Our first step was to use empirical data to identify the right words to teach kids, so that even infants would be engaged by the show, while allowing older children - 2 and 3 year olds – to also benefit. To do this, we used Stanford University's online tool called "Wordbank" a website that reports data from literally thousands of children around the world, and describes which words they use and when they begin using them. This tool allowed us to identify "foundational" words, which we felt children would be likely to be already working on but need to hear more to firm up their knowledge, and what we called "aspirational" words, which we felt most children would be unlikely to know, but were just "around the corner" developmentally. Also, we didn't settle for words that act like simple pointers to objects – like "dog" and "cat" – but tried to focus on language that requires children to build new concepts, like words that describe perception, thoughts, space, emotions, and opposites like "fast" and "slow," or "up" and "down." Finally, we grouped words into what we called "domains" – groups of words that have something in common, but contrast with each other in specific ways. This choice was based on science telling us that children begin figuring out hard, abstract, words by assembling them into groups – sometimes called "lexical classes" – and by comparing them with each other, allowing them to learn the boundaries of what counts as "red" by figuring out "pink" and "purple", or to learn what counts as "square" by seeing examples of things that aren't squares, like diamonds, triangles, and hearts. In Word Party, these groups of words were then turned into songs and chants, to help children identify meaningful lexical classes, and to learn not just by associating sounds with objects, but also by reasoning about how words contrast with one another, and how they work together to explainaspects of their world.