Asking the model to read this sentence.
A sentence nobody has read before takes a minute or two, because it means asking a language model. Nothing is lost if you leave this page: the address keeps the result.
Underline shows
What comes back
Every word with the dictionary form it comes from and the endings it carries here: case, number, gender, tense, and the job it does in the clause. Then the grammar rules the sentence uses, each one linked to the words that show it.
A sentence nobody has analysed takes a minute or two, because it means asking a language model. One that has been analysed comes back at once, and always will.
Read today's news
German published today, taken apart sentence by sentence. Every line links back to the article it came from.
vergangen
past, last
adjectiveB1-, -
Fixed in the phrase 'im vergangenen Jahr' = last year.
'die vergangenen Wahl' style: usually 'die vergangene Wahl'.
stem unchanged
The full form table has not been written for this word yet.
Seen in(7)
-
Dort sind allerdings nur private E-Roller erlaubt, die in Deutschland weit verbreiteten Leih-Scooter sind seit 2023 verboten. Auch in Deutschland sind Unfälle mit E-Rollern ein wachsendes Problem: Die Polizei registrierte bundesweit im vergangenen Jahr laut Statistischem Bundesamt fast 16.500 E-Roller-Unfälle - im Vergleich zum Vorjahr ein Plus von rund 38 Prozent. Doch eine Helmpflicht gibt es hierzulande nicht.
as vergangenen B2
-
Man macht sich starke Hoffnung, dass das in Sachsen-Anhalt gelingen könnte, und die Zahlen sprechen dafür, dass das nicht ganz unrealistisch ist." Auf der Habenseite in Sachsen-Anhalt: Mit mehr als elf Prozent das beste Landesergebnis bei der vergangenen Bundestagswahl - und damit drittstärkste Kraft hinter AfD und CDU. Heute allerdings: In Wahlumfragen mal knapp drin, mal unter vier Prozent.
as vergangenen C1
-
Erst vor zweieinhalb Jahren gegründet, folgte einem Höhenflug in ostdeutschen Landtagen das Scheitern an der Fünf-Prozent-Hürde bei der vergangenen Bundestagswahl.
as vergangenen C1
-
Rund 11.700 Verdachtsmeldungen auf medizinische Behandlungsfehler haben Experten im vergangenen Jahr im Auftrag der Krankenkassen überprüft.
as vergangenen C1
-
In Deutschland wurden im vergangenen Jahr mehr als 3.000 medizinische Behandlungsfehler bestätigt.
as vergangenen B2
My words
What you saved, and when each card comes back.
The lexicon
Every word this backend has met, commonest first. The count is what it has read, not an estimate.
Grammar rules
The sixty-nine rules the analysis chooses from, A1 to B2. Open one to read it with the sentences that use it. A rule the analysis wanted and could not find waits here as proposed until a person approves it.
How it works
Four steps, and the two places the sentences come from.
-
1
You paste a sentence
Anything up to 500 characters. Or pick one of today's news sentences instead.
-
2
A model reads it
It splits the sentence into words and records the dictionary form and endings of each one.
-
3
Rules get linked
It matches the sentence against the catalogue of sixty-nine rules and records the words that show each one.
-
4
It is stored for good
The next person to ask that sentence gets the stored answer, without going back to the model.
What you are looking at
Every word, and why it is that shape
Pick any word and its reading opens beside the sentence: the dictionary form it came from, its case, its person, and the job it does in the clause.
The underline colour repeats the word class, and the key under the sentence names every colour in words, so the colour is never the only thing carrying the meaning.
Then the rules that put it there
Under the words come the grammar rules the sentence uses. Each one names the words in this sentence that prove it, and says why it applies here rather than in general.
Open a rule to read it on its own page, with more sentences from the lexicon that use the same thing.
Questions
Where do the sentences come from
Two places. You type them, or the backend reads them. Every five minutes it fetches the German news from tagesschau.de, takes the first several sentences of each new article, and queues them. That is where most of the lexicon comes from: current German written for adults, rather than textbook sentences about a pencil on a table.
Every sentence taken from an article carries its headline, its publisher, a link back to the piece, and where in the piece it stood. Nothing here is presented as this site's own writing. If you publish German and would rather not be read this way, say so and you will be left out.
Can I trust what it says
Not blindly. A language model wrote every word of German in here. This one is better than expected: asked to conjugate helfen, it produced ihm wird geholfen for the passive, which is right precisely because helfen takes the dative and so has no personal passive.
It is also confidently wrong sometimes, and a learner is exactly the person who cannot tell. So every form, gender and plural carries two controls: right and wrong. Marking something wrong asks you for one sentence saying what is wrong with it, and that sentence is worth more than the count.
The counts show as two numbers rather than a score. One report against two approvals means something different from one report against two hundred, and a single number hides that.
What does it cost
Reading is free and needs no account. Analysing a sentence nobody has analysed spends money on a language model, so each account gets a daily allowance and can ask for more. Everything after that first analysis is free forever. The result is stored, and so is every word it taught the lexicon.
How does my list decide when to show a card again
You can save a whole word, a whole sentence, or a single form. A single form is usually the useful one: helfen, Präsens, er/sie/es is one thing to learn, and it can be solid while ihr helft is still shaky. Forms group under the word they belong to, so a verb with twenty forms is one entry rather than twenty.
You grade each card when you review it, and the grade sets the gap before you see it next. Forgot sends it back to the start. Hard stretches the gap a little. Good multiplies it. Easy multiplies it and then some. A card you keep forgetting keeps coming back sooner, for good, rather than being forgiven after one correct answer. Re-adding a set of forms never resets progress on cards you already have.
Who can read what
The lexicon is shared. Every sentence, word, form and rule is readable by anyone without an account, because a shared lexicon that everyone improves is the whole idea.
Your study list and review history are yours alone. Nobody else can read them, and the storage layer enforces that rather than the application remembering to check.
Why is the rule catalogue fixed
Sixty-nine rules, A1 to B2, written by hand. The analysis may only choose from them. That is what stops one rule appearing under four names and the links between sentences becoming meaningless.
Where a sentence uses something the catalogue lacks, the analysis may propose a rule. The proposal is stored and linked, so nothing is lost, but it stays marked as proposed until a person approves it. You can read the proposed ones under Rules.
Account
Signing in gives you a study list of your own. Reading the lexicon needs no account.
Signing in keeps a study list of your own. Reading the lexicon needs no account.
Admin
Everything here needs an API key, which is the backend itself rather than a person.
API key
Analysing a new sentence asks a language model, so it needs a key. Reading needs none. The key stays in this browser. It also opens these admin pages, so treat it as an administrator credential and do not hand it to a learner.
No key saved. Reading works without one.
The backend
Whether the thing is running. The queue is sentences waiting to be analysed; it should be draining, not growing.
Reported facts
What readers said is wrong. The note is the part worth reading; the counts only say how disputed something is.
People
Everyone who has analysed a sentence, and what they may spend a day. Requests for more are at the top.