nach oben

2014 | Buch

Kapitel lesen Erstes Kapitel lesen

Analyzing Discourse and Text Complexity for Learning and Collaborating

A Cognitive Approach Based on Natural Language Processing

verfasst von: Mihai Dascălu

Verlag: Springer International Publishing

Buchreihe : Studies in Computational Intelligence

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

With the advent and increasing popularity of Computer Supported Collaborative Learning (CSCL) and e-learning technologies, the need of automatic assessment and of teacher/tutor support for the two tightly intertwined activities of comprehension of reading materials and of collaboration among peers has grown significantly. In this context, a polyphonic model of discourse derived from Bakhtin’s work as a paradigm is used for analyzing both general texts and CSCL conversations in a unique framework focused on different facets of textual cohesion.

As specificity of our analysis, the individual learning perspective is focused on the identification of reading strategies and on providing a multi-dimensional textual complexity model, whereas the collaborative learning dimension is centered on the evaluation of participants’ involvement, as well as on collaboration assessment.

Our approach based on advanced Natural Language Processing techniques provides a qualitative estimation of the learning process and enhances understanding as a “mediator of learning” by providing automated feedback to both learners and teachers or tutors. The main benefits are its flexibility, extensibility and nevertheless specificity for covering multiple stages, starting from reading classroom materials, to discussing on specific topics in a collaborative manner and finishing the feedback loop by verbalizing metacognitive thoughts.

Inhaltsverzeichnis

Frontmatter

Introduction

Abstract

In every instructional situation, reading textual materials and writing down thoughts are the core activities that represent both causes (from learner’s viewpoint) and indicators of learning (from teacher’s viewpoint). Reading is a cognitive activity whose oral or written traces are usually analyzed by teachers in order to infer either learners’ comprehension or reading strategies. Hence reading and writing are core activities that every teacher has to assess on a daily basis. Reading materials have to be scaled or tailored to suit pupils’ actual level, and reading strategies have to be analyzed for inferring learners’ level of text processing and understanding.

Mihai Dascalu

Theoretical Aspects

Frontmatter

Individual Learning

Abstract

This chapter addresses individual learning by firstly considering the multiple facets of cohesion and coherence, their links to comprehension and textual complexity, also grounding a computational view to be discussed later (see 4.1 Measures of Cohesion and Local Coherence). Afterwards, in tight relation to Figure 1 from the Overview of Theoretical Aspects, textual complexity is regarded from a computational perspective, highlighting multiple approaches. Later on, self-explanations, seen as specific learner productions, are used to support the learning process by making it more efficient and more focused on comprehension (McNamara 2004).

Mihai Dascalu

Collaborative Learning

Abstract

This chapter creates a framing in term of collaborative learning, as it is focused on presenting chats that emerged as a viable alternative to the classic view of learning within Computer Supported Collaborative Learning (CSCL): Bakhtin’s dialogism (Bakhtin 1981) that defines the CSCL paradigms, computational approaches and Social Network Analysis (SNA) as the main Learning Analytics (LA) tool for modeling interaction between conversation participants.

Mihai Dascalu

Computational Discourse Analysis

Abstract

As previous chapters were overall oriented towards comprehension and productions from the perspectives of individual and collaborative learning, this chapter is focused on presenting automatic discourse analysis models and natural language processing techniques that ground a computational and quantifiable perspective of cohesion and coherence and that greatly impact the underlying functionalities of our developed systems (A.S.A.P., Ch.A.M.P., PolyCAFe and ReaderBench).

Mihai Dascalu

Empirical Studies

Frontmatter

Quantitative Analysis of Chat Participants’ Involvement

Abstract

The first experiments were performed with A.S.A.P. (Dascalu et al. 2008a, b) (see Table 9) whose purpose was to discover the most competent user in a chat using several analysis factors, starting with the simplest, such as the dimension of utterances, and ending with pragmatics issues such as speech acts (Austin 1962; Searle 1969; Trausan-Matu et al. 2004) and even social aspects of interaction between conversation participants (see 3.2 Social Network Analysis).

