Discourse effects on older children’s interpretations of complement control and temporal adjunct control

ABSTRACT

The reference of understood subjects (ecs) in complement control (John persuaded Peter_i ec_i to read the book) and temporal adjunct control (John_i tapped Peter while ec_i reading the book) has long been described as restricted to the object and subject of the main clause respectively. These restrictions have shaped the grammatical targets proposed for children, most of whom are reported as having acquired both subtypes by age 7. Using three picture-selection tasks, 76 children’s (34 girls; aged 6;09–11;08) interpretations of the ecs were tested. Task 1 established their baseline preferences. Task 2 weakly cued the ecs toward an alternative referent and Task 3 strongly toward an alternative referent. Complement control responses were consistent across all tasks, but in adjunct control they shifted significantly toward the object in Task 3—a pattern mirrored by 15 adults. Responses in adjunct control also exhibited a degree of fluctuation in the baseline condition that complement control did not. A follow-up study on adjunct control showed that neither children nor adults permitted an external-referent reading, even when strongly cued in that direction. Two alternative proposals are discussed: one in which the results are viewed solely as the product of a parser’s sensitivity to activation and another that proposes two possible structures for adjunct control; this permits the evident interpretation shift yet gives precedence to the highly preferred subject-oriented reading.

Acknowledgments

Foremost thanks to the staff, children, and parents at Herne CE Junior School, Kent; Wittersham CEP School, East Sussex; St Edwards Catholic Primary School, Sheerness, Kent; and St Peters Primary School, Canterbury, Kent. Thanks also to Donna Mulhall and Sam D’Elia for their help with data collection. For useful comments and discussion, I am grateful to Helen Goodluck, Christina Kim and Nausicaa Pouscoulous as well as to audiences at the LAGB 2015 and BUCLD 40. For statistical support and advice, my thanks to Gordon Craig. Finally, I gratefully acknowledge two anonymous reviewers’ constructive criticisms and suggestions.

Notes

¹ Landau (2000) draws a distinction between exhaustive and partial control within the OC classification. These subtypes are not discussed here. Examples combining a partial control verb in the matrix clause with a collective predicate in the infinitival, which invites a partial control reading for some, have been avoided.

² These were a group of typically developing (TD) children, whose performance had been compared with high-functioning children with autism. Here we refer only to the TD group.

³ The task was piloted on several younger children in Year 1 (aged 5 to 6), as it would have been interesting to see how children still at the age where temporal adjunct control can be free (for a subset) would have performed. But the task proved too difficult. They scored badly on the cued fillers, indicating they could not yet ignore an infelicitous context for a standard SVO sentence, as well as on persuade and order in complement control. They also struggled with the vocabulary check, giving responses for persuade such as “It’s when you really like someone.” A subsequent study on younger children could design a shorter version of the task with different vocabulary items.

⁴ Two types of nonobligatory control tested in the same battery are reported on in separate work.

⁵ Two further control conditions were included in the battery (one testing an SVO-embedded sentence (e.g., Harry said that Hermione is waving the wand) and another testing understanding of a cause relation (e.g., The water made Harry wet). These are relevant to the aforementioned NOC sentences reported on in separate work.

⁶ The verb tell has been standardly used in acquisition studies on much younger children than those tested here, so it was not included. Three further words (prepare, try, awkwardly) were also tested. The latter two are not relevant to the OC trials and so are omitted here. The results for prepare, a word used in one of the pragmatic leads, have been added to the appendix, as per the suggestion of an anonymous reviewer.

⁷ All p values are Sidak corrected.

⁸ Testing for this second study took place at the beginning of the academic year, unlike the testing for Study 1, which was undertaken over the summer. For this reason, only Years 3, 4, 5, and 6 children were included, as the Year 2 children were too near in age to the children who were not able to ignore infelicitous leads (i.e., in fillers and CC) in Study 1. See footnote 4.

⁹ Some participants rejected the test sentences altogether. See Janke & Bailey (2017) for another experiment testing the same phenomenon on a subset of those participants.