Words are better identified in a sequence of words when this sequence forms a correct sentence compared with ungrammatical word sequences. Here we examined whether this sentence superiority effect (SSE) can be modulated by the predictability of the target word given the sentence context. Effects of semantic predictability were measured by comparing the SSE obtained for predictable words (e.g., elle tombe dans un trou [she falls into a hole]) and less predictable words (e.g., elle tient dans un trou [she fits in a hole]) in syntactically correct word sequences, with effects being measured against ungrammatical scrambled versions of these sentences. Results replicated the SSE in both conditions, with a larger effect with less predictable words. We conclude that the SSE is primarily driven by the syntactic structure of word sequences, and this compensates for effects of semantic predictability driven by semantic associations between words.