An Alternative Method and Display for Content Analysis: Textual Visualization

The purpose of this study is to describe the usage of textual-based visualization methods as an alternative method in qualitative researches and especially in content analysis process. Qualitative researches –like social media analysis, linguistic or communicational studies that statistical methods can not be applied directly- necessitate usage of special techniques to determine patterns and structures. Textual based visualization techniques allows researcher to analyze the unstructured corpus, explore the inner relations and thus provide meaning to these texts. Decoding this structure can be achieved with content analysis method, and findings can be represented by using matrices or diagrams. However, cases of multidimensional data requires advanced visualization techniques. Data visualisation is graphical edition and common product of the fields of statistics, information technologies, broadcasting and visual communication design. Data of the qualitative research findings about determining the effectiveness and further user expectations of Kastamonu University Kuzeykent Campus wayfinding system were used for analysis and visualization. The assessments of 20 participants are collected by using a semi-structured interview form developed by the researcher. The findings have been subjected to content analysis and displayed with matrice table. These process steps were repeated with three different web-based/online visualization tool.


Introduction
Data visualization is regulating the numerical or verbal data with graphical representation methods such as charts, maps, graphs, diagrams and tables. With the help of visualization applications; it is possible to process the findings obtained from the researches and make them meaningful, to analyze the structure and relations of the contents, to question the premises and hypotheses. By doing so, it becomes easier to understand the data in complex structure. "When content is illustrated with pictures or figures, the information can be maintained in the mind of the viewer over a period of time" (Fountas and Pinnell, 2001). Arrangements such as timelines, Venn diagram, periodic table; are the best-known examples of data visualization."Many useful graphical displays result from application of multidimensional statistical analysis techniques such as factor analysis (including principal components analysis) and clustering" (Zinovyev, 2011).
Visualization is perceived much faster than intensive text or numerical data can be. In addition to the advantages it provides to the viewer, it also allows its user to be able to analyze the data with all of its dimensions. In addition to the direct involvement of the user, the main advantages of visual data exploration over automatic data mining techniques from statistics or machine learning are: • visual data exploration can easily deal with highly inhomogeneous and noisy data, • visual data exploration is intuitive and requires no understanding of complex mathematical or statistical algorithms or parameters" (Keim, 2002).

Textual Visualization
Social media analysis, linguistics/communication studies, open-ended interviews; are unstructured textual data involving subjective interpretations of people. Analyzing the words in the text, their repetition frequency and word sequences; leads to an understanding of the structure and thus, to make the text meaningful. Meirelles (2013) states that a text is a categorical type of data and making graphical representations of such data are more difficult because there is no ordering/proportionality in the construction of a text. For this reason, special methods are needed to visualize the qualitative research results. Cao and Cui (2016) classify text visualization techniques according to their purpose as follows: (1) for visualizing document similarity, (2) for revealing content, (3) for visualizing sentiments and emotions of the text, (4) for exploring document corpus, and (5) for analyzing various domain-specific rich-text corpus, such as social media data, online news, emails, poetry, and prose.
In the application phase of the study,it is aimed to analyze syntactic and lexical structures of the visualized text.For this reason, an explanation is given about visualization tools that provides syntactic and lexical analysis of text.

Relational decoding of a word (Visualizing syntactic structure)
Selected word constitutes the root of visualization and this root diverges to branches enabling the analysis of word with the text. That word tree connects the root word and related sentences with arrows. Word frequencies affect the sequence of branches, their position and size. Punctuation, repeated words and conjunctions together constitute the context, for this reason, reducing the content of the text creates negative results. "A word tree is a graphical version of the traditional keyword-in-context method, and enables rapid querying and exploration of bodies of text" (Wattenberg and Viégas, 2008).

Word's frequency in text (Visualizing lexical structure)
This visualization shows words in a cloud form. Words' proportional size depend on their frequency in text. For word cloud, the determining element is the font size of words. Words are categorical data and unlike quantitative data, they can not be classified with color values. Typeface and composition provide additional effects. The number of words can adversely affect legibility, therefore eliminating less frequent words can be considered. "Word clouds are useful when you are trying to get a quick and rough sense of some of the dominant keywords used in the text. They can be an option for working with qualitative data during the data exploration stage, more so as a means for reporting analysis to others" (Kirk, 2016).

