Related Experiment Videos
The information conveyed by words in sentences
1Department of Cognitive Science, Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218, USA. hale@cogsci.jhu.edu
Journal of Psycholinguistic Research
|April 15, 2003
Summary
This study introduces a method to quantify information in sentences using probabilistic grammars and parsing. Calculating information per word may predict reading times and cognitive load.
Area of Science:
- Computational Linguistics
- Cognitive Science
- Information Theory
Background:
- Understanding sentence information is key to cognitive load.
- Probabilistic grammars model language structure.
- Parsing algorithms analyze sentence structure.
Purpose of the Study:
- To develop a method for quantifying information conveyed by words in a sentence.
- To apply information theory to linguistic structures.
- To explore the relationship between information content and cognitive processing.
Main Methods:
- Applying Grenander's work to top-down parser states.
- Calculating uncertainty due to structural ambiguity at each sentence point.
- Quantifying word-by-word information conveyed by subtracting successive uncertainty values.
Main Results:
- A method to calculate information conveyed per word was developed.
- Information content was calculated for several small probabilistic grammars.
- The study suggests information per word correlates with cognitive load measures.
Conclusions:
- The proposed method quantifies linguistic information.
- Information conveyed per word may be a determinant of reading times.
- This approach offers insights into cognitive load during language processing.