 2004: 5).

This master’s thesis is organized in an unusual way: It consists of a general introduction dealing with a case study on teaching Biblical Hebrew in a class taught by Nicolai Winther-Nielsen at Fjellhaug International University College in Copenhagen and it gives an overview about Biblical Hebrew didactics and learning strategies for learning Biblical Hebrew vocabulary. The heart of this thesis is however formed by two publishable papers on the theoretical linguistics basis for my approach to improved vocabulary learning with Bible Online Learner and Learning Journey and one on my concrete design approach which uses the theoretical foundation developed in the first paper.

In the course of my master’s studies I have participated the International Role and Reference Grammar Conference 2013 in Freiburg, Germany, and presented a paper on Learning Journey and methods to automatically derive the meaning of verbs from the corpus used in Bible Online Learner using conceptual graphs as an alternative to the classical Aktionsarten used by Van Valin (2005) in RRG. The paper is accepted for publication in the in NIHIN, an open access series of Freiburg University dealing with linguistics. My paper for the RRG proceedings is the theoretical foundation to my approach to computer-supported vocabulary learning. It deals with the question of how computer-supported vocabulary learning can be improved by using a computationally adequate model of RRG. It suggests conceptual graphs (cf. Sowa 2000) as a new approach to semantics in RRG to develop a computationally tractable version of RRG which can be implemented in a semantic parser as an extension of existing learning software for Biblical Hebrew. The application of conceptual graphs has the advantage that computational approaches for ancient languages such as Biblical Hebrew can be developed. In this approach, a linking algorithm from syntax to semantics is reduced to a set of lexical rules which match attribute value matrices defining the layered structure of the clause against an ontology (cf. Gottschalk 2010; Gottschalk 2012a; Gottschalk in press), and uses an algorithm for the automatic determination of Aktionsarten for Biblical Hebrew developed by Winther-Nielsen (ms). With this semantic parser it is possible to derive glossing for Bible Online Learner which use exact meaning of a Hebrew word depending on their morphosyntactic context. The current problem with the glossing in the corpus Bible Online Learner is using, is namely that several Hebrew words have more than one meaning and that their means are all displayed in the glossing in Bible Online Learner.

This way it is difficult for the student to learn Hebrew vocabulary by reading the Bible with Bible Online Learner as they learners cannot easily derive lexical rules which help them to determine a proper translation of a lexeme which has multiple meaning depending on its context. My hypothesis 1) regarding computer-supported vocabulary learning is that words are best learned within the morphosyntactic context they occur and to motivate the learners to explore the hypertexed corpus of the Hebrew Bible by themselves. The idea regarding an improved glossing in Bible Online Learner is: If the learners are only exposed to ideal glossings and to perfect examples they will be able to derive lexical rules for the determination of the meaning of words which have multiple meaning depending on the morphosyntactic context they occur in. This is what I discuss in the second publishable paper, which is part of this thesis: I am dealing with the question of how the learning with Bible Online Learner and Learning Journey can be improved towards computer supported vocabulary learning and how the theoretical findings from the first paper, Gottschalk (accepted), which is part of this master’s thesis, and Winther-Nielsen (ms) can be practically applied within the tools to improve facilitation and learning? The second paper presents a design on a practical approach to how learners using Bible Online Learner will be able to acquire lexical rules to determine the exact meaning of nātan in the corresponding morphosyntactic context. It will one the hand show how exercises in Bible Online Learner needs to be designed in order to enable corpus-driven language learning and on the other hand I will show how the sound learning statistics in Learning Journey can be used for vocabulary learning. Here my hypothesis is the following: While tools like Anki and Memrise use specific-fixed algorithms which determine when recently acquired vocabulary needs to be repeated in order to be learning (cf. Edge et al. 2012), Learning Journey with its statistical framework can offer a much more flexible approach which is user-driven and enables tailoring and tunneling, which are two persuasive means, which following Gottschalk and Winther-Nielsen (2013) should enable successful language learning with Bible Online Learner. In this paper I am going to discuss the algorithms used in Anki and other tools and compare them with the approach chosen in Learning Journey. More important however is my idea that learning the vocabulary of Biblical Hebrew is best possible within contextualized vocabulary learning as done with Bible Online Learner. I present a study made with Nicolai Winther-Nielsen’s class in Copenhagen and am able to show that those students who learned the vocabulary of Biblical Hebrew contextualized with Bible Online Learner were more successful in learning the language than those students who learned vocabulary decontextualized with Memrise.

With these two papers I am on the one hand developing a formally solid framework for semantic parsing and automatic meaning determination for a specific kind of verbs within a linguistic corpus and doing work in the area of theoretical functional computational linguistics and on the other hand I am developing a persuasive approach to computer-supported vocabulary acquisition within the area of applied computational linguistics, which practically enables learners to improve their language skills.
Udgivelsesdato22 dec. 2014
Antal sider100
ID: 207213665