Methods and Procedures
This study was designed to provide an alternative analysis and presentation method recommendation to the content analysis process. For this purpose, the data gathered from the opinions of the users of Kastamonu University Kuzeykent Campus about the wayfinding process and signs are used. Descriptive research data gathered during 2014-2015 Fall-Winter Semester are evaluated with content analysis, and then, organized with application of visualization methods. Two main terms about gathered data, wayfinding and signage/wayfinding system can be explain as: -Wayfinding refers to information systems that guide people through a physical environment and enhance their understanding and experience of the space (SEGD, 2014).

Online Journal of Communication and Media Technologies
Special Issue: December 2017 16 -Signage/wayfinding system consist of assistant tools such as maps, signs, arrows and pictograms, kiosks, floor/building plans, landmarks which designed and combined to enable people to reach their destinations independently, easily and without wasting time.

Content Analysis Procedure
Interviews, as a qualitative method, are conducted in order to compile campus users' views on their way finding process in terms of their perspective, experiences, feelings and perceptions. The data were collected by using the interview form developed by the researcher in Turkish. As a result of the literature review made during the question preparation phase; signage/way finding systems have seen to be evaluated on the basis of syntactic, semantic and pragmatic levels (Teker, 2002). Designated concepts are transferred to everyday speech; the qualities such as noticeability, legibility, and sufficiency of the signs have been questioned. Moreover, by asking opinions on experiences such as getting lost, finding easily, loss of time, anxiety and on points of the system deficiencies/corrections; indirect effects of signs have been tried to be detected. An interview form approach has been adopted with the anticipation that the responses of the participants could change the course of the study.Preliminary studies have been carried out for analysis by selecting important parts such as meaningful words and sentences in the texts derived from the interview records. This kind of analysis, which Yıldırım and Şimşek (2013) characterized as induction, is called content analysis. Codes as the key words that correspond to the responses of the participants and themesas the headings that categorize and group the codes are developed. Matrix notation is used in the presentation of the data. Data entry in the form of this table representation is possible in various forms such as short text blocks, expression quotations, classifications, abbreviations, symbolic figures, marked lines and arrows (Miles and Huberman, 1994). Participants are coded with numbers such as K1, K2, K3 and placed horizontally on the line, themes/ codes forming themes are placed vertically in the column. User joining the statement marked with x in the cell where the user and the code coincide, while unmarked cells mean that they do not participate. Cells covered with different colors belong to the participant who has not made assessment because they did not notice the direction signs on the campus. Yıldırım and Şimşek (2013) indicate that the percentage calculations in the systematic digitization of interview data are important.In this direction; the total number of participants who expressed their opinion on the subject was taken and the percentages within the group were calculated and added to the last columns of the matrix. By doing so, a digitized visualization of the content analysis was created.  and what they mean in the given circumstances" (Frey, Botan and Creep, 1991).In order to reflect in a more striking way the users' experience of navigating the campus and the effects of these experiences; the data were examined through descriptive analysis, with direct cues from responses.
In the beginning of the interview, question of "Did you find the places you did not know/visited for the first time at Kuzeykent Campus easily?" was asked, 30% of the respondents answered positively and stated that they found it easily, 60% answered negatively and stated that they could not find it/had difficulty. The behaviours of the participants to find their ways/directions in the campus and their expectations related to present signage/wayfinding system are seen in the content analysis matrix and interpretations.

Visualization Procedure
It is possible to determine the codes and themes in the text prior to data collection. If codes and themes are to be developed after data collection, researchers can make use of manual methods or can perform analysis automatically by using computer programs. In addition to Atlas. ti, Maxqda and NVivo programs for professional users, there are also web-based/online visualization tools for the average user group. "Many of these visualisation tools and techniques include a welcomed aim of bringing corpus analysis to student (Allen,  It is envisaged that, among the other key terms that can be examined in this direction, the word more has significant connotations.The percentage of participants who indicated that larger signs/letters or more noticeable colors would be useful was 65%. A visualization made with this term expands with terms bigger, more striking, more colorful. Thus, statements that show the expectations of the users about the signs can be reached.

Conclusions
While there are a large number of visualization methods that can be used in the analysis of quantitative data, there are relatively few options in qualitative data display. It can be stated that these data are usually presented as quotations from the text (in plain text form) or digitized. Text visualization applications with advantages such as easy access, user centricity, intuitive, fast and flexible usage, brought an alternative dimension to the analysis process.
The qualities of the data to be analyzed and the format of display are factors for determining the appropriate type of visualization practice. Word cloud visualization, which provides an explanatory and inductive structure, is example of a commonly used type. Word tree visualization is exploratory and deductive in nature, allowing researchers to search for new root terms in the text. These techniques, which researchers can use simultaneously with content analysis, will contribute to the interpretation and display of content in easy and understandable manner.