One of the studies that established that the key characteristic of so-called infant-directed speech is not more canonical or more consistent vowels, but more extreme vowels (a larger vowel space--bigger mouth for /a/, wider spread lips for /i/, etc.)--which is good not only in and of itself but because it reminds us that phonemes are abstractions and not real things that we learn. We start with difference, and only from there home in stochastically, approximately, on saying things the way we say them (as opposed to starting with some pristine abstract structure, whether innate or acquired, and only then getting sloppy). Liu et al. also devoted some time to demonstrating that this expanded vowel space (which they call "clarity of speech," a bit confusingly, since that could also refer to canonicity as well as distinctness) has nothing to do with socioeconomic status, which was a dumb thing that people thought once. Paper appeared in Developmental Science. ( )