Analyze and Visualize

Analyze and Visualize with CATMA

Tutorial 2

A German version of this tutorial with Franz Kafka’s Erstes Leid as sample text can be found here: Jan Horstmann (2019): Analyse und Visualisierung mit CATMA. In: forTEXT. Literatur digital erforschen. https://fortext.net/routinen/lerneinheiten/analyse-und-visualisierung-mit-catma [accessed: December 9, 2019].

Key Data

  • Object of investigation: Edgar Allan Poe’s The Tell-Tale Heart (1843)
  • Methods: analysis, visualization, semi-automatic annotation
  • Goals: quantitative analysis of text and annotation data; creation of queries and interactive visualizations; modification of visualizations; automatic annotation of selected keywords
  • Duration: approx. 90 minutes
  • Level of difficulty: easy

Components

  • Application Example
    Which text and annotations will you explore?—Here you learn how to analyze Poe’s short story The Tell-Tale Heart and visualize analysis results.
  • Preliminary Work
    What do you need to do before the analysis?—Here you get information about necessary preliminary work.
  • Functions
    Which functions can you use in CATMA’s Analyze module?—Get to know the individual components of the module and solve sample tasks.
  • Solutions to the Sample Tasks
    Have you solved the sample tasks correctly?—Here you find answers.

Application Example

This tutorial shows you how to analyze and visualize textual data with CATMA. In part, we will build on results of the tutorial → Manual Annotation: we will analyze and visualize textual data from Poe’s short story The Tell-Tale Heart (1843) in addition to the annotations created there on the narrator’s style and attitude. However, you can perform most of the functions and tasks described here without having set manual annotations before.

Preliminary Work

In order to work with this tutorial, you must have completed at least the section “Preliminary Work” from the first tutorial → Manual Annotation. In that section, you learn how to register for a CATMA account, create a project and upload the text you want to investigate. The section “Functions” from the manual annotation session is optional. If you did not complete it in advance, you can still follow this tutorial at large. To use the functions of the Analyze module in CATMA, go to https://app.catma.de and log in. On the next page (see Fig. 1), CATMA’s home screen, you should find the project created in the previous session (we called it “CATMA Tutorial” there). Enter the project by clicking on the tile.

 

Fig. 1: CATMA’s home section with projects

Functions

CATMA supports hermeneutic text investigation by integrating text analysis and annotation into a cyclical, iterative workflow. Usually, you will typically combine your work in the Annotate module with using the merits of the Analyze module. While you practice close reading when annotating, the Analyze module also supports a distant reading approach: you can investigate your texts, even without reading them.

There are two ways to reach the Analyze module from the Projects module:

  • You can open the text by double-clicking it. Once you are in the Annotate module, you click the “ANALYZE” button below the text (this way you analyze the text together with the associated annotation collections).
  • Alternatively, you can directly click on “Analyze” in the navigation area on the left. In the opening latch you have to select the texts and annotation collections you want to analyze (see Fig. 2). In this tutorial, you select both the text and your annotation collection (contents can also be selected altogether by clicking in the upper box next to “Name”).

 

Fig. 2: Selection of documents and annotation collections in CATMA’s Analyze module

What does it mean to select texts and annotation collections for analysis? The Analyze module performs quantitative operations. These can be either text-only data (such as word frequencies or distributions) or annotation data you create yourself (such as tag frequencies or distributions). CATMA also offers the possibility to create complex queries, i.e. a combination of several searches, that can also be, for example, text and annotation data in combination.

Whenever you add or delete one or more annotations in the Annotate module, you modify the annotation data so that you potentially get different results in the Analyze module. You can open a new analysis tab using the small plus sign in the upper right corner of the Analyze module in order to, for example, integrate further annotation collections into the analysis. Similarly, whenever you click on the “ANALYZE” button in the Annotate module again, a new tab will open in the Analyze module to keep the different databases separate. The individual analyses have a time mark so that you can keep track of their chronological order (see Fig. 3).

 

Fig. 3: CATMA’s Analyze module with two analysis tabs

If you look at the structure of the Analyze module, you will see that it is generally divided into two areas: “Queries” on the left and “Visualizations” on the right. A query defines which information is retrieved from the text and/or annotation data. You can view, explore and process the results of the query in various interactive visualizations. Text visualization is thus understood as an integral part of text analysis.

