Grammarlands are notional places where learners can explore simple language. My first example was John and Mary, which was published in Computers in Language Learning by Higgins and Johns in 1983, including a full listing in Sinclair BASIC. The screen showed two rooms, a connecting door and two figures identified as John and Mary. The learner's task was to ask questions to find out what the computer knew and could understand, or to manipulate the figures by giving instructions. I have tried to create something similar here, adding a third room, a second door, and a third character — Rex the dog.
Two people, now joined by Rex, in a small house of three rooms and two doors. Ask where everyone is, work the doors, and find out who can see whom — or leave the input box empty and press Enter to let them get on with things by themselves.
A second grammarland, from 1985: six people, each shown doing something that has nothing to do with their job. Ask what they do for a living, ask what they're doing right now, teach the machine things you think are true, and see how well it keeps everything straight.
A new grammarland: a country-house murder. Six suspects, a plan of the house, and a gunshot everyone heard from somewhere different. Question each suspect about what they were doing when the shot was heard and what they did afterwards, then catch the culprit out by finding the two accounts that don't add up. Watch an interrogation unfold, or conduct one yourself.
A gentler grammarland, with a nod to A. A. Milne's Tigger and his very particular tastes. Somewhere in the larder is one thing Tiglet actually likes; offer him things and he'll turn them down — too sweet, the wrong colour, a food-group he won't touch — until you narrow it down to the one he wants. Type an offer, or leave the box empty and press Enter to watch him try things himself.