DE102010027146A1 - Method for displaying degree of difficulty of readability of text in word processor, involves comparing numerical readability subscripts in overview of text, so that readability assessments of passages are compared with each other - Google Patents

Abstract

The method involves determining degree of difficulty of readability of text in a word processor by an application depending on quantifiable linguistic units e.g. letters, syllables and words, in the text in narrow passages. Numerical readability subscripts are simultaneously and separately represented for the individual passages of the analyzed text as similar closed linguistic units. The subscripts are represented in overview of the text such that readability assessments of the individual passages are compared with each other.

Description

Name of invention

• »Method for presenting the readability assessment of all individual passages of a text in a comparative overview« (Hereafter abbreviated as "Readability Assessment").

Affiliation to the technical field

The "readability assessment" belongs to the algorithmically supported improvement of texts recorded in word processing, as eg This is the case, for example, with spelling corrections and other analysis methods for eliminating weaknesses in the text.

This serves for the proofreading of texts by the author himself as by a foreign-Lektor.

In our estimation, the "readability assessment" is associated with the IPC classification "G06F Data Processing", Subgroup "17: Digital Computing or Data Processing Equipment or Procedures, Especially Adapted to Special Functions", in which "G06F17 / 21 word processing" takes precedence over »G06F17 / 28 Processing or Translating Natural Language« and in turn precedence of »G06F17 / 27 Automatic analysis, eg. Parsing, spelling correction «.

Previous state of the art

Speech analysis to determine the readability of a text has been the subject of research since at least the 1920's ( Kitson, HD, 1921. "The mind of the buyer." New York: Macmillan ).

Legibility here is the assessment of the comprehensibility of a text passage, based on the writing style used by the author.

To date, there is a wide differentiation of various formulas, all of which provide a useful assessment but only the overall readability of a text. The most important criteria here are the text-immanent features "word difficulty" and "sentence complexity", determined statistically in the form of letters or better syllables per word as well as letters, syllables or words per sentence, whereby these criteria are compared in different formulas.

Overall, there is a lack of opportunities for the German market to analyze texts for readability ( Felsch, Claudia, 2004, »Frequential Readability Analysis«, Grin Vg. For akad. texts ), however, no further approaches can be found on an international level that specifically support finding the weak points of a text.

An implementation in word processors or even direct, up-to-date software programs for text analysis such as »TextQuest« ( http://www.textquest.de ) has not yet or only inadequate. Basically, a formula is always applied to the entire text and is reflected in only one result number. A vulnerability-oriented intervention in the text is not possible.

Selection of a formula for readability analysis (belongs to »state of the art«)

There are a variety of formulas for readability analysis that provide meaningful results. Be it the original, first formula of Rudolf Flesch, 1948 (Dubay, William H. (Editor), 2007: Unlocking Language: The Classic Readability Studies, BookSurge, Llc. - reprint of the original edition out of print ), Modifications thereof for the German language as at Toni Amstad from 1978 (Toni Amstad, 1978: "How understandable are our newspapers?", University of Zurich: Dissertation ) and the like dozens more.

Which formula of the readability assessment is deposited is only incidental importance. The best-known formulas all have a high degree of usability for assessing readability and vary only in details that are the subject of current linguistic research, but generally have a weak impact and emphasize different linguistic aspects.

Many authors are already working with "their" formula, so they have opted for one of the existing formulas and will want to continue doing so.

Accordingly, in our "readability assessment" method, one can deposit any formula based on the ratios of letters and syllable numbers per word and sentence as well as words per sentence or even the evaluation of frequency lists of words of a particular language.

In principle, our innovation in the comparative display for individual text passages (such as paragraphs) applies to all imaginable formulas.

However, we had to opt for a well-usable formula as the basis for the user. Thus, users who are not fully familiar with the mathematical principles of the individual formulas can also use the procedure.

So we have invested considerable work in finding a formula that makes the most sense for the modern German language in the field of fiction and journalism right through to textbooks.

This was done through elaborate statistical analyzes and comparisons of many megabytes for the "readability assessment" of meaningful texts.

