A growing body of work has demonstrated that children, like adults, can be
primed to produce sentences with a particular structure (e.g., [1-2]).
These priming studies have focused on sentence structures with very similar
meanings (e.g., active vs. passive, prepositional object vs. double object
datives). It is unclear, however, whether a particular interpretation can
be primed when a sentence is compatible with multiple meanings (i.e., it is
globally ambiguous). For example, in the sentence in (1), the pipe cleaner
could either be the instrument of pointing – i.e., the sheep uses it to
point ([VP [VP points at the pig] [PP with the pipe cleaner]]) – or modify
the description of the pig – i.e., the pig holds it ([VP points at [NP the
pig with the pipe cleaner]]).
(1) The sheep points at the pig with the pipe cleaner.
Previous work demonstrated that children rely on verb biases to determine
how to interpret these sentences [3]. Verbs fall into 3 classes:
instrument-biased, modifier-biased, and equi-biased. The current study
examines whether children’s interpretation of sentences with equi-biased
verbs can be primed by producing sentences with one of these
interpretations.
*Control* (N=16) The classification of the 8 verbs used in the study (
*scratch*, *throw*, *pinch*, *feel*, *drag*, *turn over*, *blow on*, & *point
at*) as equi-biased was based on adult sentence continuations [3]. Thus, we
first conducted a control experiment to uncover children’s bias. Both
experiments utilized a picture selection task that the child played with a
puppet. In a particular trial (*n*=24; 8 targets, 16 fillers), the puppet
would describe a picture and the child had to select the matching picture.
For the target trials (Fig1), the puppet described the picture using an
ambiguous sentence like (1) containing one of 8 equi-biased verbs. Despite
this classification, children (mean age=5;2, range=4;1-6;4) selected the
picture matching the instrument interpretation 82.12% of the time. Thus,
this value will be used as the baseline in the main experiment.
*Exp *(N=32) Unlike in the control experiment, the child and the puppet
took turns describing a picture and selecting the matching picture from two
options. For the production portion of the task, an ambiguous sentence like
(1) described the pictures, but the image itself disambiguated its meaning.
When it was the child’s turn to select the match, the choices depicted the
two alternative interpretations of the target sentence (Fig1) and
technically either answer was correct. Crucially, the child’s production
could serve as an interpretative prime; if they produced a sentence with a
particular interpretation, they may be primed to select that same
interpretation when encountering another ambiguous sentence of the same
type. Children (mean age=5;7, range=4;5-6;7) were randomly assigned to
either produce instrument or modifier interpretations. Each trial (*n*=8)
consisted of a pair of picture selections. First, the puppet selected a
picture based on the child’s production, then the child selected based on
the puppet’s production. The modifier priming group was less likely to
select the picture corresponding to the instrument interpretation than the
instrument priming group (Fig2; *β*=1.25, *p*<0.001). Older children were
significantly more likely to select the instrument interpretation (*β*=1.49,
*p*<0.05), but there was no interaction between age and priming group.
Compared to the baseline established in the control experiment, priming
increased the selection of the picture with the matching interpretation for
both priming groups (instrument: *t*=3.55, *p*<0.01; modifier: *t*=-2.19,
*p*<0.05).
These results suggest that interpretation can be primed by the production
of a sentence with a similar interpretation. Implications for priming and
the role of production will be discussed. Also, an in-progress follow-up
experiment aims to compare the priming effects arising from ambiguous
primes to those following unambiguous primes (e.g., *The sheep points at
the pig* [instrument: *by using* / modifier: *that has*]* the pipe cleaner*
).
*References* [1] Huttenlocher, Vasilyeva, & Shimpi 2004. *JML*. [2] Savage
et al. 2004. *Developmental Science*. [3] Snedeker & Trueswell 2004. *Cognitive
Psychology*.