摘要：Producing a sentence involves retrieving lexical items and integrating them into a functional (semantic-togrammatical) structure. Agrammatic speakers are impaired in this grammatical encoding (GE) process (e.g., Bastiaanse& Zonneveld, 2004); however, the nature of the deficit remains unclear. This study investigated how linguisticknowledge is used during real-time GE in young, older, and agrammatic speakers, testing two competing sentenceproduction models. In the incremental model, GE proceeds in a word-by-word manner; thus, speakers begin theirutterance upon planning the first word only (Griffin, 2001). In the structural model, GE is guided by verb argumentstructure (VAS); hence, speakers plan up to the verb at least before speech onset (Ferreira, 2000).
