The distinction between linguistic and non-linguistic knowledge is particularly relevant because it is related to the principle of compositionality during sentence comprehension. Hagoort, Hald, Bastiaansen, and Petersson (2004) challenged the distinction between linguistic and non-linguistic knowledge. Here, we investigate how linguistic and non-linguistic violations are processed in a setting adapted from Hagoort et al., whilst in contrast to Hagoort, keeping the critical word identical. In line with the findings by Hagoort et al., our results showed largest N400 amplitudes for semantic violations (‘Journeys are stripy’), followed by non-linguistic world-knowledge violations (‘Ladybirds are stripy’) and contingent sentences (‘Trousers are stripy’), and finally by correct sentences (‘Zebras are stripy’). Traditional fractional area and relative criterion measures of peak and onset latencies showed no effect of violation type. Interestingly, the semantic violation condition crossed a fixed criterion earlier than the word-knowledge violation condition. In conclusion, our data suggests that the question regarding the distinction between linguistic- and non-linguistic knowledge in terms of language integration remains open. Implications for future studies addressing the difference between linguistic and non-linguistic knowledge are discussed.