For two years, @mbraginsky, @danyurovsky, Virginia Marchman, and I have been working on a book called "Variability and Consistency in Early Language Learning: The Wordbank Project" (@mitpress).
Here's our draft: langcog.github.io/wordbank-book/…
[+ a thread with a few of our findings]
Here's our draft: langcog.github.io/wordbank-book/…
[+ a thread with a few of our findings]
We look at child language using a big dataset of parent reports of children's vocabulary from wordbank.stanford.edu, w/ 75k kids and 25 languages. (Data are from MacArthur-Bates CDI and variants). Surprisingly, parent report is both reliable and valid! langcog.github.io/wordbank-book/… 
First finding: It's long been known that children are variable with respect to language. The striking thing is that the level of variability is very consistent across languages. The world around, toddlers are all over the place with respect to language! langcog.github.io/wordbank-book/…
Girls also have a bit bigger vocabularies than boys on average. This female advantage is quite consistent across languages as well. (There are tons more demographic analyses here: langcog.github.io/wordbank-book/…).
On the other hand, the "noun bias" - how much young kids over-represent nouns and under-represent verbs in their early language - varies quite a lot across languages! langcog.github.io/wordbank-book/…
You can also apply this analysis to semantic categories and see that there are some categories that are very consistently over-represented in children's early language, like body parts or words for social routines. Others, like clothing, are more variable. langcog.github.io/wordbank-book/…
Children with bigger vocabularies also have stronger grammatical abilities, and this linkage is really quite strong and consistent across languages with pretty different grammars. (This confirms work by Liz Bates and colleagues). langcog.github.io/wordbank-book/…
But despite all this consistency, children do show stable individual differences in both the rate and the style of their language learning. Again confirming previous work, some are more "referential" (noun-y) and some more "expressive" (combining words). langcog.github.io/wordbank-book/…
You can even quantify which of these findings are relatively more consistent across languages than others, leading to a new data-driven approach to what aspects of language learning are relatively more "universal" (at least in our sample) than others. langcog.github.io/wordbank-book/…
This naturally leads to a different perspective on language learning than you see in classic nativist/empiricist clashes. If you're interested in reading more about the theoretical perspective that emerges, we try to tie it all together here: langcog.github.io/wordbank-book/…
This work wouldn't have been possible without the pioneering research of Larry Fenson, Philip Dale, Liz Bates, and many others who developed the CDI forms and then shared the data and encouraged others to adapt the form across dozens of languages! mb-cdi.stanford.edu
Finally, all this work is done in bookdown and should be fully reproducible, with code at github.com/langcog/wordba…. It's also work in progress, so we'd love any and all comments. Feel free to email or DM me, or file a github issue: github.com/langcog/wordba…
