🤽🏼 👃🏻 🤙🏽 Experiment: how to learn to create popular texts in English (and why the English-speaking Habrists read so little) 📬 👩🏼‍🎓 😳

For many years I have been engaged in marketing, writing texts, while I am fond of English and use it in my work. There are not many articles on content marketing and business promotion in the USA on the Russian-speaking Internet as a whole. And those that are often just tell how to do on the analysis of any specific errors.

I thought it’s worth comparing the approach to creating content that native speakers use and we non-inactive when we write in English. And here are the results of my experiment.

Note : the mini-research methodology is not particularly scientific, but, in my opinion, practical. To me, the results seem interesting and useful, but do not consider them to be the ultimate truth.

Introduction: what we will analyze

To begin with, I needed to find some way to collect in one place English-language articles of originally Russian-speaking authors who “cooked” in our network segment, but would like to bring their content to an international audience. Such a collection of Russian-English content I found on a hub

Habr is a popular resource that in the last year or two has declared the goal of transformation into an international resource. Here appeared the English version, the possibility of conducting discussions in English. The problem is that so far it does not seem that the content that is written in English by Russian-speaking users is of interest to foreigners.

The success of articles by the same authors in Russian and English is significantly different. To understand why this is so, I decided to start by evaluating what kind of content they create.

Methodology

I do not consider myself an expert of a high enough level that I can evaluate it on my own, and there are no familiar media editors who would help with this for free. Therefore, I will use three text evaluation and error correction tools. These are Grammarly , Linguix and ProWritingAid . I actively use the first two (I pay for two subscriptions - Grammarly has a great web editor, and Linguix has a faster browser extension - they are discussing it at ProductHunt now ), ProWritingAid met in a couple of reviews and decided to take it as a “third opinion”.

All these tools belong to the category of Writing assistance software, that is, they find errors in English texts, offer ways to correct them, and help evaluate the “readability” of the resulting text.

The hypothesis that we want to confirm or refute - English-language texts of Russian-language authors are poorly read because:

their level of knowledge in English is not good enough;
they write too hard, while carriers prefer lightness;
texts as a whole are not in the format that a foreign audience likes.

We will look at:

length of articles
the number of errors per word,
readability indicator.

For clarity, the best-rated articles from the English-speaking Habr then we compare with the top articles that got to the main page of the Hacker News website - the top American site on the topic of startups and technologies.

Well, that’s it, let's go!

Level of English: Runet authors vs foreigners

For the analysis I took 10 articles from Habr. These included materials from the list of the best for the year, for the month, corporate blogs, and fresh posts of private authors. The total volume of the text is 22437 words.

Grammarly found in this volume of text only 986 errors, that is, by mistake for about every 23 words .
Linguix showed a similar result - 933 errors, that is, by mistake for every 24 words .
ProWritingAid – , , , 88 100 , 50. , , , Grammarly Linguix 160 , PWA 350. 1343 . 16 . , .

Errors that intellectual assistants find in English-language articles of Russian-language authors.

But we will compare these indicators with what software has demonstrated when evaluating the content of foreign authors.

The volume of English-language publications reached 22,631 words. For this volume, Grammarly detected 617 errors ( 1 error for 36 words ), Linguix found 421 errors ( error for 53 words ), PWA found 542 errors ( error for 41 words ) - but here again there were anomalies.

In general, the result is eloquent - runet authors who write in English for every 20-25 words have an error, while for foreign authors, grammar checking software generates recommendations (that is, these are not even necessarily gross errors) once for 30-40 words .

"Readability" of texts

It is clear that native speakers write on it with fewer errors. But if the article is interesting, it should be read well even with errors, but this is not observed with posts from our selection. This means that it affects not only the number of errors per word, but something else.

In English, there is the concept of readability - that is, the complexity of the text to perceive. Often these levels are associated with the level of education - for example, "the text is so complex that only people with a university education will understand it" or vice versa, "the text is so simple that it will be easily understood by a person with 8 classes of education."

A good indicator of readability is considered a level of 60-70 points- in this case, the text will be able to easily understand about 80% of native English speakers. Grammarly and Linguix in their editors provide Readability data along with other statistics for free, while ProWritingAid analyzes the readability of only the first 500 words for free. Therefore, in this segment we will have data from two services, where I already had premium access.

For materials in English from Russian-speaking authors according to Grammarly, the average readability was 55.5 . The average score for Linguix is 57.3. In general, both services give almost the same readability ratings, the difference is not more than a couple of points. As you can see, the materials of our authors (and those who maintain corporate blogs) do not reach the desired 60 points of readability.

What is the situation with the content of foreign authors? Grammarly gave an average readability score for such articles at 77.8 . Linguix - 75.4. Again, the gap between the two services is small. Much more is it between our and non-our authors.

Another conclusion is that it’s more difficult for us to write accessible texts. This is a common "disease" of non-carriers - I want to "wind up" the text more seriously, as a result, it is difficult to read.

Text Length

In this experiment, we review blog posts from private and corporate authors. Within this category, various publication formats are possible.

Domestic authors often publish rather long materials - the volume of 22,437 words was contained in ten articles. That is, the average volume is 2243.7 .

To collect a similar volume (22,631 words) of material for the analysis of English-language content, 13 articles were required. The average volume of materials is about 1740 words .

That is, foreign authors write shorter, and their audience gets used to reading texts of this particular format.

What does all this mean

According to the results of my, albeit quite superficial and not particularly scientific, but practical analysis, we can say the following:

If you want to be read by foreigners, write as a foreigner!

In practice, this means exactly three things:

Read your articles . Hire the corrector, and if it is not, use the automatic verification services (or combos from such services).
Try to make the texts as simple as possible . Read more English-language blog posts to learn style. After some time, it will be easier for you to find unnecessarily complex structures in your texts.
Write shorter . Consumers of English-language content prefer not particularly long articles - there are a number of studies on this topic. So for a greater effect, it is worth abandoning longreads for any reason.

That's it for today, thanks for watching! I will be glad to answer questions in the comments.