A Corpus-Based Study of must, should, and have to

Abstract

Must, should, and have to are different in many respects while they are nearly synonymous in meaning in English. This study aims to review the meaning and usage of must, should, and have to described in the previous studies, conduct corpus-based research and discuss the corpus findings focusing on British English. For this purpose, British National Corpus has been searched including spoken and written corpora and THINNING by random selection function has been performed. The findings are as follows: (1) Should has the highest frequencies in spoken and written texts as a whole and must is in the middle; (2) Should accounts for almost half in written and in spoken English. Have to in spoken categories occurs about 2 times more commonly than must; (3) Regarding the frequency distribution of them by four major registers, conversation has the highest frequencies of have to, followed by should, and the lowest of must; (4) Regarding the meaning, unlike must and have to, the meaning of suggesting uniquely belongs to should only; (5) Must occurring with be or have is associated with logical/epistemic necessity. Epistemic meaning involvement can be a factor which might influence higher collocate ranking of be and have in the list of should.