Grammar anomalies occur in all kinds of written texts – whether the author is a second language learner or a native speaker. Existing language comprehension models are based on processing studies with either constructed anomalies (e.g. Bornkessel & Schlesewsky 2006, Friederici 2002, Hagoort 2005) or with no anomalies at all (e.g. Ferreira & Patson 2007). These models give much insight into the cognitive foundation of grammatical processing, but little insight into the comprehension of everyday “broken grammar” and its role in communication.

The aim of the research project Broken Grammar and Beyond (BGB) is to explore patterns in grammar anomalies in naturally occuring texts and carry out experiments that examine the processing of naturally occuring grammatical anomalies.
In the talk I present the BGB corpus which contains texts from both 71 highly proficient speakers of Danish (high school essays) and from learners of Danish (232 essays from a language school in Copenhagen). I will give examples of how the BGB group have analyzed data from the corpus and used the corpus to construct experimental stimuli for reading experiments, and I discuss some of the findings that surprised us along the way.