Research methods in psycholinguistics: developing a tool for the automatic generation of pseudo-verbs

Abstract

The use of pseudo-words has benefitted psycholinguistic studies dedicated to the different aspects of the acquisition and processing of language, including reading and writing. However, in most of the studies that make use of pseudowords, the combination of graphemes in a string is often based on the intuition of researchers, who generally do not take the frequency of structural occurrences of the language into consideration. The present study aims at contributing to the development of procedures to aid research which makes use of pseudowords. In this paper, we present a computational tool developed for the generation of pseudoverbs. The process of development of this tool involved the analysis of the structural patterns of the 500 most frequent verbs in Brazilian Portuguese. This analysis was carried out with the use of NLP techniques and data mining in order to collect and extract patterns. The present study adopted an unsupervised machine learning approach since it used an algorithm of clustering for the automatic generation of pseudo-verbs.