For example, I started reading a new book and want to highlight unknown words in it. So this service collects my known words and filters them from the others (unknown), then highlights new words (with possible translations).

As a result, I can create my own set of vocabulary, and keep words that I have to learn (export them via CSV or print as PDF).

After learning new words I can move them from the 'unknown' list to the 'known' list, and in the next text (that I'm reading), I will see only new 'unknown' words to continue reading.

  • When you say you’re looking for something to highlight the unknown words, do you mean with yellow in what you’re reading? If so, what software are you reading in (eg Chrome)? – Laurel Dec 16 '20 at 21:37
  • Yes, any browser tools(chrome extension etc.) or site. – Sasha B. Dec 17 '20 at 8:35

Here are two different kinds of resources of which I am sure there are many other instances.

If you're looking for something for you yourself to mark words that are new to you, to remember them in a list maybe for language learning, there is the Chrome extensions:


If you add it to your browser, it allows you to click on a word to translate it between any two languages. YOu can then add this word to a flashcard list for review later.

If you are looking for new words for everybody, or at least for one large publication, there is a twitterbot called

New New York Times @NYT_first_said

which describes itself as:

Tweets words when they appear in the New York Times for the first time.

It's not perfect: the corpus is not huge (just the NYT), it includes some obvious misspellings, one-off foreignisms, , and anything goes in quotes. But it is just what is wanted for that 'first citing' for OED as a place of record for neologisms.

  • Thank you @Mitch! Readlang is wonderful service! – Sasha B. Jan 7 at 13:31

Natural language processing might make this achievable. One could filter a corpora of their messages, literature, tweets, etc. (after clean-up) using nltk. Subtracting that set from all the words in the dictionary (or a book) would yield the "unknown" (or, more accurately, "unused") words.

One would have to

  • Provide and tidy up the data
  • Pick a suitable model
  • Interpret the results

Though a tool can help, I wouldn't recommend this for primary vocabulary building because it is unnatural. It takes considerable time and patience. Moreover, one needs intermediate programming skill and a good grasp of data science and mathematics to implement this.

If you are still considering it, I would recommend watching this video.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .