The About the word widget shows the grammatical and semantic features of the word.

For nouns, adjectives, verbs and adverbs, you can get the most complete grammatical information.

The semantic features of homonyms are shown in separate lines.

The widget shows a word frequency scale consisting of six ranges.

For the searched word the frequency (IPM) is defined as the ratio of the number of occurrences of all word forms divided by the corpus size and multiplied by one million. Depending on the resulting IPM value, the word will fall into one of the following ranges:

>10000 high frequency, the word is very common
1000..10000 quite high frequency, the word is common
100...1000 rather high frequency
10...100 rather low frequency
1...10 rather low frequency, the word is rare
<1 low frequency, the word is very rare

By hovering the mouse over the widget, you can see the numerical value of the IPM.

The frequency information is available if there are examples of word usage and in disambiguated corpora only.

The widget shows the morphemic structure of a word: prefixes, roots, suffixes and endings are highlighted using the geometrical signs adopted in the school Russian language teaching.  

The word structure annotation in most corpora is based on the morphemic dictionary specially developed for the corpus, which provides analyses for more than 75,500 lemmas (over 312,000 non-unique morphemes) as of December 2024. Automatical annotation is added for the lemmas that are absent in the morphemic dictionary, including some fairly frequent lexical items. For example, the word эстетика is not included into the morphemic dictionary, so its structure (эстет-ик-а) is predicted by the algorithm. Such analyses are tagged by a special attribute "generated by NeuroRNC".

A special version of the widget is implemented in the Educational corpus. Here the morphemic structure of the word is determined in accordance with the practice of morphemic analysis at school and is based on the "Morpheme-Orthographic Dictionary" by A.N. Tikhonov (2002), containing about 100 thousand lexemes. For words that are not included in the dictionary, the morphemic structure is not determined.

For more information about the derivational / morphemic structure in the RNC, see the article.

Errors in automatic annotation may always occur. Please note that the morphemic structuring of words may differ from what you are accustomed to (see "Annotation principles"). 

The widget shows the same root words of a query word.

The root is displayed on the left, the query word and up to 10 most common words with the same root are displayed on the right.

The roots are highlighted thanks to the word structure annotation, made with the help of a morphemic dictionary specially developed for the corpus with automatical annotation added. This morphemic structuring of words may differ from what you are accustomed to.

In the current version of the widget, words with the same root are shown

• for words annotated manually, with the help of the morphemic dictionary. However, the "nests" of same root words contain not only words from the dictionary, but are also extended with the help of NeuroRNC. For example, for the word актер, all the words of the same root, except актриса and киноактриса, are selected by NeuroRNC.
• for words that are not in the dictionary, if the NeuroRNC has found at least 5 words with the same root. For example, эстет
• for words with multiple roots, same root words for one of the roots are shown
• only in the Word at a glace in the Main corpus.

When hovering the mouse over a word you can see the IPM (the number of occurrences of the word form divided by the volume of the corpus and multiplied by a million). At the same time, if a  word in a word family belongs to different parts of speech, the total IPM will be shown.

Click on any word in the widget to see the Word at a glance section.

Errors in automatic annotation may always occur. If you find any discrepancy with "Annotation principles", please report errors using the "Rate" button. 

The widget shows word forms that occur in the corpus more than 5 times. Different spellings can be displayed for the same word form if they occur in the corpus.

The cell color depends on the frequency of the word form: the more occurences of it found in the corpus, the richer the color. Hovering the mouse over each word form you can see its IPM (the number of occurrences of the word form divided by the size of the corpus and multiplied by one million). If the word form does not occur in the corpus, a dash is displayed in the cell.

Click on the word form to view its usage examples in the corpus.

Information about the word forms is currently available only for nouns in the Main, Educational and Old East Slavic corpora. Errors are possible in the annotation of lemmas and forms.

