This paper aims to validate the argument that SpeechActPhrase shell structure is needed to explain some constructions with modal expressions in English and Korean. Specifically, some epistemic modal constructions in English and –keyss-, -ullay, -la, -ca, and -ma constructions in Korean, with their speaker/hearer oriented meaning, should be distinguished from root modal constructions in their structure. This paper also shows that simple syntactic hierarchy does not fully explains the scope relations between modal constituents and other tense/aspect constituents.