Human and computer estimations of Predictability of words in written language

Bruno Bianchi,Diego E Shalom,Gastón Bengolea Monzón,Luciana Ferrer,Juan E Kamienkowski,Diego Fernández Slezak

doi:10.1038/s41598-020-61353-z

Bruno Bianchi, Diego E Shalom + Show 4 more

Open Access

https://doi.org/10.1038/s41598-020-61353-z

Copy DOI

Abstract

When we read printed text, we are continuously predicting upcoming words to integrate information and guide future eye movements. Thus, the Predictability of a given word has become one of the most important variables when explaining human behaviour and information processing during reading. In parallel, the Natural Language Processing (NLP) field evolved by developing a wide variety of applications. Here, we show that using different word embeddings techniques (like Latent Semantic Analysis, Word2Vec, and FastText) and N-gram-based language models we were able to estimate how humans predict words (cloze-task Predictability) and how to better understand eye movements in long Spanish texts. Both types of models partially captured aspects of predictability. On the one hand, our N-gram model performed well when added as a replacement for the cloze-task Predictability of the fixated word. On the other hand, word embeddings were useful to mimic Predictability of the following word. Our study joins efforts from neurolinguistic and NLP fields to understand human information processing during reading to potentially improve NLP algorithms.

Highlights

When we read printed text, we are continuously predicting upcoming words to integrate information and guide future eye movements
We used different versions of four widely used algorithms (N-grams, Latent Semantic Analysis (LSA), Word2Vec and FastText) to build different computer-based Predictability models, and we used them as replacement for cloze-Predictability in a statistical analysis of gaze duration
We trained different computational models drawn from the Natural Language Processing (NLP) field in a larger corpus

Summary

Introduction

When we read printed text, we are continuously predicting upcoming words to integrate information and guide future eye movements. We show that using different word embeddings techniques (like Latent Semantic Analysis, Word2Vec, and FastText) and N-gram-based language models we were able to estimate how humans predict words (cloze-task Predictability) and how to better understand eye movements in long Spanish texts. Both types of models partially captured aspects of predictability. Having a computational estimation of cloze-Predictability would be a great step forward for neurolinguistics, from the methodological point of view, and to enable researchers to experiment with different contributions of the components of the computational model This would allow us to better understand the sources of the effect of this variable that is involved in the prediction of an active sampling of the visual world[1,2,3]. We used different versions of four widely used algorithms (N-grams, Latent Semantic Analysis (LSA), Word2Vec and FastText) to build different computer-based Predictability models (computer-Predictability), and we used them as replacement for cloze-Predictability in a statistical analysis of gaze duration

Objectives

Methods

Results

Conclusion