Tools for learners expanding their English vocabulary.

You meet a word you do not know, in an article, a chapter, or the subtitles of an episode, and you want to make it yours without losing the thing you were doing.

The tools here take the mechanical work out of that moment, and they leave the judgment, which words are worth your time, where it belongs: with you.

Two projects belong to the organization, on that one problem, by one person.

Wordwinnow reads a text you are about to read, gives every word a CEFR level, and places each word against the level you have and the level you are working toward, so you can see, before you start, which words deserve your attention now, which to keep for later, and which you likely know already.

It is a command-line tool, feature-complete at v1.0.0, and source-available under the PolyForm Strict License 1.0.0, which permits using it for any noncommercial purpose, personal study and research included.

Its repository holds everything needed to run it and to understand it: how to get it running, what it is for and what it will not do, how it is built and why each technology is there, what was measured, and where the idea came from.

An error found in it may still be corrected, and an issue on its repository is the way to report one.

Wordbench is the later project on the same problem, and a separate one: neither project is a version of the other.

Wordwinnow was shaped by its technologies more than by its learner; Wordbench starts from the product question, what a vocabulary tool should actually help a learner do and which decisions must stay the learner's, and chooses its technology to serve the answer.

Wordbench is not yet public.

Generative AI did a substantial part of the engineering work on both projects, through models of two families: Claude, by Anthropic, and GPT, by OpenAI.

A proposal, whether a model's or mine, is questioned, tested against the running software, and then kept, reworked, or dropped, and every change answers to the same engineering standard.

Working this way is itself part of the engineering, and it is not finished: how the collaboration is organized changes with what each project teaches, so the software and the way of working grow together.

I am Oleg Turzhanskii, and the decisions and the results are mine to answer for: I decide what each project is for and what it leaves to the learner, set the requirements and constraints, weigh the proposals and the evidence, and settle what is accepted, revised, or rejected.

• Wordwinnow: run it on a text of your own, or read why it exists.
• My profile: the person behind both projects, and how I work.
• The engineering standard: the conventions both projects follow, with the tooling that enforces them.