Consider the sentence “The little girl blicked the kitten on the stairs.” As English-speaking adults, we’re more likely to interpret this as an event involving a little girl doing something to a kitten and that event happening on the stairs (as opposed to, say, a kitten doing something to the stairs, and that event happening on a little girl). This is because we’ve solved the problem of how thematic roles link to their syntactic positions for English verbs -- even if we’ve never seen the verb before. Current proposals for the nature of the linking theories we use to do this vary on (i) how thematic information is used (fixed categories like UTAH vs. relativized like rUTAH), and (ii) whether the links are innate or derived from language experience.

Here, I investigate several linking theory proposals from the perspective of acquisition. I use an empirically-grounded quantitative acquisition framework involving realistic child input, child behavioral data, and developmental modeling to determine which proposals are compatible with English children's behavior at different ages. My findings support either innate links or derived links, but find only relativized thematic information is consistently compatible with children's behavior. I also use empirically-grounded developmental modeling to investigate concrete proposals for how children could successfully derive linking knowledge from their input. I again find support for relativized thematic information over fixed categories.