When processing sentences, the words come in one after the other, as if they are beads on a string. From the outside, then, sentences are linear, but underneath that linear sequence is a certain type of structure. That structure is hierarchical, meaning that words are combined into constituents, which are combined again with other words into larger constituents, and so on. To understand sentences, we have to construct their hierarchical structure on the basis of linear, word-by-word input. This thesis aimed to shed light on different aspects of the role hierarchical structure in language use, asking question such as: are people biased to interpret language hierarchically or linearly? Are artificial neural network models well-equipped to learn to behave like humans? How does the brain infer hierarchical structure during language comprehension? And is the hierarchical structure of language found in other domains of cognition? By showing strong effects of hierarchy in both behavior and brain activity, this thesis shows how one of the organizational properties of our linguistic competence strongly drives our linguistic performance.
Share this page