Priming syntactic ambiguity resolution in children and adults

ABSTRACT

Adults use their recent experience to disambiguate ambiguous sentences: Structures that have recently been primed are favoured in the resolution of different types of ambiguity, an example of structural priming. Research on children's use of recent information for disambiguation is scarce. Using a forced-choice task with a tablet, we asked whether 5–6-year-old French-speaking children could also be primed in the resolution of attachment ambiguities, as well as whether listeners are affected by the proportion of primes of each structure, and whether priming is cumulative. We found that both children and adults can be primed, and are sensitive to the proportion of structures in the input, and that priming effects cumulate as the experiment progresses. This is the first study showing priming of ambiguous sentences at 5–6 years, suggesting that children, like adults, use recent experience as a source of disambiguating information.

KEYWORDS:

• Adaptation
• priming
• ambiguity
• language
• children

Acknowledgements

We would like to thank Anne-Caroline Fiévet, our babylab manager for her help with recruitment and testing, and Marion Beretti and Alex de Carvalho for their help in testing. We thank Shravan Vasishth for statistical advice, Alex Cristia for allowing us to use the tablet application, Sharon Lim Sautarel for drawing the stimuli for the experiment, and the schools, parents and children for their collaboration.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Including these three children in the sample does not change our pattern of results.

2 Some sentences are more conductive to noun-attachment interpretation than others. For example, “the boy is kissing the girl with her/his tedding bear” is a better noun-attachment candidate than “the cat is hitting the mouse with the cheese”. In order to control for this variable, we used counterbalanced lists. An equal number of participants saw the first sentence at the beginning of the experiment as those who saw it towards the end. However, precise models of how the number of previous primes affected a specific target might be noisy, because which primes were preceding that specific target was not fully randomized.

3 That is, we first fit a maximal model including all possible random intercepts and slopes, and then ran a principal component analysis of the random effects structure to determine the number of variance components and correlation parameters supported by the data (using function rePCA from the package RePsychLing, Baayen et al., 2015). This function identifies the most complex model supported by the data, and then, to simplify the model, relies on comparisons of goodness of fit of nested models with likelihood ratio tests (LRTs). We chose this procedure rather than the more prevalent use of the maximal effect structure allowing the model to converge (Barr et al., 2013), because this procedure allows for a robust random effect structure without including unnecessary random effects thereby losing power. Indeed, a maximal random effect structure did not converge with our analysis, which is often the case with complex random effect structures.

Additional information

Funding

This work was supported by a postdoctoral grant to the first author from the French Embassy in Israel and The Victor Smorgon Charitable Fund, and by the French Agence Nationale de la Recherche [grants n° ANR-13-APPR-0012 LangLearn, ANR-17-CE28-0007-01 LangAge, and ANR-17-EURE-0017 FrontCog].