Tải bản đầy đủ - 0 (trang)
V-TM – A Methodological Framework for Social Science

V-TM – A Methodological Framework for Social Science

Tải bản đầy đủ - 0trang

4.4. Conclusions and Further Analyses


against their political enemies. At the same time, we observe an

increase of statements of demarcation against left- and right-wing

politics alike co-occurring with statements on right-wing demarcation.

From this, I conclude that in contexts of right-wing extremism speech

acts more often morally point to the far-left side of the spectrum or

to all extremists alike as severe issues for German democracy. This

can be seen as a hint for increased relevancy of the extremism model

and its demand for equidistance underlying the political discourse.

As far-left wing politics has lost some of its threatening character

for representative democracy, political opponents of leftist ideas not

longer demand for legal exclusion, but discredit their opponents by

putting them discursively on an equivalent level with neo-fascist or

right-wing extremist threats to democracy.

4.4. Conclusions and Further Analyses

In this chapter, I have introduced basic theoretical thoughts on the

concept of democratic demarcation. Subsequently, I have described

contents of selected relevant documents by exploring them with the

help of visualizations as SECGs showing co-occurrence of topics and

topic defining terms in single time periods. Many important issues of

contemporary history in the political debate on state and conditions of

democracy in Germany could be revealed by these exploratory descriptions. The rather qualitative insights were backed up by automatically

extracted sets of representative sentences containing important semantic propositions derived from co-occurrence patterns in single

SECGs. Qualitative findings were further extended by a quantitative

approach of machine classification to determine on the development

of content categories expressing statements on democratic demarcation. These quantitative measures of categories exhibited interesting

differences in discourse developments of perceived societal threats for

democracy from the far-left or the far-right. Moreover, it allowed for

interpretation of shifts in discourse strategies of demarcation from

left- or right-wing politics. While in recent history, demarcation from


4. Exemplary Study: Democratic Demarcation in Germany

the far-right is disputed along with legal measurements of fortified

democracy, demarcation from the far-left is expressed largely without

that reference. Instead, we can observe an increase of speech acts

equating it to the threat of the far-right side or by referencing to an

unspecific conception of ‘extremism’ as a whole.

The quantified view on categories of democratic demarcation revealed such shifts in discursive patterns and strategies. Ideally, the

interpretation of this data should be supported by a qualitative analysis. For this, sentences of each category identified by classification

can be selected and filtered by meta-data, e.g. their publication origin

and time. Selected and filtered sentences further can be clustered

by their specific content features in a separate unsupervised machine

learning process.21 Investigation of these clusters is helpful to explore

underlying discursive events of a category in a very efficient manner.

For example, we can answer questions such as: What is the reason for

the increase of right-wing demarcation statements during 2004 and

2005 in Die Zeit? Clustering of around 1,000 sentences from these

years by lexical features (uni-/bi-grams) and topic features showed

that in these years there is an intense debate on growing scenes of

neo-Nazis and right-wing extremism in Germany, resulting in electoral

successes of the party NPD which entered two federal parliaments.

At the same time, there is an intense debate on history of the Third

Reich and politics of commemoration. A further exploration of such

clustered sentences from interesting time frames, e.g. peaks in category trends, would help to improve the understanding of discourse

processes and contents allowing for inference on development of scopes

and limits of democracy.

For the hypotheses guiding this exemplary study introduced in

Section 4.1, some important insights could be generated by the TM

supported analysis of large newspaper corpora. In fact, exploration of


Clustering can be done analogue to the process of clustering years to identify

time periods presented in Section 3.2.3. But, instead of aggregating topical

features of documents from a whole year, topic proportions of single documents

can be utilized together with lexical features extracted from single sentences to

reveal clusters of contextually similar sentences.

4.4. Conclusions and Further Analyses


the retrieved sub-collection of relevant documents contained strong

reference to national-socialism and socialist regimes of the Eastern

bloc, largely framing disputes on democratic demarcation. By conceptualization of the political spectrum as a one-dimensional left-right

scale we could observe strong demarcation against far-left politics

during the first decades of the FRG, but largely diminishing after

1990. Right-wing demarcation interestingly did not emerge as a very

prominent topic on its own before the second half of the sixties, then

almost vanished, before it occurred again in the eighties and has been

dominating the dispute after the German reunification. We also saw

differences between the more conservative FAZ and the more liberal

Die Zeit, although long-term trends between the two publications

correlate to a large extent—nonetheless, correlation is much higher

on left-wing than on right-wing demarcation. Further, we saw that

reference to fortified democracy as a special version of German democracy conception is tightly coupled to positive references to democratic

identity. Yet, both categories are more frequently used in contexts

of right-wing demarcation than in demarcation towards far-left politics. This is true especially after German reunification, where we

observed that left-wing demarcation itself only plays a minor role, but

right-wing demarcation is increasingly accompanied by pointing to

problematic issues of leftist or generally ‘extreme’ positions alike. In

conclusion, the integrated interplay of qualitative and quantitative

information extracted from the large amount of hundreds of thousands

of newspaper documents allowed us a deep and insightful analysis,

giving answers to a complex, abstract research question which we

would have been unable to generate otherwise.

5. V-TM – A Methodological

Framework for Social Science

Chapter 3 has introduced a selection of Text Mining (TM) procedures

and integrated them into a complex workflow to analyze large quantities of textual data for social science purposes. In Chapter 4 this

workflow has been applied to a corpus of two newspapers to answer a

political science question on the development of democratic discourses

in Germany. In this final chapter, I extend the workflow to a general

methodological framework describing requirements, high-level design

and evaluation strategies to support Qualitative Data Analysis (QDA)

with the help of TM.

The method framework is drafted in analogy to concepts from the

field of Software Engineering (SE) or, more specific, to Requirements

Engineering (RE). In (business-oriented) computer science RE is

a well-established field of research. It is defined as “a coordinated

set of activities for exploring, evaluating, documenting, consolidating,

revising and adapting the objectives, capabilities, qualities, constraints

and assumptions that the system-to-be should meet based on problems

by the system-as-is and opportunities provided by new technologies”

(van Lamsweerde, 2007, p. 6). Heyer et al. (2014) suggested that

employing concepts from SE and RE for Digital Humanities (DH) can

help to develop interdisciplinary research designs more systematically.

Picking up this idea, I will conceptualize the methodological framework along the V-Model known from SE. The V-Model of the software

development cycle distinguishes two phases, the verification phase and

the validation phase, to coordinate development and testing activities

during the entire process (Bucanac, 1991). By assigning a testing

task to each development task around the actual implementation of

© Springer Fachmedien Wiesbaden 2016

G. Wiedemann, Text Mining for Qualitative Data Analysis in the Social Sciences,

Kritische Studien zur Demokratie, DOI 10.1007/978-3-658-15309-0_5


5. V-TM – A Methodological Framework for Social Science


Tài liệu bạn tìm kiếm đã sẵn sàng tải về

V-TM – A Methodological Framework for Social Science

Tải bản đầy đủ ngay(0 tr)