Linguistic structures that appear to be different on the surface may be linked. This dissertation contains three sets of studies that investigate the cognitive mechanisms behind the relationships among linguistic structures on the syntactic, morphological, and phonological levels. Chapter 1 examines how speakers learn to generalize word orders in sentences in three artificial language learning experiments. Results suggested that learners have linguistic biases that mirror typological differences, which help them go beyond simple statistics tracking. Chapter 2 investigates whether structuring mechanisms are shared across linguistic units of different grain sizes. In three structural priming experiments, we tested whether priming of attachment preferences occurs between words and sentences. Results showed that priming only occurred within- but not across-grain size, suggesting that structuring mechanisms for words and sentences are not shared. Chapter 3 explores phonological representation in the production of tonal languages. In two speeded repeated production experiments, we confirmed that Mandarin Chinese speakers use syllables (rather than segments such as consonants and vowels) as basic planning units. Additionally, we discovered that lexical tone is special in phonological processing, in that it is integrated with syllables relatively late in processing and in ways that are different from how segments in non-tonal languages are represented and combined.