Mapping “Easy ” and “Hard ” Messages onto Language: Conceptual and Structural Variables Jointly Affect the Timecourse of Sentence Formulation

Sentence formulation requires mapping pre-verbal messages onto linguistic structures. This message-to-language mapping is often evaluated in eye-tracking tasks where speakers describe pictured events (The dog chased the mailman). Speakers can begin sentence formulation by quickly selecting the first-fixated character as the sentential starting point (lexical incrementality), or generating a rudimentary sentence plan based on their construal of the event gist before selecting a starting point (hierarchical incrementality; Kuchinsky & Bock, 2010). Lexical incrementality predicts fast divergence of fixations while hierarchical incrementality predicts slower divergence of fixations to the two characters within 200ms of picture onset. Speakers have some flexibility in their choice of a more lexically or more hierarchically incremental planning strategy (Konopka & Meyer, 2011). Here we tested whether the timecourse of sentence formulation in early (0-200ms) and later time windows (200ms-speech onset) changes with the ease of encoding the event gist (the who-did-what-to-whom event structure) and generating a suitable linguistic structure. 48 eye-tracked speakers described a series of pictures, including 60 two-character target events that elicited descriptions with the preferred active or dispreferred passive construction. In “easier ” (more codable) events, the characters performed unambiguous actions (e.g., hitting) that were described with a small set of verbs, while “harder” (less codable) events were described with a wider range of verbs depending on speakers ’ interpretation of the event (e.g.