How to read the 3D grammar coordinates

Why three dimensions

On a flat whiteboard a sentence is one line: subject, verb and object next to each other, with attributives and adverbials squeezed in between, and clauses nested on top. Once layers stack up, the structure blurs. The 3D view separates those layers: the backbone runs along one direction, modifiers hang off another, and clauses get their own level. “Which noun does this attributive modify?” and “which part does this clause belong to?” stop being guesses.

The eight component roles

Click “3D analysis” and the sentence unfolds by role:

• Subject — what the sentence is about
• Predicate — the action or state
• Object — what the action affects
• Predicative — the complement after a linking verb
• Object complement — extra information about the object
• Adverbial — time, place, manner, reason
• Attributive — what modifies a noun
• Clause — a whole level that is analysed again by the same rules

Six sentence-pattern skeletons

The toolbox ships hundreds of blocks covering six skeletons: subject + verb, subject + linking verb + predicative, subject + verb + object, subject + verb + two objects, subject + verb + object + object complement, and there-be sentences, plus tense transformations, pattern changes and phrase/clause components. Sentences generate live as learners build, with gentle hints when something is off — structure learned by doing sticks better than rules memorised by heart.

Offline engine vs. AI

Offline static engine (no network, no API key)

• The five major sentence patterns
• Tense changes
• Passive voice
• Questions and there-be sentences
• Comparatives

For everyday practice sentences, the offline engine already returns a full 3D structure.

AI deep analysis (optional)

For long sentences, nested clauses, non-finite structures or rare words, connecting an AI provider returns precise sentence-type labels and a multi-level structure tree. DeepSeek and OpenAI-compatible providers are supported; the key is entered by you and stored on your device only.

Using it in class (teacher view)

1. On a projector, load an example of the target pattern from Project Learning (Alt+P) instead of building it live.
2. Ask students to guess the backbone first, then run 3D analysis to check — “guess, then verify” sticks far better than showing the answer.
3. Have students build similar sentences themselves in free build (Alt+B); Esc returns to the previous screen.
4. Export the classroom sentences as JSON and hand them out as homework templates.

FAQ