A Quantitative Study on the English Headline Grammar: Focused on the NYT Headlines

Abstract

English headlines are said to constitute a special type of language, often called Headlinese , that has its own grammar. There are some, mostly fragmentary and sketchy, generalizations made in the literature on Headlinese, among which the following two are most salient and frequently mentioned: (i) Semantically light elements such as articles and copula/auxiliary be are generally/often deleted and (ii) tenses are simplified to the present tense. This paper examines how much truth there is in the two (and some other) generalizations by comparing an actual sample corpus drawn from the New York Times headlines and the British National Corpus (BNC), one of the most authentic English corpora. The major findings made in this paper include the following. First, the two corpora display a clear contrast in the frequency of the definite article, but not in the frequency of indefinite articles. Second, copula/ auxiliary be is more often retained than deleted, contra the general belief. Third, past tense does exist in Headlinese and must be used in certain cases, again contra the general belief. Fourth, Headlinese cannot exclusively consist of lexical categories, although an array of headline properties apparently conspires to indicate so.