1

Dubay, William H. (Herausgeber), 2007: »Unlocking Language: The Classic Readability Studies«, BookSurge LIc. (Nachdruck des vergriffenen Originalaufsatzes) oder auch leichter findbar: » www.wikipedia.de « -> Lesbarkeitsindex

(bekannt als »Flesch«-Formel) als Basis gewählt, die sich auch für die deutsche Sprache als tauglicher erwies als andere Formeln. LI = 206,835 – (1,015·DSL) – (84,6·DSW) mit:

• • DSL: Die durchschnittliche Satzlänge ergibt sich, indem die Anzahl der Wörter im Text durch die Anzahl der Sätze des Textes dividiert wird.
• • DSW: Die durchschnittliche Silbenanzahl pro Wort ergibt sich, indem die Silbenanzahl des gesamten Textes durch die Anzahl der Wörter im Text dividiert wird.
• • LI: Der Lesbarkeitsindex als Maß für eine Einschätzung der Lesbarkeit.

For the basic implementation, the original 1948 formula was developed by Rudolf Flesch 1

1

Dubay, William H. (Editor), 2007: "Unlocking Language: The Classic Readability Studies," BookSurge LIc. (Reprint of the out-of-print original essay) or easier to find: » www.wikipedia.de «-> readability index

(known as "Flesch" formula) as the basis, which also proved to be more suitable for the German language than other formulas. LI = 206,835 - (1,015 · DSL) - (84.6 · DSW) With:

• • DSL: The average sentence length is obtained by dividing the number of words in the text by the number of sentences in the text.
• • DSW: The average number of syllables per word is calculated by dividing the syllable count of the entire text by the number of words in the text.
• • LI: the readability index as a measure of readability.

This "Flesch" formula was adapted to the German language by reducing the syllable count for this word by "1" for words ending in "e", based on the consideration that many words with a terminal " This is not perceived as an added syllable. Relevant here, however, is not sensation, but elaborate statistical analysis.

Statistical analyzes have shown that the original "Flesch" formula is perfectly suited for the readability analysis of the German language due to this additional modification.

The problem underlying the invention

For the realization that an improvement of the readability of a text is desired, one does not have to go so far as William DuBay ( DuBay, William H., 2004: Principles of Readability, (http://www.impact-information.com/impactinfo/readability02.pdf ) or in DuBay, William H., 2006: "Smart Language: Readers, Readability, and the Grading of Text, BookSurge LIc. ). He argues that there are fewer dead children if more people understand the instructions for poorly installed child seats better, which in turn would be the case if they were written more comprehensible.

Even a commercial interest is worth every innovation in itself. For example, DuBay describes in the same sources that magazines have experienced drastic sales increases of up to 60% by increasing readability. However, as stated in the "prior art" above, so far there is no way to find out why a text is classified as difficult to read. An overall text that is classified as difficult to read does not show in the overall analysis where the passages are that render a text more difficult to read.

There is also the problem that, compared to the overall text, easily readable passages compensate for poorly readable passages. The text appears easy to read from the overall index, but this is not the case in a number of "weak" passages.

Being able to assess the readability of individual text passages at a glance and in comparison with adjacent passages in order to be able to improve the readability at the specific place and thus gradually by finding the particular focus of poor readability throughout the text is a hitherto unresolved one Task.

The display of readability is done so far only in numerical form, a graphical implementation, for example, in a color display does not exist yet.

Texts must be subjected to a readability analysis in a targeted manner, according to previous procedures, until a result is obtained.

It was not possible to simply add a readability analysis while editing a text normally. The possibility of directly changing a found vulnerability with immediately updated display of the new readability assessment did not exist so far.

Problem solving and improvements of the known state of the art through the »readability assessment«

Resolution of the readability analysis on text passages (as paragraphs)

The "readability assessment" takes a new path, in that the entire text is not the goal of the readability analysis, but the text is separated into individual passages of text for analysis.

These text passages are similar, self-contained linguistic entities. The selected unit for the text passages examined is each paragraph of the text (other types of units, such as sentences, can also be set).

The innovation lies in the technical implementation of making the readability of all individual text passages visible in the entire text at the same time, so that hard-to-read passages are immediately apparent when the text is being edited by the editor and can be directly corrected.

