Log in to Account

Lost your password?
No account? Register free X


Already have an account? Log in here

Your personal information will always be kept private

Personal Information

Language Information

Account Details

I accept the terms and conditions

Get a real browser... Some functions are limited with your browser.
...about anything you want
...as your practice text is corrected and analyzed
...from your corrections on your profile

Writing Fluency Explained

Everything you write is corrected by a native and compared to hundreds of native texts with the Martian analysis technology. Your total overall score is made up of five different categories. Select the chart to learn more about each statistic.

how will you score?

The percentage of your text that was corrected from the native speaker is the main statistic that goes into your final score. After all, who can judge language better than a human? You will also notice that several other scores depend on what words were corrected from your text.

The Correction Score is simply the percentage of how closely your original text resembles the corrected text. If you wrote an entry of 10 characters and 2 of these characters were changed, your score would be an 80%.

This score makes up 32% of your Overall Score.
The Vocabulary Size Score is found by first calculating your Lexical Diversity. Lexical Diversity is the ratio of unique words used divided by the total number of words in your text. Imagine you wrote something that is 500 words long and used 500 different words to do this. This would mean you have a 100% lexical diversity. Now let's pretend that, in the same text, you only used 100 different words. This means your lexical diversity would be 20%.

Imagine both cases. Writing with 100% lexical diversity may sound great, but you probably wouldn't sound like a native writer with 100% word diversity (even native speakers repeat words). The same goes for the 20% score. A native writer would probably be somewhere in the middle. Your Vocabulary Size Score is found by comparing your Lexical Diversity to the normal range of Lexical Diversities expected from a native. If your lexical diversity is 80% and a native writer's is usually between 60% and 70%, your score will be the percent error between 80% and 70% (86% in this example).

Before we even start calculating lexical diversity, however, we have to remove anything that isn't a word. For example, punctuation can skew the results. We remove all punctuation from your text and expand all word combinations so we are strictly looking at your word use. Those Martian language analysts knew what they were doing when they designed this process.

It wouldn't be fair for us to count words toward your Vocabulary Size Score that are incorrect. If we did, you could write nonsense and it would increase your lexical diversity, artificially. Remember the percent corrected metric? We use a list of all of the words left over from the text you wrote that were correct and use only those words while calculating your Vocabulary Size Score.

The above description of lexical diversity is the simple method called Type Token Ratio (TTR). However, this varies greatly depending on the length of text being analyzed. This is why we use something different called MTLD (Measure of Textual Lexical Diversity). This is similar to TTR but is able to handle variance in text length much better. This is important since we are comparing your short texts to an entire database of native texts of varying lengths. However, the longer the text, the better the results. This is why we do require that your entries be 50 words or more. For even more information on MTLD, check out the reference below.

MTLD handles text length well while calculating lexical diversity but is sensitive to the order of your words. For example, if you have a great lexical diversity at the beginning of your text, and a poor lexical diversity toward the middle, your results may be skewed. Therefore, we randomize all of your correct words before we get your score. We do this hundreds of times and average the result to make sure that we get the most accurate score possible.

This score makes up 30% of your Overall Score.


McCarthy PM. Doctoral dissertation. 2005. An assessment of the range and usefulness of lexical diversity measures and the potential of the measure of textual, lexical diversity. Available from Proquest Dissertations and Theses. (UMI No. 3199485)

McCarthy PM, Jarvis S. Voc-D: A theoretical and empirical evaluation. Language Testing. 2007;24(4):459–488. doi: 10.1177/0265532207080767.
The average word length of your text actually says a lot about your writing style, especially when compared to that of a native speaker. Average word length is simply the total number of characters of words (punctuation not included) divided by the total number of words. Of course, only the words that were written correctly are counted here. Don't try and boost your score by using large words that you don't understand... it won't work.

As before, this value should not be above or below the average native range to sound native. If a native speaker of your target language writes an average word length between 4.0 and 4.3 letters per word and your text has an average word length of 7, you probably don't sound like a native.

In the example mentioned, you would recieve a score of 37% (since the percent error is 63%). On the other end, if your average word length was 3.0, your score would be 75% (percent error between 4.0 and 3.0).

This score makes up 17% of your Overall Score.
The sentence length score is calculated similar to the word length score. Why is sentence length important? We have seen that users who are able to match the average sentence length of native speakers sound more like a native in their writing.

Average sentence length is found by summing the total number of words of your text and dividing that by the total number of sentences. Unlike the word count score, even your incorrect text is used to calculate the sentence length score. At this point, the program is less concerned with the correctness of your writing, and more concerned with the flow. Therefore, it's important that all of your words (correct or not) are accounted for.

As in the other scores, your score will go down if your sentences are too long just like when your sentences are too short, when compared to a native.

This score makes up 13% of your Overall Score.
Word frequency comparisons are important to consider when comparing writing styles. Word frequency is the percentage of time you use each word in your text. Our Martian friends built this program to calculate the percentage every one of your words in your text was used and compares the top ten words used by natives, to your top words.

Before the program starts looking at the word frequency, it breaks down all words into their expanded form. This way, even words that make up the combined words are included in the individual word counts. Each word is counted and divided by the total number of words used. This would take a lot of work if we had to count by hand... us penguins have no fingers.

Words at the top of the native frequency list are most important. The further down the list we go, the more we see words that are more specialized. This is why the program has chosen to only consider the top ten words, and to weight the words at the top of the list higher than the ones at the bottom.

This score makes up 8% of your Overall Score.
Penguin Member


No card required

  • ✔ Corrections from natives
  • ✔ Accurate writing scores
  • ✔ Access bilingual library
  • ✔ Smarter feedback learning
  • I'm excited to have a way to track my improvements.  It keeps me motivated!
  • Finally there's a way to actually see my language progress over time! Why is this just now becoming a thing?