Measuring the vocabulary spurt: a reply to Mervis & Bertrand