Abstract
In this poster, I suggest a three-stage model for L1 teaching at primary and secondary school level and argue for a different type of corpus data for each stage. Although the proposal is modelled to fit Czech education system with its two compulsory levels of primary school, the recommendations strive to be fairly universal and applicable to other languages as well.
At stage 1, pupils are to get acquainted with corpus materials and methods via texts that are most accessible to them, i.e. popular children's fiction. The purpose of this stage is to introduce a new look at language, emphasising the role of context and word combinations (collocations).
At stage 2, new corpus data should be introduced, such as different text types (non-fiction and newspaper texts) and registers (non-standard spoken language in addition to the standard written). At this point, students learn about the variability of language and relevant linguistic means for different communication situations.
At the final stage 3, students at secondary schools already know a lot about how language works, so they can focus more on specific types of word combinations, such as terminology (based on academic texts), set phrases and idioms (especially those used in contemporary texts, such as newspapers) and typical grammatical/lexical patterns (also in comparison to foreign languages based on parallel corpora).