For the simplest tasks, CATMA offers you a few predefined queries. Click on the small arrow pointing downwards next to “Select or enter a free query” and select the upper option “Wordlist (freq>0)” (see Fig. 4). This action gives you a list of all words in the text, sorted by frequency (see Fig. 5). By the way: the technical expression “freq>0” in brackets after “Wordlist” is the query language expression for the task of showing all words with a frequency above zero.

 

Fig. 4: Selection of the Wordlist query

 

Fig. 5: The Wordlist shows all words sorted by frequency


Task 1:
How many words does Poe’s story contain? What is the most common content word (= a word with “more” semantic meaning than function words such as articles, pronouns etc.)? What else do you notice about the given “words”? And how should a query look that returns all words in the text that occur more than five times?


If you have entered another query to search for all words occurring more than five times, you will have noticed that the result of this query is displayed above the previously created wordlist. The wordlist itself does not disappear. It is generally possible to perform several queries one after the other and collect the result lists in this form. Each query result has a small search field in the header to quickly search the respective list (see Fig. 6). You can delete a result list by clicking on the small eraser symbol. The small three dots to the right contain export options for your queries in CSV, XLSX or JSON format. The arrow symbol allows you to expand and collapse the result lists in order to keep a better overview of all your queries.

 

Fig. 6: Search bar and symbols in CATMA’s query result lists

Word frequencies can also be explored visually in the form of a wordcloud. This is one of the visualization possibilities for query results currently implemented in CATMA. To create a visualization, run a query (such as freq>0) and then click the desired button on the right; in this case “Wordcloud”.

The wordcloud view opens and on the left you see the query result list again. From this list you can now select the words to be displayed in the wordcloud by using the down arrows in the “Select” column. This way, you can manually compile a list of words that shall appear in the wordcloud. Hint: the “Select all” option available via the three dots of the query result list integrates all words into the wordcloud. You will receive a wordcloud in which the size of the words represents their frequency in the text (see Fig. 7). The second list created this way allows you to delete individual values (such as punctuation marks, numbers, or words belonging to possible paratexts) using the eraser icon so that they also disappear from the visualization.

 

Fig. 7: CATMA wordcloud for word frequencies in Poe’s The Tell-Tale Heart

Under the wordcloud, you have several possibilities to manipulate your visualization via slider. You can change the number of words, the size of the wordcloud, as well as the contained words and also the distance between the single words (“Word padding”).

Side note: all visualizations in CATMA have three pale dots in the upper right corner that allow you to export your visualization as an image (PNG) or vector (SVG) file. You can also view the source code of each visualization here. CATMA visualizations are based on the Vega visualization language. With some training, Vega allows you to create any (even interactive) visualization that fits your data. You open the Vega editor, potentially create a completely new visualization based on selected query results and load the code of this new visualization back into your CATMA project. In this introductory session, however, we will skip these expert functions. It is easier to click on the small button to the right of the three dots. The visualization code appears in a column on the right: manipulations in this code (after a click on the circle arrows above) will change the displayed visualization. Why don’t you try to change the font as shown in Fig. 8 for example?

 

Fig. 8: Excerpt from the Vega code for a CATMA wordcloud with changed font

To close a visualization and return to the Analyze module, click on the button with the two arrows pointing to each other at the top right. The visualization is minimized and collected in the visualization area of the Analyze module like the query results (see Figure 9). From here, you can return to individual visualizations at a later point.

 

Fig. 9: The Analyze module with collected queries and several visualizations

Now let’s turn back to the queries. Among the proposed queries, you will find the word list, a tag list (optional incl. properties) and the option “Wildcard”. With the wildcard option you can search for word beginnings, for example. “Wildcard” stands for placeholders (these are the “%” symbols in CATMA’s query language). The suggested wildcard query “a%” will therefore show you all words that begin with a small “a”.


Task 2:
How many words in Poe’s story begin with the letter “a”, how many with “b”?


To create more queries, CATMA provides the “BUILD QUERY” feature, which uses common language questions to help you compose your query command. After each run through BUILD QUERY, the created query appears at the top of the Analyze module. Paying attention to the form of these queries allows you to quickly learn CATMA’s query language. You can also find a systematic overview of CATMA’s query language here.

In this tutorial, we will now use the BUILD QUERY feature. If you click on the corresponding button, you will first be asked what you want to search for (see Fig. 10): words or phrases, similarities, tags, collocations or frequencies. First select “by frequency” and then click on “CONTINUE”.

