Rivera-Vera, N. 1 , Monaghan, P. . 1 , Andringa, S. 1 & Kronmüller, E. 2
1 Universiteit van Amsterdam
2 Pontificia Universidad Católica de Chile
When acquiring a new word, learners must determine the co-occurrence of the word with its meaning, as well as assess the speaker?s competence in labeling objects. We adapted a CSWL task by manipulating the reliability of word-object co-occurrences, resulting in two experimental conditions that translate into how consistently (?reliably?) the speaker mapped a word to an object. We aimed to test whether learners weigh the speaker?s reliability in CSWL. In the reliable condition, the speaker taught two sets of eight reliably mapped words (co-occurrence target probability of 1.0). In the unreliable condition, the same speaker mapped one set of eight words unreliably (co-occurrence probability of 0.25), and the other set reliably. We predicted that reliably mapped words would be more difficult to learn in the unreliable condition, and that learners? judgment about the unreliable speaker would account for this result above and beyond our manipulation. Descriptive statistics of pilot data (n=8) show that participants in the reliable condition tended to select the target object at a higher rate (M=89%, SD=11%) than participants in the unreliable condition (M=81%, SD=15%). We are currently collecting data (n=60) to assess whether the observed difference is statistically significant, and whether participants weigh the speakers? competence.