Semantic space models are computational representations of word meaning based on co-occurrence counts from large corpus data. This means that the meaning of each word is described by the contexts in which it occurs. Although these data-driven models have proven to be well suited for capturing a wide range of semantic information (such as similarity of synonyms and relevance of correlating words), the meaning aspects they cover have not been fully explored. In this talk, I will present studies on the suitability of automatically built semantic space models for representing meaning in /frame semantics/, an empirical semantic theory that emphasizes on the relation between language and experience