 

Fig. 10: The BUILD QUERY feature in CATMA: How do you want to search?

In the next window, you can specify which words should be displayed: all words that appear “exactly” x times in the text, all words that appear more or less than x times (“more than” and “less than”), “more or equal than” or “less or equal than” x times, or “between” x and y times (see Fig. 11). You set the numerical values yourself. At the bottom of the window, you can already see what the query should look like.

 

Fig. 11: The BUILD QUERY feature in CATMA: search for frequencies


Task 3:
How many words occur five or more times, but less than 16 times? How does the query look like? Which physical senses seem to be predominant in the story?


This way, you can now build a large number of queries without having to know the query language (if you want to analyze a text or a corpus with many queries, however, we recommend that you learn the rather simple language—you will quickly save a lot of time. Additionally, not all of the more complex queries can be created using BUILD QUERY). Now click your way through the other BUILD QUERY functions (except “by tag”) and try to answer the following questions.


Task 4:
How many words have a 70% similarity to “heart”? Use the “by grade of similarity” query option. Increase the similarity to 80%. Do it again with 85%. And how often does the word “mad” appear near the word “me” (within a span of ten words). Use the “collocation” function for this. What do the respective queries look like?


CATMA offers two visualization options for examining word contexts: KWIC (keyword in context) and doubletree. Both visualization types can display selected words in their respective contexts; KWIC does this in the form of a table that respectively displays the five words that precede and follow the selected word. This contextualizing exploration of selected words enables a selective and efficient insight into semantic constellations. To display a keyword in its context, first create a word list again and then click on the “KWIC” symbol on the right; you will be forwarded to the KWIC visualization view. Here you can select words from the list at the top left (like you did before) and thus compile another list of selected keywords at the bottom left. On the right side, the chosen keyword appears as a context list (see Fig. 12). Further selected keywords are added to this list; you can remove keywords by clicking on the eraser symbol in the list at the bottom left. You can enlarge or reduce the individual columns of the keyword list with the mouse (this generally applies to all panels in CATMA) or, for example, arrange them according to the text chronology by clicking on “Start Point” in the header. The numbers at “Start Point” and “End Point” are the so-called character offsets, i.e. the number the first letter of the selected word received in the text indexing process, and the number of the last letter. For a keyword with five letters, the number at “End Point” will thus always be exactly five values above the number at “Start Point”.

 

Fig. 12: The KWIC list in CATMA with selected keywords

If you find a passage in the KWIC list that seems interesting to you or for which you need more context, you can always jump directly from this list to the corresponding text passage in the Annotate module by simply double-clicking the line in the KWIC list that interests you. The keyword (token) is then underlined in two colors. A click on “Analyze” in the navigation area on the left takes you back to the list.


Task 5:
Analyze the word “I” in its contexts using KWIC visualization. What do you notice?


The doubletree visualization provides a more interactive form of context exploration. Close the KWIC list by clicking on the arrow symbol in the upper right corner and then click on the “DOUBLETREE” button. The creation of the visualization works according to the known scheme: select the desired keyword on the left, which is then visualized on the right as a doubletree (see Fig. 13). The difference is that only one keyword at a time is displayed, another one can be selected by clicking on it in the bottom left list.

 

Fig. 13: Doubletree visualization in CATMA

You see the keyword in the doubletree highlighted in red and displayed only once in the middle. The contexts are displayed as word types to the left and right of it. If a word occurs frequently in the context of the keyword, it is displayed larger than if it occurs only once in this context. If you click on one of these larger words, the further context unfolds (by the way, you can freely move the doubletree on the screen with your mouse). The selected word will turn red, too, as will those words that occur in the same context as that word on the opposite context side. This offers you a possibility to reflect upon the sentence structure. The doubletree visualization thus provides a structured overview of the way vocabulary is used in a specific text, and offers, for example, the possibility to draw conclusions about characterizations. For this visualization, the “case sensitivity” common in CATMA can be switched off, i.e. upper and lower case word variants can be combined.


Task 6:
Explore the contexts of the word “man” with the doubletree visualization. Can you say anything about the characterization of the character? Try the same with “I”.


