Modal verbs through the years: Computational linguistics perspective

Abstract

This diachronic study which implements computational linguistics approach investigates the linguistics events of modal verbs in English song lyrics across time. This study aims to identify the frequency distributions of modal verbs, and to analyze the language uses of modal verbs in the Diachronic Corpus of English Song Lyrics. Quantitative and qualitative methods of analysis were employed. First, quantitative data consisting of frequency distributions of modal verbs and collocations were produced via LancsBox. Collocational patterns are illustrated through collocation graph created via Lancsbox. Frequency distributions of modal verbs were compared to general reference corpus namely the Corpus of Contemporary American English, to generalize the findings from this study as representative of English language. The numerical data were submitted for four statistical tests of significance namely Chi-square, Mutual Information, Log-likelihood, and t-score. Second, qualitative data are consisting of corpus annotations. Corpus annotation was conducted via CLAWS for assigning part-of-speech C7 tagset to identify modal verbs. Semantic categories of modal verbs were identified via UCREL Semantic Analysis System. Findings uncovered the significantly high frequencies and uprising trend of semi modals verbs want to and wanna as compared to core modal verbs in English song lyrics from l960s towards l990s. These two semi modals reflect social actions, high preference for personal pronouns I and you, and carry meanings of obligation and necessity similar to core modal verb should. The aforementioned characteristics characterized English song lyrics as a spoken-like genre.