Are conservative quantifiers easier to learn ? Evidence from novel quantifier experiments

Natural language quantificational determiners seem to always be conservative. Some researchers suggest that one explanation for this link between syntactic form and semantic interpretation may be due to conservative quantifiers being easier to learn. Previous experimental research showed that children learned a conservative novel quantifier (i.e. not all more easily than a novel non-conservative quantifier (i.e. not only) [Hunter and Lidz, 2013]. In a series of four experiments we further investigated this learnability claim. Experiment 1 replicated the original study, but with adult participants. We found no evidence that the conservative novel quantifier was easier to learn. Experiment 2 replicated and extended the original study, testing children. Again, we found no difference between the quantifiers. Experiments 3 and 4 introduce a new training and testing method, situation verification with correction. In Experiment 3, we tested this new method with known quantifier meanings, conservative all and non-conservative only, using a novel quantifier expression. Children were highly successful, and there was no difference between the two quantifier meanings. In Experiment 4, we used the same method to teach children the novel conservative and non-conservative meanings not all and not only. Most children were unable to learn either quantifier, and we again found no difference between the conservative and non-conservative quantifier in terms of learnability. In conclusion, in four experiments, we failed to replicate findings that conservative quantifiers are easier to learn. These results suggest that a learnability advantage is not a plausible explanation for con-servativity's universality.