Prosodic Disambiguation of Syntactic Clause Boundaries in Korean

This study explored the effects of prosodic boundaries on the comprehension of temporarily and globally ambiguous sentences in Korean. Previous work on English, Japanese, and Korean has demonstrated that prosodic structures can carry critical information about the meaning and/or syntactic structures of spoken sentences (Kjelgaard and Speer 1999; Jun and Oh 1996; Misono et al. 1997; Venditti 1993; among others). These languages provide an interesting contrast in terms of the prosodic and syntactic information available to the listener during sentence comprehension. In contrast to English, Korean and Japanese are head-final, pro-drop languages so that both temporary and global syntactic ambiguities are frequent. For example, in a head-final language, multiple NPs are encountered before the verbs to which they are attached. In addition, syntactic clausal boundaries are marked by a complementizer at the end of the clause so that clause boundaries cannot be detected until this morpheme is encountered. Thus, some or all the NPs encountered are subject to misanalyses in sentences with more than one clause. The pro-drop nature of Korean also leads to syntactic ambiguity. Unlike English, in which all the arguments of verbs are usually realized phonologically, pro-drop languages allow deletion of verbal arguments. Hence, a string of words can be interpreted in many different ways depending on where those dropped pro are posited. Prosodic phrasal boundaries can play a fundamental role in restricting the possible number of (preferred) interpretations for syntactically ambiguous strings in Korean. Moreover, prosodic disambiguation can be directly tied to prosodic phrasing, because of the distinctly phrasal nature of Korean intonation. In contrast to English and Japanese, Korean lacks lexically-associated