Nowadays, computational approaches to language are mainly based on the concept of distributional semantics. But how do humans learn language? And how to operationalise these processes computationally? In this talk, I will discuss how to operationalise one of the theories of usage-based language acquisition computationally. This theory states that humans learn language through the processes of intention reading and pattern finding. Intention reading is the process in which children try to reconstruct the intention behind linguistic utterances. The process of pattern finding is the process of finding similarities and differences in the form and meaning side of linguistic observations. These processes can be operationalised using computational construction grammar. Specifically, pattern finding can be operationalised through anti-unification over form and meaning. This leads to grammars that are robust and adaptable. To conclude the talk, I will discuss how similar construction grammars can be used for frame extraction and value detections in social media.