Parsing factivity and alternating factivity verbs

A complex sentence may indicate if the information it expresses is presupposed by the presence of an embedding verb. Factive verbs, such as “know,” are presuppositional lexical triggers that presuppose their complements, whereas nonfactive verbs, such as “believe,” do not. A special type of factivity verb, an alternating verb such as “remember,” is ambiguous between factive and nonfactive readings. Theories differ on the proposed structural complexity of factivity verbs, and factors influencing the alternating verb’s structural selection are poorly understood. Using eye tracking, this research investigates these questions by observing the verbs’ parsing profiles in relation to theoretical predictions. The first experiment compares the parsing of complement structures selected by the Chinese factive verb 知道 (zhīdào, “know”) and the nonfactive verb 相信 (xiāngxìn, “believe”), to ascertain which factivity structure takes longer to process. Using information gathered from the first, the second experiment investigates whether the ambiguous alternating verb 记得 (jìdé, “remember”) is initially parsed as a factive/nonfactive, testing predictions from minimal attachment versus the principle of parsimony. Our results indicate that the factive zhīdào incurs a lower processing cost than xiāngxìn, suggesting that factive structures are less complex, and the alternating jìdé, when presented without context, has a nonfactive processing profile due to the parser adopting a meaning-based parse. The implications for the theory of factivity verbs and structural selection are discussed, along with the role of meaning guiding structural selection in the case of ambiguous alternating verbs.