What Weighs for Word Stress? Big Data Mining and Analyses of Phonotactic Distributions in Brazilian Portuguese

摘要

For about four decades, phonological theories have claimed that word stress assignment depends on the word's syllabic phonotactic complexity in relation to syllabic position. This study analyzes the phonotactic implications for word stress Brazilian Portuguese. After creating a phonotactic corpus and applying Random Forest modeling, phonotactic distributions for word stress were found to be bound to stress pattern and word length in number of syllables. To account for these observations, models of word naming must be extended with aspects of word stress.