Basic Setting Options. OKAPI! has a number of features that you can set to change the appearance or content of the CBA probes or readability analyses:

Title. Type a

Figure 1: Title & Author Text Boxes

title for the CBA Probes or readability analysis into the Title text box (Figure 1). NOTE: This step is optional.

Author. Designate an author for the text sample in the CBA Probes or readability analysis by typing the author's name into the Author text box (Figure 1). NOTE: This step is optional.

Font Type. Use the Font drop-down list (Figure 2) to select the font-type. You can select one of three font-types for your CBA probes: Serif, Sans Serif, and Courier. (NOTE: Fonts determine the style and appearance of the letters that make up the text). Serif resembles well-known fonts such as Times New Roman or Palatino.

Figure 2: Options for Font Type & Text Size

Sans Serif resembles such fonts as Arial or Helvetica. Courier is a font that resembles the letter-forms produced by old-fashioned typewriters.

Text Size. Use the Text Size drop-down list (Figure 2) to select letter-size for the CBA probe. The application gives you several choices when setting the size of the text. As a general rule, more advanced student readers are able to handle text of smaller size while younger or less-skilled readers read better when given text of larger size.

Readability Formula. The Readability Formula

Figure 3: Options for Readability Formula

drop-down list (Figure 3) allows you to choose from two widely-used formulas to calculate the reading difficulty of a text sample. The Spache Readability Formula is typically used to calculate the difficulty of text that falls at the third grade level or below. The Dale-Chall Readability Formula is most often used to calculate the difficulty of more advanced text (i.e., fourth grade and up). See the section below, About Readability Formulas, for a more detailed discussion of readability calculations.

Text to Be Analyzed. Type or paste your text sample into the Text to Be Analyzed box (Figure 4). Although you can place text of any word-length into the Text to Be Analyzed box, OKAPI! will format and return samples of no more than 180 words. (If you submit samples longer than this, the application automatically trims their length before processing and returning them.)

Figure 4: Text to Be Analyzed Box

Take care that your text samples do not contain misspellings and use correct end-of-sentence punctuation.

Keep in mind, too, that OKAPI! is generally quite forgiving of strange text formatting. You can, for example, copy text directly from a web page and paste it into the Text to Be Analyzed box. Even though the text will probably seem erratically formatted in the text box, OKAPI! will automatically correct the formatting before displaying the finished CBA Probes or Readability Analysis.

'Magic' Characters. OKAPI! has several reserved ("magic") characters that you can use to fine-tune the final appearance your CBA Probes or to increase the accuracy of the selected reading formula. These are described below:

• Asterisk ("*"): Tag Words to Be Ignored in Readability Analysis: Put an asterisk just in front of any word that you want to be counted automatically as an 'easy word' in the Spache or Dale-Chall Readability Formula. (Do not leave a space between the asterisk and the target word.) You need to tag a word only once and OKAPI! will 'remember' that 'easy word' no matter how many additional times it should appear in the passage.

  For example, you would tag the name Eddie as *Eddie to signal to OKAPI! that this word is to be automatically counted as an 'easy word'. (Note: For the Spache Formula, you should mark any person's first name with an asterisk to exempt it from analysis. For the Dale-Chall Formula, you should mark with an asterisk any names of individual people (to include all first, middle, or last names that appear), as well as the names of places. )

• Backslash ("\"): Force Additional Sentence Breaks: Place this character just before any point in the text where you want to force OKAPI! to break up a longer sentence and count it as 2 shorter sentences. (Do not leave a space between the symbol and the target text.) OKAPI! will interpret this symbol as an additional sentence break, helping to reduce the average number of words per sentence in the final readability calculation.
• Right Bracket (">"): Force Paragraph Breaks: By default, OKAPI! formats all passages as single paragraphs, regardless of the formatting of the text that you originally typed or pasted in to the Text to Be Analyzed box. If you want to break your text sample into multiple paragraphs, place a right-bracket (">") just in front of any text that you want to appear indented on a new line. (Do not leave a space between the symbol and the target text.) You can insert as many forced-paragraph breaks into the text as you like.
• Underscore ("_"): Ignore End-Of-Sentence Punctuation in a Word: Sometimes your text sample may contain a word that includes periods (e.g., Y.W.C.A.). Or your text sample may include a quoted clause (with its own end punctuation) that is embedded within a larger sentence (e.g., "I am running out of time!", he said.). OKAPI! permits you to place an underscore before any word to indicate that the program should ignore any end-of-sentence punctuation that might appear in that word. (Do not leave a space between the symbol and the target word.) If you wished to tag the word time in the example above so that OKAPI! ignores the exclamation point that appears with it, you would place an underscore character directly in front of the targe word: e.g., "I am running out of _time!", he said.

Tips to Improve the Accuracy of Readability Calculations. OKAPI! can remove a great deal of drudgery from the task of computing a readability index. The application is highly accurate in computing average sentence length and percent of 'difficult' words not found on a standard wod list--values that play a part in both the Spache and Dale-Chall formulas.

A passage's 'true' readability estimate is not likely to be higher than that returned by OKAPI! However, OKAPI! may sometimes slightly overestimate a passage's level of reading difficulty. This overestimation occurs because OKAPI! 's internal dictionary is not extensive enough to contain the names of all people or places that might qualify as 'easy' words under the Spache or Dale-Chall formulas. (The OKAPI! dictionary does include a list of the 200 most common male and female first names, though, based on data collected during the 1990 U.S. Census.) Use the guidelines below to 'fine-tune' readability results for your passage.

