Functional motivations behind direct object fronting in written Swedish: A corpus-distributional account

Abstract

In Swedish, grammatical functions are primarily encoded by word order. In prototypical transitive sentences, the subject precedes the direct object. However, Swedish also allows for fronting of the direct object, although such sentences are potentially ambiguous with respect to grammatical functions. This study therefore investigates direct object fronting in written Swedish with respect to 1) which functions this construction serves and 2) whether the use of direct object fronting is dispreferred when the grammatical functions cannot be determined on other information types. These questions are investigated on the basis of quantitative differences in the distribution of NP prominence properties (e.g., givenness and animacy) and formal, morphosyntactic cues to grammatical functions (e.g., case marking and verb particles) between OVS and SVO sentences, and between OVS sentences and passives. The results indicate that direct object fronting is used when the object either is topical and highly discourse prominent, or when it is contrastive. I also argue that direct object fronting is used to introduce new topics into the discourse. Subjects are more frequently high in discourse prominence in object-initial sentences than in subject-initial sentences. I suggest that this stems from a motivation to keep the information in object-initial sentences following the sentence-initial object “informationally light” and predictable. Unambiguous formal markers of grammatical functions are used more frequently in OVS sentences than in SVO sentences, but less frequently in passives than in SVO sentences. OVS sentences also more frequently contain an animate subject and an inanimate object than SVO sentences, and in passives, animate subjects and inanimate objects are even less frequent. Writers therefore seem to prefer the structurally unambiguous passive construction over the potentially ambiguous object-initial construction, when grammatical functions cannot be determined on the basis of other formal markers or an NP argument animacy difference. Further, sentences with two animate arguments more frequently contain formal markers than sentences with at most one animate argument. These findings indicate that writers actively avoid direct object fronting when it potentially results in an ambiguity, and provide evidence for the hypothesis that writers are inclined to actively avoid ambiguities more generally.