How can the annotations created in the previous tutorial be examined in the Analyze module? (If you haven’t completed the tutorial → Manual Annotation, you can jump to the paragraph after figure 15). Close the doubletree visualization. You can generate the entire list of all your tags assigned in the annotation process using the offered query tag=”%”. The annotated phrases are displayed first. If you want to sort the results by tag category instead, select “Group by Tag Path” from the three-dot menu. In the result, you can see which Tag was assigned how often (see Fig. 14).

 

Fig. 14: Tag list in CATMA’s Analyze module

You can also display the specific annotations and properties using the options “Display Annotations as flat table” or “Display Properties as columns”. If you then hover over the individual annotations, a bit more context is displayed (see Fig. 15).

 

Fig. 15: Tag list in CATMA’s Analyze module with Annotations

You have the possibility to generate distribution graphs for words or tags—the fourth visualization option in CATMA. These distribution charts are created in the same way as the other visualizations, i.e. by transferring individual lines from the query result to a visualization-specific second list, which in turn can be manipulated to edit the visualization. In this case, we want to try out the distribution for assigned tags. (If you did not set any annotations, simply create a wordlist and select individual words for the distribution visualization.)

Build a tag list with the given query and select the option “Group by Tag Path” in the three-dot menu. Then click on the “DISTRIBUTION” button on the right. Choose “Select all” to transfer all lines of the query result into the visualization list. A grid appears on the right, showing the distributions of the tags you have assigned (see Fig. 16). The x-axis refers to the entire text length, which is divided into 10% segments here (based on the number of letters). The number of occurrences appears on the y-axis. You can enlarge the entire visualization using the zoom slider.

 

Fig. 16: Distribution visualization for assigned Tags in CATMA


Task 7:
Visualize distribution graphs for the tags you assigned. What do you find noteworthy? What purpose is this form of visualization suitable for? In which respect would another visualization be more useful?


You can also use “BUILD QUERY” to search for individual tags. There you select the option “by Tag”. After a click on “CONTINUE”, the tagset appears from which you can select the desired tag. A corresponding query would look like this: tag=”/attitude/questionable morals%”. The query tag=”/style%” looks for all annotations of the four tags assigned to the area “style” (i.e. “parallelism”, “address to the reader”, “repetition” and “exclamation”).

It is also possible to generate complex queries by clicking CONTINUE instead of FINISH at the end of the Build Query dialog. You will then be asked if you want to “add more results”, “exclude hits from previous results”, or “refine previous results”. In the last two cases, you must also determine the type of relationship between previous and further results: there are the options “exact match”, “boundary match” and “overlap match” (see Fig. 17). The “exact match” option shows results that occur exactly at the same point (e.g. two annotations). With “boundary match”, you can search for instances that contain each other (e.g. all occurrences of a tag within another tag category). Finally, the third option (“overlap match”) allows you to display overlaps (such as different annotations that can overlap each other, or overlapping words/phrases and annotations).

 

Fig. 17: Build a complex query in CATMA


Task 8:
Search for all incidents in which words beginning with “hear” overlap with the tags of questionable factual claims. What does the query look like? How many results do you get?


The last function of the Analyze module that we want to approach in this tutorial is that of semi-automatic annotation. You may have noticed by chance that the wordcloud and distribution chart visualizations you have generated in CATMA are interactive, i.e. you can click on individual areas in the visualizations and get more options. We will try this out with a simple wordcloud visualization of all words in the text.

Open the previously created wordcloud visualization again and add all words (if this is not still the case). If you have closed the respective Analyze tab in the meantime, create a new word list, click on “Wordcloud”, then on “Select all” and increase the number of displayed words (types). If you now click on individual words in the wordcloud, a KWIC list opens below the visualization (see Fig. 18). Do this with the words “I”, “me”, “my” and “myself”.

 

Fig. 18: Wordcloud with selected keywords

With KWIC lists in CATMA, annotations can generally be created semi-automatically for selected keywords. To do this, click on the selection symbol to the left of the heading “Keyword in context” and select all lines by checking the top box next to “Document”. Alternatively, you can select individual lines from the list. After having selected all lines, you find the option “Annotate selected rows” in the three-dot menu, which you then click.

A dialog window opens (see Fig. 19), which allows you to either select existing tags or to create new tagsets and tags to be used for semi-automatic annotation. Now create a tagset named “Characters” and, in this tagset, the tag “hero” (see Fig. 20). Select this new tag in the tag list and click on “CONTINUE”.

