UC Santa Cruz

Compounding, the creation of words by combining two or more words, has long been a topic of interest in various fields of linguistics. This is because compounds seem to straddle the boundary between “words” and “phrases” and have some amount of internal structure (Scalise and Vogel 2010). This raises a question for research on the syntax-prosody interface about how syntactic structure is mapped onto prosodic structure when compound words are concerned. Whereas the fundamental difference between syntactic structure and phonological structure has traditionally been held to be that syntactic structure allows recursion and phonological structure disallows recursion, phenomena like compounding call to question whether disallowing recursion in phonological structure is tenable. If recursion is allowed in phonological structure, however, the issue becomes just how much recursion is allowable. Compound words provide a crucial case for investigation because while they seem to act like words on the one hand, a well-known property of compound words in many languages is that they are infinitely recursive, such that novel compounds can be created productively. Furthermore, compounds have been noted to be able to include phrasal structure such as sentence fragments. It is plausible, then, to expect that recursion may occur in phonological structure when considering such recursive cases and cases in which phrasal structure is involved.

Compounds in Japanese have long been observed to exhibit a large variety of compound prosodies. Ito and Mester (2003, 2007, 2018a, 2021) have developed a theory that accounts for the variety of compound prosodies in Japanese by crucially proposing that recursive structure is involved. The theory they develop predicts a set of structures of which a subset is observed in Tokyo Japanese. In this dissertation, I demonstrate that Kansai Japanese exhibits a compound type, which I refer to as the word-phrase compound, which was predicted by Ito and Mester’s theory, but which was not observed in Tokyo Japanese because Tokyo Japanese does not have the prosodic phenomena required to diagnose it. Accordingly, this serves as a confirmation of the theory. This dissertation demonstrates, based on the crucial similarity of compounds with non-compound words and phrases in Kansai Japanese, that recursive structure naturally and elegantly predicts and explains the typology of compound prosodies in Kansai Japanese, and that an approach that does not make use of recursive structure requires positing additional prosodic categories which may not otherwise be well-motivated.

Although a theory involving recursion in prosodic structure predicts the word-phrase compound in Kansai Japanese, however, an interesting problem arises when attempting to account for the prosodic structure through typical syntax-prosody mapping mechanisms. Whereas the compound types in Tokyo Japanese and non-word-phrase compounds in Kansai Japanese are generally straightforwardly derived based on the length of their second component, word-phrase compounds cannot be. In this dissertation, I explore the possibility that non-syntactic, non-phonological, and non-morphological factors are involved in this mapping, extending Bell and Plag (2012)’s work, which suggests that informativeness, a gradient, frequency-based factor, has an influence on right-hand stress in English compounds. Based on novel fieldwork data collected for this work, I demonstrate that informativeness may play a role in whether a compound in Kansai Japanese can have the word-phrase prosodic structure, suggesting that non-syntactic, non-phonological, and non-morphological factors may be important for syntax-prosody mapping as well.