Mihai Dascalu

PolyCAFe - Polyphonic Conversation Analysis and Feedback

Abstract

The PolyCAFe system (Polyphonic Conversation Analysis and Feedback generation) (see Table 14) (Trausan-Matu et al. in press) was designed, implemented and validated within the FP7 2008-212578 LTfLL - Language Technologies for Lifelong Learning project (Trausan-Matu et al. 2008; Trausan-Matu et al. 2010a; Trausan-Matu et al. 2011). Moreover, it represented the starting point of the joint work between the two universities (University Politehnica of Bucharest and University Grenoble Alpes), as both were partners in the same work package.

Mihai Dascalu

ReaderBench (1) - Cohesion-Based Discourse Analysis and Dialogism

Abstract

In contrast to the previous systems that focused on analyzing CSCL conversations, ReaderBench (Dascalu et al. 2013a; Dascalu et al. 2013b) (see Table 19) addresses a wider spread of activities and can be used within more complex educational scenarios (see 10.3 Educational Implications). Nevertheless, A.S.A.P., Ch.A.M.P and PolyCAFe have provided valuable insight and some features were reused, of course with the necessary improvements.

Mihai Dascalu

ReaderBench (2) - Individual Assessment through Reading Strategies and Textual Complexity

Abstract

As an overview, in terms of individual learning, ReaderBench encompasses the functionalities of both CohMetrix (McNamara et al. 2010) (see 2.2.2 Textual Complexity Computational Approaches) and iStart (McNamara et al. 2007a; Graesser et al. 2005) (see 2.3 Reading Strategies), as it provides teachers and learners information on their reading/writing activities: initial textual complexity assessment, assignment of texts to learners, capture of metacognitions reflected in one’s textual verbalizations, and reading strategies assessment (a detailed comparison is presented at the end of this chapter).

Mihai Dascalu

ReaderBench (3) - Involvement and Collaboration Assessment through Cohesion and Dialogism

Abstract

Although participants’ involvement in chat environments has been studied in previous systems, as mentioned in Overview of Empirical Studies, ReaderBench has brought a series of remarkable improvements in terms of collaborative learning:

Emphasis and better support of the dialogical and polyphonic model previously proposed in PolyCAFe with new visualizations and evaluation factors.
Refinement of the initial collaboration assessment model (Trausan-Matu et al. 2012b; Dascalu et al. 2010a) based on the social knowledge-building effect, through the use of the cohesion graph (Trausan-Matu et al. 2012a; Dascalu et al. 2013b).
A novel collaboration evaluation model based on the overlapping effect of voices seen as semantic chains (see 7.5 Dialogism and Voice Inter-Animation) pertaining to different participants.
The validation of the evaluation mechanics on a long-term discussion group, seen as an aggregation of multiple threads across a longer timespan, and not only the assessment of individual chat conversations (Nistor et al. 2013a; Nistor et al. 2013b, submitted; Nistor et al. 2013c).

Mihai Dascalu

Discussions

Abstract

Starting from the integrated view presented in Figure 1, we have designed a cohesion graph (see 7.2 Cohesion-based Discourse Analysis) that was later on used to aggregate individual and collaborative learning through the underlying discourse structure. Without limiting the overall perspective, we opted here for focusing solely on ReaderBench, as it introduced the integrated cohesion-based analysis addressing both individual and collaborative learning, incorporating and refining nevertheless more functionalities than the other systems.

Mihai Dascalu

Conclusions

Abstract

The development of multiple systems, a constant growth in terms of the complexity of the approach, the multitude of considered factors, the unified approach that addresses both general texts and conversations and the emphasis on providing effective support for tutors and students in their learning and CSCL activities, are just the highlighting points of our research.

Mihai Dascalu

Backmatter

Titel: Analyzing Discourse and Text Complexity for Learning and Collaborating
verfasst von: Mihai Dascălu
Verlag: Springer International Publishing
Electronic ISBN: 978-3-319-03419-5
Print ISBN: 978-3-319-03418-8
DOI: https://doi.org/10.1007/978-3-319-03419-5