Using the previously explained learning mechanisms, the agent has acquired a number of item-based and lexical constructions. This learning mechanism covers the cases where previosuly acquired constructions cover the observed utterance, but where the slot-and-argument combination has not been observed before. In such cases, the agent simply adds the slot-argument link to the categorial network.

| shape-category |
| |
| score: 1.000 |
| |
| score: 1.000 |
| |
| score: 1.000 |
| |
| score: 1.000 |
| |
| score: 1.000 |
| |
| score: 1.000 |
| |
| score: 1.000 |
| |
| score: 1.000 |
| initial structure | ||||||||||
| application process | ||||||||||
| applied constructions | ||||||||||
| resulting structure |
| shape-category |