 

Fig. 19: Dialog for semi-automatic annotation in CATMA

 

Fig. 20: Creating new tagsets/tags within the dialog box for semi-automatic annotation

In a final step, you are asked in which annotation collection the new annotations should be stored. Here you select your own annotation collection (see Fig. 21) and click on “FINISH”. The new annotations have been set.

 

Fig. 21: Selection of the annotation collection for semi-automatic annotation

You have now annotated all keywords in the list with the tag “hero”. A double click in the list will bring you back to the Annotate module (see Fig. 22), where you can see the new annotations (you may have to click on the eye symbol next to the corresponding tagset on the right side, if it is crossed out).

 

Fig. 22: Poe’s The Tell-Tale Heart with semi-automatically generated annotations of the main protagonist


Task 9:
Choose words from the wordcloud or create a KWIC list from the usual wordlist and semi-automatically annotate them with an “opponent” tag. Which words or phrases qualify for being annotated as “opponent”? For what kind of annotation tasks is the semi-automatic annotation feature particularly suitable? In which cases is caution recommended?


You are now familiar with all the basic functions of the Analyze module in CATMA: You can create simple and complex queries, build and manipulate lists for visualizations, modify visualizations and generate KWIC lists for semi-automatic annotation to enrich your annotation data.


Task 10:
Build a distribution chart containing the tags “hero” and “opponent”. What can be seen from the representation of the two tags?


Solutions to the Sample Tasks

Task 1: How many words does Poe’s story contain? What is the most common content word (= a word with “more” semantic meaning than function words, such as articles, pronouns etc.)? What else do you notice about the given “words”? And how should a query look that returns all words in the text that occur more than five times?

Query results are always displayed in brackets in the gray column above the list as well as at the bottom of the list. The Tell-Tale Heart accordingly contains 2624 word occurrences (tokens) and 763 word types. But beware: the words from the paratext, which contains information about the edition, the text, its author and the editor, are counted as well. The words “Poe” or “Wikisource” therefore do not appear in the story The Tell-Tale Heart as indicated. When analyzing word frequencies, always keep an eye on how the underlying document is designed. 

The word “was” appears 33 times (or, in case you do not consider pronouns function words, “I” appears 118 times). Even from this quick glance into the wordlist and without knowing the text, it is possible to infer the main characters (homodiegetic first person narrator; no females) and the relationship between narrative time and narrated time (subsequent narration) as well as an emphasis on physical senses.

Punctuation marks are also counted as “words” in CATMA (if you deactivate “Filter punctuation” in the three-dot menu, you can also display them). The reason for this is that a word could basically be defined by the spaces before and after the word. According to this rule, punctuation marks (always displayed without spaces next to a word) would be counted together with the words (i.e. “interpret” and “interpret,” would be two distinct words). At the same time, one does not want to ignore the frequency of punctuation marks, as they can provide valuable impulses for the interpretation of a text (e.g. if it contains more question marks than exclamation marks, or more dots than commas, etc).

Finally, you may have noticed that CATMA does not count uppercase and lowercase words together (e.g. “my” occurs 28 times, “My” two times), i.e. it is case sensitive, and also counts word forms separately (e.g. “see” occurs five times, “seen” two times). So if you want to find out how often a word is used in its basic form (actually a lemma), you should add all forms of the word. The search bar in the header of the query result is suitable for this.

The query for all words that occur more than five times in the text should be: freq>5.

Task 2: How many words in Poe’s story begin with the letter “a”, how many with “b”?

167 tokens beginning with “a” and 90 with “b” are displayed. The second query is: wild = “b%”.

Task 3: How many words occur five or more times, but less than 16 times? How does the query look like? Which physical senses seem to be predominant in the story?

75 words (types) between five and 15 times appear in the text (the words of the paratexts are also counted here). The query for this is: freq = 5-15. CATMA includes the words that occur five or 15 times with this expression. Especially sound but also vision are the predominant physical senses.

Task 4: How many words have a 70% similarity to “heart”? Use the “by grade of similarity” query option. Increase the similarity to 80%. Do it again with 85%. And how often does the word “mad” appear near the word “me” (within a span of ten words). Use the “collocation” function for this. What do the respective queries look like?

Twelve words have a 70% similarity to “heart” (be aware that an instance like “heart—for” is counted as one word here), nine words with 80%. With 85%, only “heart”, “hearty”, “earth”, “hear”, “tear”, “Heart” and “HEART” are listed. The query is: simil=”heart” 70%. It is more time-saving to simply adjust the percentage in the query bar than to use the BUILD QUERY function again for each query.