• Spache. According to Spache guidelines (Spache, 1953), all first names are considered familiar words--but OKAPI! does not have the capability to identify all first names. Therefore, when using the Spache formula, you should place an asterisk (*) in front of any person's first name that appears in the text. (Do not leave a space between the asterisk and the target word.) The asterisk tells OKAPI! that the word should be treated as 'familiar'.

• Dale-Chall. According to Dale-Chall guidelines (Dale & Chall, 1948), all names of people and places are considered familiar words--but OKAPI! does not have the capability to identify all names or places. Therefore, when using the Dale-Chall formula, you should place an asterisk (*) in front of the name of any person or place that appears in the text. (Do not leave a space between the asterisk and the target word.) The asterisk tells OKAPI! that the word should be treated as 'familiar'.

Estimating the Readability Level of a Text. You can use OKAPI! to estimate the average reading level of larger sections of text, including chapters and even entire books (Dale & Chall, 1948). Follow these steps:

1. Select at least 3-5 text samples from the document to analyze. Be sure to take these samples from different sections of the document (e.g., near the beginning, in the middle, near the end). Each sample should be 100-150 words in length.
2. Using OKAPI!, run Spache or Dale-Chall readability statistics for each of the sample passages.
3. Once you have collected readability values for all of the text samples, add those values together and divide by the number of samples analyzed to get an estimate of the mean or average readability level for the larger document. (NOTE: With the Dale-Chall, use raw scores when computing the mean. Then find the grade equivalent for that mean raw score value by consulting the Raw Scores to Grade Conversion Table.)

Using Public-Domain Texts As Source Material for CBA Probes. There are a number of web sites on the Internet that contain

Table 1: Some Recommended Public-Domain Text Collections: Project Gutenberg Bartleby.com Blackmask.com

collections of stories, fables, legends, and other literature of high quality whose copyright protection has expired. Because these texts are now in the public domain, there are no restrictions on how you might use them. Using OKAPI!, you can quickly and conveniently turn sections of these public-domain texts into CBA reading probes. Here's how:

1. Open a web browser window and go to the OKAPI! web site.
2. Open a second browser window.
3. While in that second browser window, go to one of the public-domain text collections listed in Table 1.
4. Browse that site to find a text that you would like to convert into a set of CBA reading probes. Using your cursor, highlight about 100 to 150 words from that passage and copy it to your computer's clipboard (memory).
5. Return to the web browser window displaying the OKAPI! page. Paste the selected public-domain text from the clipboard into the Text to Be Analyzed box. Add title and author information for the passage (optional). Click the appropriate button to create CBA probes. Print off these CBA probes for later use.
6. Repeat steps 3-5 to create as many CBA probes as you need.

When I compare readability analyses run with the Spache and Dale-Chall methods on the same text sample, the results are significantly discrepant from each other. Why? The Spache and Dale-Chal readability formulas differ in significant ways, which explains why each returns different results when used to analyze a text sample. Here are just two noteworthy differences between the Spache and Dale-Chall: (1) While both formulas use a word list to determine whether a specific word in a text sample is easy or difficult, the Spache formula's wordlist contains fewer than 1000 words--while the Dale-Chall formula employs a list of 3000 words. As a result, the Spache will identify many more words in a sample as 'difficult' than will the Dale-Chall. (2) The Spache formula requires that a word be counted as 'difficult' only once, no matter how often it appears in the text sample. In contrast, the Dale-Chall counts a word as 'difficult' each time that it appears in a text sample. Differences in the size of the wordlists used and in the convention followed for counting up the total number of difficult words go a long way to explain why the two formulas return discrepant results when used on a single text sample.

Although these formulas can appear to return conflicting results, in fact the Spache and Dale-Chall were designed to complement each other (Spache, 1953). The Spache was constructed to be a reasonably accurate indicator of reading difficulty for texts up through the end of third grade. The Dale-Chall was explicitly designed to estimate readability for texts beyond the third grade level. (Read through About Readability Formulas for a more complete discussion of how each of the readability formulas works and for formula-selection guidelines.)

OKAPI! takes a long time to display CBA pages or readability analyses. OKAPI! requires a lot of computer processing power to format text samples into CBA probes and to run readability analyses. It is not unusual for OKAPI! to take up to 90 seconds from the the time that you click the Create CBA Probes button to have those finished probes downloaded and displayed. When Internet traffic to the OKAPI! web site is heavy, you may find that it takes even longer for the program to process text samples. One solution is to visit OKAPI! during times when traffic is less heavy (e.g., early mornings or late evenings).

OKAPI! overlooks common names of people or places when computing readability levels. Because it contains an internal dictionary of only limited length, OKAPI! cannot identify most common names and places, even though such words should be counted as 'easy' under the Spache or Dale-Chall guidelines. (The OKAPI! dictionary does include a list of the 200 most common male and female first names, though, based on data collected during the 1990 U.S. Census.) You can 'fine-tune' readabiilty results by hand, though. Consult Magic Characters and Tips to Improve the Accuracy of Readability Calculations to find out how you can improve the accuracy of readability scores.

When I try to submit a text sample, OKAPI! returns an 'error' message. OKAPI! is a flexible program and usually is able successfully to read and analyze most text that you submit. However, once in a great while, a user will submit a text sample with 'strange' individual characters or combinations of characters that OKAPI! cannot understand. When the application cannot figure out how to analyze a non-standard text sample, it will respond with an error message. If this happens to you:

1. take a close look at the text sample that you are trying to submit;
2. edit or delete any non-standard text characters or combinations of text characters that you believe might be causing the error(s), and
3. resubmit the text sample to OKAPI!.

I am having other problems in using OKAPI! If you are having significant problems of any kind with OKAPI!, please email me, Jim Wright, at jwright@lefthandlogic.com to share details about the problem. I will get back to you as soon as I can with an update about possible fixes or workarounds to the problem!