Large-scale evidence for logarithmic effects of word predictability on reading time

Area of Science:

• Cognitive Science
• Psycholinguistics
• Computational Linguistics

Background:

• Real-time language comprehension involves rapid decoding of complex meanings.
• Word predictability influences processing difficulty, but the underlying cognitive processes are debated.
• Existing theories propose linear (anticipatory processing) or logarithmic/superlogarithmic (probabilistic inference) effects.

Purpose of the Study:

• To investigate the cognitive processes underlying predictability effects in language comprehension.
• To determine the precise relationship between word predictability and processing demand.
• To differentiate between anticipatory processing and probabilistic inference models.

Main Methods:

• Analysis of six large-scale reading datasets.
• Estimation of next-word probabilities using diverse statistical language models.
• Modeling of reading times with advanced nonlinear regression techniques.

Main Results:

• A consistent logarithmic effect of word predictability on processing difficulty was observed across datasets.
• This finding supports models where processing demand scales logarithmically with predictability.
• Results challenge theories predicting a purely linear relationship.

Conclusions:

• Probabilistic inference is a crucial component of human language processing.
• The cognitive cost of language comprehension scales logarithmically with word predictability.
• This research provides strong empirical support for the role of inference in real-time language understanding.