This book investigates the syntax and usage of fragments, such as "A coffee, please!" which serve the same communicative function as the full sentence "I'd like to have a coffee, please!" Despite their common use, these utterances challenge the central role of the sentence in linguistic theory, particularly from a semantic viewpoint. The first part focuses on the syntactic analysis of fragments through experimental methods, addressing several competing theoretical analyses that rely mainly on introspective data. The experiments conducted provide a systematic evaluation of crucial predictions, supporting an in situ ellipsis account of fragments, as suggested by Reich (2007). The second part explores why fragments are used and when they are preferred over complete sentences. While syntactic accounts impose licensing conditions on fragments, they do not clarify why fragments are sometimes preferred when their usage is permitted. This book proposes an information-theoretic account, predicting that fragment usage is influenced by a tendency to distribute processing effort evenly across utterances. Two empirically confirmed predictions emerge: speakers tend to omit predictable words and add redundancy before unpredictable ones.
