In the current studies we investigated English-speaking pre-school and young school-age children’s choice of referential expressions in a referential communication game. In study 1 we crossed two variables: the type of question asked (general vs. specific) and the availability of the listener (present vs. absent). In study 2 we manipulated the number of referents (one vs. two) and the availability of the listener (present vs. absent).
The results showed that all children were very sensitive to the type of question asked, the number of referents was also a reliable predictor of type of referential expression, while the presence of the listener was the variable that posed the largest number of difficulties, even to the oldest six-year-olds. These findings are interpreted against the relative contribution of lexical- and construction-specific knowledge and of perspective-taking abilities in the presence of converging and diverging discourse and perceptual cues.