The words “mad” and “me” co-occur two times within a span of ten words. The query for this is: “mad” & “me” 10.

Task 5: Analyze the word “I” in its contexts using KWIC visualization. What do you notice?

It is noticeable that the word “I” occurs very often, so it is probably a story with a homodiegetic first-person narrator. A closer look at the KWIC list confirms this impression, because the “I” never appears in the direct speech of another character.

Task 6: Explore the contexts of the word “man” with the doubletree visualization. Can you say anything about the characterization of the character? Try the same with “I”.

The doubletree allows you to explore a word in its various contexts. For example, the man is obviously old, scared and he dies. This may lead to conclusions about the character of the man as well as his feelings to and/or his behavior towards the narrator. For the word “I” the doubletree is not the most suitable visualization, because it occurs so frequently. Nevertheless, one can draw conclusions about the character of the main character: he is very active, sees, moves, could, knew etc.

Task 7: Visualize distribution graphs for the tags you assigned. What do you find noteworthy? What purpose is this form of visualization suitable for? In which respect would another visualization be more useful?

It is noticeable that, due to the rather short text length, the 10% segments are correspondingly short, which means that in each segment only a few annotations of each category occur. Furthermore, several of the annotated categories have similar distributions; the most striking is the exclamations tag, since 17 such exclamations take place in the last 10% of the text alone.

Distribution charts usually get more interesting if occurrences (e.g. between tags and words) are visualized and investigated. The connecting lines between the individual points of the distribution chart should also be considered critically, as they imply a continuous development. This is not given in texts in which separate words form the data basis. Distribution charts are particularly suitable for longer texts. For a mere representation of the frequencies of assigned tags, the wordcloud is a more fitting visualization, which in this case could look somewhat like this:

Fig. 23: Tag cloud for narrator’s style and attitude in Poe’s The Tell-Tale Heart

Task 8: Search for all incidents in which words beginning with “hear” overlap with the tags of questionable factual claims. What does the query look like? How many results do you get?

The query is: (wild=”hear%”) where (tag=”/attitude/questionable factual claims%”) overlap. The result amount depends somewhat on which decisions you made in the annotation process. In our case, we get nine matches (four times “heart” and three times “heard”).

Task 9: Choose words from the wordcloud or create a KWIC list from the usual wordlist and semi-automatically annotate them with an “opponent” tag. Which words or phrases qualify for being annotated as “opponent”? For what kind of annotation tasks is the semi-automatic annotation feature particularly suitable? In which cases is caution recommended?

Words/phrases like “old man”, “eye”, “heart”, “corpse”, “he”, “him”, “his”, “body”, “police”, “officers”, “they”, “them”, “their” etc. qualify to be annotated with the “opponent” tag.

Semi-automatic annotation offers a very time-saving support especially for the annotation of named entities (like the main character in our example). But also time forms of verbs could be annotated semi-automatically with the corresponding tags (e.g. “past”, “present”, “future” or also “simple past”, “present perfect”, “simple present”, “future I” etc.). Be careful here because there are words which can be verbs in a certain time form as well as other types of words depending on the context. The tag “hero” could also lead to difficulties in narratives with a homodiegetic narrator: here the “I” does not always have to stand for the narrator, since it can refer to every other character in direct speech. The context in the KWIC list should therefore be taken into account, especially during semi-automatic annotation.

Task 10: Build a distribution chart containing the tags “hero” and “opponent”. What can be seen from the representation of the two tags?

The distribution of the hero and opponent annotations could look somewhat like this:

 

Fig. 24: Distribution chart for “hero” and “opponent” in Poe’s The Tell-Tale Heart

By clicking on the individual dots, you can examine single passages of the story in a KWIC list. Considering the rather parallel distribution of the two categories, it is noticeable that the curves drift apart in the section between 50 and 60%. Examining the contexts of this passage, one finds out that this is exactly the point where the “hero” kills his “opponent”, the old man. The mere distribution of frequencies thus hints to this central and important passage of the narrative.


Cite this article as: Jan Horstmann: “Analyze and Visualize”. In: CATMA, published at: 9 December 2019, [last access: 31 March 2020,] https://catma.de/how-to/tutorials/analyze-and-visualize/.