Graphical display of readability indexes directly on the text passages

The readability analysis results can be displayed numerically by presenting the readability index (LI) next to or on the paragraphs.

For easy, direct detection by the lecturer of the readability assessment of each text passage, the readability index with a function was mapped to a color space as a variant of a graphic image that is more easily recognizable than a mere number representation.

The default setting is an illustration of the range of the readability index on the colors of the rainbow or the corresponding section of the color wheel from HSV values from red to orange, yellow and green to blue, since this is intuitively easy to grasp.

Also usable are other assignments that the processor perceives as suitable, for example, an area of the color space which images the traffic light colors from red to yellow to green. For color-blind users, an assignment to gray values can also be set.

Readability assessment directly when editing

In principle, even during the display of readability, a normal processing of the text in the word processing system is possible. Changes to the readability assessment are thus immediately visible during editing.

In the implementation of the "Readability Assessment" in the word processing system, a switch in the text entry window allows you to switch between normal view and readability assessment, if the author feels disturbed by the appearance of the colors.

Otherwise, the possibility of text editing with and without the color representation of the readability indexes is equally given.

The advantages achieved by the invention

By breaking down the analysis into individual text passages as linguistically closed text units (such as paragraphs) and their graphically implemented readability assessment, the particularly difficult to read passages of a text are directly recognizable. The fact that the text remains editable while the readability is displayed can be improved immediately after being linked back to the display.

Thus, the editor of the text can specifically intervene in the most meaningful places in order to make the text more punctual as well as overall better readable. This is difficult to achieve after the technically possible overall analysis of the readability of a text. This is not possible when statistical fluctuations in the text cancel each other out so that hard-to-read passages are not noticed and the author does not notice them by analytical means.

Also, an author can recognize whether any readability index "deteriorating" passages quite a stylistic device, that is, phrases and passages that represent so-called "strong" passages of text and tolerated by their expressiveness quite an algorithmically assessed "difficult" legibility.

In the overall text, a total negative judgment of readability would leave only a "bad feeling" without the author knowing where the negative rating came from.

All in all, the "Readability Assessment" presented here is a method by which writers and authors in the field of subject matter and specialist text can specifically write better understandable texts.

In the field of fiction, this leads to a more "fluent" reading feeling (quite important for an increase in the sales figures of a novel). In the area of technical instructions, however, a better readability instantly brings a better understanding on the part of the reader for the mediated matter. Better - understandable - explained technique is also better used.

A correspondingly high level of buying interest in an implementation of our "readability assessment" is expected.

Further developments and refinements

The data determined for individual passages of text (such as paragraphs) can be used for statistical evaluations of the text passages in relation to the entire text.

Thus, a sequential representation of the readability indices of all paragraphs can be done as a height point in a coordinate system, so that even for long texts the readability assessments of all paragraphs in the optical overview.

An illustration of the standard deviation of the readability indices across all text passages (such as paragraphs) of the text provides a measure of how uniform a text is readable. This measure shows whether there is a consistent flow of reading throughout the text, or whether a medium readability index for the entire text has been determined only by many paragraphs that are considered to be very readable, compensating for many badly readable paragraphs.

Similar findings can be seen from a histogram that shows the frequency of all paragraphs for specific areas of the associated readability index.

Readability is also influenced by algorithmically recognizable writing style weaknesses. These can be doubly used words in a narrow text area, furthermore, with the aid of corresponding word lists, typical filling and swelling words can be recognized, which are neutral or even negative for the meaning of a text.

By combining the readability index determined by the above formulas with such style weaknesses, a malus added to the readability index for each style defect found can result in a more precise assessment of readability, which takes into account other parameters besides the values determined only from the formulas.

Embodiment of the invention

The following text example shows the individual paragraphs classified differently in terms of their readability with the colors assigned to the respective readability indices (LI) (otherwise as the background color of the paragraph, in this case as an HSV value).

Substantially meaningless sample text for a »readability assessment« I think that with weighty argumentation aids, which I have simply sucked from my fingers, built in horrible construction extremely disgusting megaband worm sentences that will look even diehard multi-text readers with fright. If such overlay sentences are now also summarized in a long paragraph, then my expectation is certainly quickly fulfilled that just this gigantic paragraph should turn out to be the most ghastly, difficult readability.

