We mined the vowel database for the possibility that certain experimental features would affect effect sizes.
What is the best ISI to use? Some research suggests that infants need a considerable silence (interstimulus interval, ISI) to detect a change, since they fail with short ISIs as shown in conditioned headturn (Cowan, Suomi, & Morse, 1982) and event-related potentials (Pihko et al., 1999). Taking the database as a whole, there is no clear relationship between ISI and effect size, once differences across methods are taken into account.
Are infant-directed, female-spoken, natural stimuli the best discriminated? We couldn't compare results based on voice, because CHT and HAS only used male voices, whereas AEM, HPP, and CF only female voices. We couldn't investigate register either, since AEM, CHT, and HAS relied on adult-directed speech only, whereas CF and HPP studies used infant-directed speech only. Finally, there is no clear advantage for fully natural over synthetic stimuli (controlling for method).