Theories of adverbial modification can be roughly distinguished into two sorts. One kind of theory takes logical form to follow surface grammatical form. Adverbs are treated as unanalyzable logical operators that turn a predicate or sentence into a different predicate or sentence respectively. And new rules of logic are stated for these operators. -/- A different kind of theory does not suppose that logical form must parallel surface grammatical form. It allows that logical form may have more to do with (...) deeper structures that might be studied in transformational grammar. Adverbs are treated as surface forms of the underlying predicates represented by corresponding adjectives and verbs. 'Slowly' is derived from 'slow'; 'intentionally' from 'intentional' or 'intend'; etc. And new rules of logic are avoided where they can be. -/- In this paper I attempt to state some of the advantages of the second sort of theory. My procedure will be this. First, I will try to say in outline what theories of logical form are. Then I will state five principles for evaluating such theories. Next, I will sketch the sorts of analyses acceptance of principles (1)-(5) leads to. In particular I will talk about adverbial phrases (e.g. locatives) that are best analyzed in terms of implicit references to events, relative modifiers (like 'large') which relate something to a comparison class, and 'that' clauses taken as names of propositions. By appealing to principles (1)-(5) I will defend these analyses against certain others, one that appeals to many logical operators, a second that treats all sentences as names of propositions, and a third that sees implicit reference to possible worlds in the language being analyzed. Finally, I will offer a pragmatic defense of my approach in terms of principles (1)-(5) as against a different approach that appeals to possible world semantics. (shrink)