Here I really formulate without any effort to write an understandable paragraph, but I also do not make an effort to purposely slash the paragraph in its readability. In fact, I am extremely curious to see if and how a well-recognizable difference in readability and comprehensibility emerges. If all goes well, then these four sentences should be well distinguishable in terms of color, becoming increasingly simpler from top to bottom.

Now I know how this has to work. And that brings us closer to the perfect result of the new papyrus readability assessment. Here then comes the final sentence, which I still have to write. It was an overwhelming ease to write this, even if it's nonsense.

It is easily recognizable by the colors that a reworking of the first paragraph shows the biggest effect in order to improve the readability of the entire text.

By targeted improvement in the first paragraph you can quickly reach a serious improvement in readability.

I think that with weighty arguments, which I have simply sucked from my fingers, built in horrible design extremely disgusting megaband worm sentences that will look even diehard multi-text readers with horror. If such overlay sentences are now also summarized in a long paragraph, then my expectation is certainly quickly fulfilled that just this gigantic paragraph should turn out to be the most ghastly, difficult readability.

For example, if you change the easily recognizable filler words shown in bold by removing or rewriting them, the legibility assessment will change - while you are editing.

I think that I can get tapeworm sentences built with arguments that I have just sucked from my fingers, even read the multi-text reader with great fright. If such long sentences are now also summarized in a long paragraph, then my expectation is certainly quickly fulfilled that just this gigantic paragraph of the most difficult legibility should be.

Further application example: This application

A first version of this application already contained all text passages and arguments.

Passages marked as particularly difficult to read have been purposefully improved by simplifying sentence constructions and wording. Thus, the readability could be greatly improved with less effort of less than one hour. The original readability of the entire text of »16.4«, normal for a complex technical text, could be raised to much more pleasant readable »30.4«.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited non-patent literature

• Kitson, HD, 1921. "The Mind of the Buyer." New York: Macmillan [0004]
• Felsch, Claudia, 2004, »Frequential Readability Analysis«, Grin Vg. For akad. Texts [0007]
• http://www.textquest.de [0008]
• Rudolf Flesch, 1948 (Dubay, William H. (Editor), 2007: "Unlocking Language: The Classic Readability Studies," BookSurge, Llc. - Reproduction of the Out of Print Original Edition [0009]
• Toni Amstad from 1978 (Toni Amstad, 1978: "How understandable are our newspapers?", University of Zurich: Dissertation [0009]
• Dubay, William H. (Editor), 2007: "Unlocking Language: The Classic Readability Studies," BookSurge LIc. (Reprint of the original edition which was out of print) [0017]
• www.wikipedia.de [0017]
• DuBay, William H., 2004: Principles of Readability, (http://www.impact-information.com/impactinfo/readability02.pdf [0020]
• DuBay, William H., 2006: "Smart Language: Readers, Readability, and the Grading of Text, BookSurge LIc. [0020]

Claims (3)

Display the level of difficulty of the readability of a text in a word processor in the form of a numeric readability index by using any formula as a function of quantifiable textual units of the text (such as letters, syllables, words, sentences, words from different lists of the analyzed speech, in narrow passages multiple words), characterized in that the representation of the readability index separated for individual passages of the analyzed text as similar, self-contained linguistic units (such as sentences or paragraphs) simultaneously and in an overview happens in the text, so that the individual Text passage is in its readability assessment in comparison with the adjacent passages. Display of the readability of a text according to claim 1, characterized in that the assessment of readability is additionally or alternatively also represented by an immediately visible graphical representation of the readability index. The graphical representation is done in the form that every readability index of a text passage is assigned a color value. The readability index is mapped into the color space over the entire area that it can assume, via a predetermined or selectable function, so that different readability assessments of adjacent text passages are noticeably different for the editor. As an alternative to the color representation, the readability index can be displayed in a graph adapted to its value, which is provided to the associated text passage, eg. B. in a value corresponding to the wide bar. Display of the readability of a text according to claim 1, characterized in that the representation of the readability index on changes of a text passage (eg the editing of an affected paragraph in a word processing system) reacts immediately and is updated automatically.