Let’s drop the object : An experimental study of the implicit object construction in American English

The topic of transitive structures with implicit (phonologically null) objects has received much attention in Construction Grammar (CxG). Among natural languages, English is characterized by allowing mainly non-referential implicit objects (‘They build _ for a living’), whereas referential implicit objects are highly constrained and/or require conversational implicatures (Fillmore 1986 ; Cummins & Roberge 2004 ; Goldberg 2001). Using the CxG framework, this article investigates discourse-based acceptability of the Implicit Object Construction (IOC) in two context types: Generic (no antecedent for an implicit object) and Definite/Indefinite (+antecedent for an implicit object). One of the hypotheses explored is that English allows implicature-based referential construals for implicit objects (Cummins & Roberge 2004). Two online experiments were conducted. Experiment 1 included 31 native speakers of American English (AE) who provided acceptability rankings for the IOC in three experimental conditions (Generic, Indefinite, Definite). Experiment 2 included a different group of 24 AE native speakers and used the same materials as in Experiment 1; however, the ranking component of the task was replaced with a multiple-choice format. Experimental results converge on two findings: an IOC is clearly favored in the [-antecedent] contexts, whereas overt object counterparts of an IOC are preferred in [+antecedent] contexts. However, implicit objects, when used with some verbs, are ranked as quite acceptable with a [+specific] antecedent, which attests to the role of individual verbs in speakers' judgments.1