Poster D29, Friday, August 17, 4:45 – 6:30 pm, Room 2000AB

EEG Evidence for Different Syntactic Expectations in Parsing Chinese Subject and Object-relative Clauses

Jiayi Lu1, Matthew Walenski1, Cynthia K. Thompson1;1Northwestern University

In English and other languages with head-initial relative clauses, where an extracted noun phrase precedes the relative clause, object-relative clauses incur higher processing costs than subject-relative clauses. In Chinese, relative clauses precede the extracted noun; thus Chinese subject-relative clauses begin with a verb, similar to pro-drop sentences like imperatives, whereas object-relative clauses begin with a noun phrase, similar to SVO sentences. Mixed results for the relative difficulty of these structures are found; some studies find increased processing costs for subject-relatives and others increased costs for object-relatives. We used event-related potentials (ERPs) to investigate whether expectations for competing structures (e.g., pro-drop for subject-relatives, SVO for object-relatives) may contribute to increased processing costs at different points for subject- and object-relative clauses in Chinese. We recorded EEG from twenty native Chinese speaking college-age participants. We examined four sentence types (n=38 each; 76 ungrammatical relative clause sentences were included but not discussed here): Subject-relative clauses (Xiaoming shuo: [[zhui mao de] daheigou dique zhang-zhe chang wei ba]; “Xiaoming says: The black dog that is chasing the cat indeed has a long tail”), Pro-drop imperatives (Xiaoming shuo: [zhui mao qu ba]; “Xiaoming says: Let’s go chase the cat”), Object-relative clauses (Xiaoming shuo: [[mao zhui de] daheigou dique zhang-zhe chang wei ba]; “Xiaoming says: The black dog that the cat is chasing indeed has a long tail”) and Simple SVO sentences (Xiaoming shuo: [mao zhui-zhe daheigou]; “Xiaoming says: The cat is chasing the black dog”). Sentences were visually presented word-by-word with an 800ms ISI; participants made an end-of-sentence acceptability judgment after each sentence. ERPs were time-locked to the onset of the first word of the relative clause (bolded), with a 100ms baseline and 2400ms epoch. At the initial word (zhui vs. mao), subject-relative clauses (1) evinced a larger P600 component (600-800ms) than object-relative clauses (2). At the relativizer position (de), there were no significant differences between subject- and object-relative clauses, though both had anterior negative (LAN) components (300-500ms) relative to the aspect marker (zhe) in the SVO sentences (4). Finally, a P600 component (600-800ms) was observed at the imperative marker (qu) in the imperative sentences (2) compared to the relativizer (de) in the subject-relative sentences (1). The P600 results indicate that the initial nouns in subject-relative clauses are more difficult to process, and likely less expected than initial verbs in object-relative clauses. However, compared to another verb-initial structure (imperatives), subject-relative clauses are the more expected structure. Thus an initial verb may trigger an expectation for a subject-relative clause structure. In addition, the similar LAN components at the relative clause marker suggest similar structure-building costs for subject- and object-relative clauses, despite the initial difference in structural expectations.

Topic Area: Grammar: Syntax