Spoken sentence production: incremental planning for fluent output.
Speaking is a demanding task involving several stages in which linguistic representations must be retrieved and structures must be generated in order to convey our thoughts. Put simply, speakers need to select appropriate words and articulate them in the correct order, ideally in a timely and a fluent manner. There is now a great deal of evidence that we plan our utterances incrementally, initiating the articulation of early parts of an utterance and planning the rest as we speak. Investigating the nature of the planning increments employed during fluent sentence production can provide a great deal of information about speech production processes as well as the factors that constrain them: in particular information about the relationship between words and structure. According to some theories, syntactic planning is controlled by processes responsible for the retrieval of words. Alternatively, syntactic structure may be generated independently of lexical access and may even function to constrain it. Moreover, at the interface with sound structure, we can generate representations that correspond to neither lexical nor grammatical units. I will review research that has focused on planning scope during the production of fluent sentences. I will consider to what extent planning scope is determined by linguistic structures and to what extent it is flexible and subject to cognitive constraints, as well as whether the answers to these questions differ for different levels of linguistic representation – grammatical, lexical and phonological.
