This dissertation is an exploration of counterdirectionality, a semantic notion encompassing various sorts of reversals, involving either movement along a path, or---more abstractly---the restoring of an object or a state of affairs to a prior condition. Counterdirectionality is a relationship between an asserted event and a presupposed event in a strict temporal ordering. Across languages, it is frequently expressed by presuppositional adverbs that mean BACK (as in English Ali flew back from New York, Bina hugged Ali back, The door swung back open). The distribution of BACK-adverbs tends to overlap in a systematic way with that of repetitive/restitutive adverbs, i.e. those that mean AGAIN (as in English swung back open/swung open again), a puzzling fact given that counterdirectional adverbs have no repetitive component. This work ties into an older literature in which restitutive readings have been used as tools to probe syntactic structure very low in the VP. I demonstrate that Hindi-Urdu allows some adverbs to be structurally low, modifying very small sub-structures within the VP, as evidenced by highly constrained word order in restitutive readings, as opposed to the freedom of scrambling available in other circumstances. I show that a basic semantics for counterdirectionality can derive the effect of restitution independent of repetition. The core contribution of this dissertation is a semantic treatment of BACK-adverbs that moves beyond the basic task of capturing restitutive readings. I show that two core readings of BACK bear a family resemblance that is explained by a view of scalar change that unifies paths and scales: restitutive readings (restoring a state) and reversed path readings thus convey the same kind of meaning, with their slightly different flavours due to the specific scale involved in each case. Once we adopt this view, we no longer need to posit counterdirectionality or reversal as a semantic primitive; rather, the presupposition of BACK is derived correctly in each case by copying some (and critically not all) of the content of the assertion. My proposal radically simplifies the semantics of counterdirectionality, reducing it to concrete components already present in the assertion, and linking specificities (verbal selection, word order) of the various readings to independently available syntactic and semantic processes. I build my proposal on a detailed case study of vaapas 'back' in Hindi-Urdu, which exhibits a range of readings shared by BACK-adverbs in other languages. Finally, the dissertation contributes to the crosslinguistic study of adverbs and focus-sensitive presupposition triggers through the study of vaapas 'back', which is both: it can create different licensing conditions for a sentence depending on which constituent it associates with. This focus-association has word order consequences: I show that there are multiple possible surface positions available to focus-bearing nominals between the adverb and the verb, complicating the established FocP account of positional focus across South Asian languages.