The Good Enough Approach and Lingering Effects of Error Correction in Antonymic Pairs: Evidence from Eye Movements
Author
Summary, in English
This study is concerned with how the reading process reacts to detection of errors in discourse, and how this differs from cases where an error is missed or if no error is present. It sets out to investigate these questions using eye-tracking methodology. In order to determine the influence of detection of errors during the reading process, native English speakers were asked to read short news items containing antonymic incongruities in an eye-tracking experiment.
Experimental items were coded as discovered or not discovered based on answers given during the experiment and post-experiment interviews. A significant decrease was found in first fixation durations on discovered incongruities compared to the control condition, suggesting that language processing is instantaneously interrupted by detection of an error, in order to revise the representation of the text.
No difference was found in first fixation duration for undetected incongruities compared to the control condition. This supports the Good Enough approach to language comprehension. It argues that the heuristics providing an analysis of what is expected from the text enable the reader to dismiss an error before it undergoes a full semantic analysis. However, a significant increase in the total number of fixations on the word immediately prior to an undiscovered incongruity was found, as well as a marginally significant increase on the incongruity itself. This suggests that while the parser is capable of initially compensating for textual errors during reading, some cost is still accrued, evidenced in late measures – a finding that mirrors the garden path linger.
Experimental items were coded as discovered or not discovered based on answers given during the experiment and post-experiment interviews. A significant decrease was found in first fixation durations on discovered incongruities compared to the control condition, suggesting that language processing is instantaneously interrupted by detection of an error, in order to revise the representation of the text.
No difference was found in first fixation duration for undetected incongruities compared to the control condition. This supports the Good Enough approach to language comprehension. It argues that the heuristics providing an analysis of what is expected from the text enable the reader to dismiss an error before it undergoes a full semantic analysis. However, a significant increase in the total number of fixations on the word immediately prior to an undiscovered incongruity was found, as well as a marginally significant increase on the incongruity itself. This suggests that while the parser is capable of initially compensating for textual errors during reading, some cost is still accrued, evidenced in late measures – a finding that mirrors the garden path linger.
Department/s
- Master's Programme: Language and Linguistics
Publishing year
2013
Language
English
Document type
Student publication for Master's degree (two years)
Topic
- Languages and Literatures
Keywords
- garden path linger
- eye-tracking
- antonyms
- Good Enough approach
- inattentional blindness
- shallow processing
Supervisor
- Carita Paradis
- Marcus Nyström