Social Media and Language Processing: How Facebook and Twitter Provide the Best Frequency Estimates for Studying Word Recognition
University of Milano-Bicocca
Frequency effects, Social media, Lexical decision, Text corpora
Herdağdelen, A. and Marelli, M. (2017). Social Media and Language Processing: How Facebook and Twitter Provide the Best Frequency Estimates for Studying Word Recognition. Cognitive Science, 41, 976–995.
- The higher the frequency of a word, the faster it will be processed.
- How can we estimate word frequency? We count occurrences in text corpora.
- Traditionally, statistics from books and newspapers are considered.
- We show that frequencies from social media (Facebook and Twitter) provide a better explanation for word recognition data.
We often use pronouns like it or they without explicitly mentioned antecedents. We asked whether the human processing system that resolves such indirect pronouns uses the immediate visual-sensory context in multimodal discourse. Our results showed that people had no difficulty understanding conceptually central referents, whether explicitly mentioned or not, whereas referents that were conceptually peripheral were much harder to understand when left implicit than when they had been mentioned before. Importantly, we showed that people could not recover this information from the visual environment. The results suggest that the semantic–conceptual relatedness of the potential referent with respect to the defining events and actors in the current discourse representation is a determining factor of how easy it is to establish the referential link. The visual environment is only integrated to the extent that it is relevant or acts as a fall-back when the referential search within the current discourse representation fails.