This paper aims to examine hedging expressions expressed by epistemic modal auxiliaries and to analyze their grammatical function, intrinsic meaning and scope in English. Meaning difference in sentences merged with an epistemic modal auxiliary is revealed by the scope parameter between the unmarked and marked interpretation of it. At the C-I interface, the unmarked use of an epistemic modal auxiliary doesn't contribute to the truth-conditional meaning, whereas the marked use of it sometimes contributes to the truthconditional meaning. An epistemic modal auxiliary moves to C from T at the C-I interface, then it functions subjectively as a discourse-related information marker connoting an illocutionary force feature there. But when it functions objectively at T without movement, an epistemic modal auxiliary links thematic relation to its subject. In order to confirm the scope of epistemic modals, 20 native speakers checked the grammaticality of sentences that containing both an epistemic modal auxiliary and a quantifier concurrently. The result shows that all the native speakers interpret the epistemic modal auxiliary as denoting wide scope, but they don’t agree one another on the interpretation of it as denoting narrow scope.

This paper examines the change of frequency of the core modal auxiliaries will/would, shall/should, can/could, may/might, and must in the Corpus of Historical American English (COHA) covering 1810 to 2009. The most common modals used in this period were would (average 2,827.69 per mil), followed by will (2,262.94), can (1,783.77), could (1,782.83), may (1,138.85), should (1,081.55), must (1,019.18), might (729.83) and shall (609.54). The top four modals accounts for 65.4%. The general trend is a significant decline in use of core modals with time: In a simple comparison between 1810s and 2000s, shall reveals the most remarkable decline (-97.23%), then follows must (-78.34%), may (-77.76%), will (-65.48%), should (-59.21%), can (-29.53%), might (-27.17%), would (-17.13%), though can and would shows repetition of rises and falls, and could shows a rise in frequency. This paper shows that the decline of core modals goes along with the increase of such lexical verbs as semi-auxiliaries and main verbs, and argues for grammaticalization and democratization as the determinants of this change: The increase of lexical verbs is due to a 'renewal' in a grammaticalization process and the notable loss of must and shall is related to the sociocultural change of democratization.