| Popis: |
In this project, we study whether readers use non-lexical cues to infer discourse relations, by examining whether discourse relation interpretations and expectations for discourse relations are dependent on the clause structure. More specifically, we examine discourse relations with and without a gerund free adjuncts (henceforth: GFA). GFA's are subordinate clauses that start with a present participle and do not contain a subordinating conjunction, as in Walking to school, the girl sang a song. Previous studies have mostly focused on connectives as cues that readers use to interpret discourse relations and base their expectations for discourse relations on. However, many discourse relations are implicit. They have been argued to be signalled by different cues (Das & Taboada, 2018), as certain linguistic features correlate more strongly with some discourse relations than with others (Asr & Demberg, 2015; Webber, 2013) However, little is known about how readers adopt these cues in processing discourse relations, as these correlates are less specific than discourse connectives, and primarily have a different function. A few studies have suggested that negation and tense influence the processing of discourse relations (Crible 2021, Grisot & Blochowiak, 2017), but no study has looked at the role of structural cues. The present research examines whether readers use clause structure as a cue for discourse relations. A corpus investigation has shown that discourse relations containing a GFA are often cause-consequence or synchronous relations. Readers were shown to expect result and synchronous relations more after prompts containing a GFA than a full matrix clause in a continuation study. However, in an interpretation study, no difference was found. In this study, we extend the previous studies, by examining how clause structure influence readers' processing of discourse relations during reading. We will compare reading times on the same result continuation after a GFA to those after an FMC. Since a potential slow-down after a FMC might also be caused by the full stop, we will also include a control condition. This condition, in which the full matrix clause is connected with the underspecified connective, and, contains no punctuation at all. If readers have a stronger expectation for result relations after a GFA, they are expected to process result continuations faster than after a FMC. This will reveal which information readers use in processing discourse relations online. |