The Similar words widget displays the closest semantic associates of the word. The proximity coefficient of words, which can be seen by hovering the mouse over a word in the Word cloud, is calculated using distributive semantics models based on the actual materials of the main corpus of the RNC. The closer the coefficient value is to 1, the larger the word in the Word cloud is, and the more similar the contexts with this word should be to the contexts with the keyword.

The current version of Similar words works only in the Main, Media, Educational, and some other corpora and only shows semantic associates of the same part of speech for nouns, verbs, adjectives and adverbs. For proper names, toponyms, abbreviations and words that have non-standard spellings or are rarely found in the corpus, similar words are not displayed.

We use word2vec models trained on texts of a specific corpus to search for word associates in RNC. Currently, models have been trained for seven corpora: Main, Media, Educational, Middle Russian, “Russian Classics” and “From 2 to 15”. The Continuous Bag-of-Words algorithm (implementation from the gensim library) was used for training. All models use a vector dimension of 300 and a window of 5 words. The threshold depends on the case and is:

• 5 entries for the Main, Middle Russian, “Russian Classics”, “From 2 to 15” corpora and the corpus of Central Media;
• 7 entries for the Regional Media corpus;
• 10 entries for the Educational corpus.

You can download the vector models trained on the RNC corpora data from the “The RNC neural network models” page.

With "Similar Words" widget at the Main corpus users can now examine context-based associated words not just across the entire corpus but within specific historical periods. All texts within the Main Corpus (1700–2020s) have been divided into 11 time spans. If the number of occurrences of a word in the texts of the selected period is lower than the threshold for the corpus, similar words are not displayed.

The "Similar Words" service in "Russian classics β" corpus is available not only for the entire corpus, but also individually for works by nine authors whose corpora are the most extensive. This feature allows users to compare how a word is used across the distinctive individual styles of different writers. If the number of occurrences of a word in the texts of the author is less than 7, similar words are not displayed.

Users can view similar words from a single time span or one author, compare word clouds across two different time spans or two authors, and download a screenshot of the results.

The widget is marked with a special sign "Generated by NeuroRNC". It means, the selection of associates is completely automatical and errors may occur in the lists, for example, incorrectly formed words or word associations that are not intuitively clear.

Text distribution visualization is available in the form of bar charts and pie charts.

The bar chart shows the ratio of the number of occurrences of a word in a category, divided by the volume of that category, multiplied by one million (ipm).

The pie chart shows in which types of corpus texts the query word occurs. If you choose the "word" unit of measurement, the pie chart displays the exact number of times the word appears in the category. If you choose the "text" unit, the chart shows how many texts in the selected category contain the desired word.

You can select a meta-attribute for which to build a chart from the list of the most representative attributes of the corpus. The chart shows the distribution of the top 10 meta-attribute values. The remaining values are merged into the Other category. To the right of the chart is the list of values and its percentages. When you hover the mouse over a sector of the chart, you can see the name of the value and the corresponding number of words or texts that include the query lemma.

Text distribution visualization is not yet available in all corpora.

In several corpora, a word usage frequency by date graph (ipm, frequency per million word forms) is available. In the Regional Media corpus graphs are now plotted using days, months or years as units of measurement, in other corpora - only years.

You can use a ready-made graph that includes all examples of the word usage, or refine the displayed results by changing the time period or smoothing.

For more information about graph analysis, see the article Chronological distribution of search results.

The widget contains five usage examples of the word in the corpus. To select the examples, the lexical and grammatical search by lemma and part of speech is used. Display settings:

• random sorting
• no more than one example from each document
• user's subcorpus is not taken into account

By clicking Show all examples, you can switch to the full search results (Concordance mode).

The word portrait is built based on the full corpus without taking into account the user's subcorpus. However, if you had specified the subcorpus earlier, by clicking the Show all examples button, it will be included. In this case, the word usage examples given in the portrait may not match the first examples in the search results.

Word at a glance feature (word portraits) is available in all corpora in the new interface. In some corpora, some of the widgets are not available.