We introduce the task of predicting adverbial presupposition triggers such as\nalso and again. Solving such a task requires detecting recurring or similar\nevents in the discourse context, and has applications in natural language\ngeneration tasks such as summarization and dialogue systems. We create two new\ndatasets for the task, derived from the Penn Treebank and the Annotated English\nGigaword corpora, as well as a novel attention mechanism tailored to this task.\nOur attention mechanism augments a baseline recurrent neural network without\nthe need for additional trainable parameters, minimizing the added\ncomputational cost of our mechanism. We demonstrate that our model\nstatistically outperforms a number of baselines, including an LSTM-based\nlanguage model.\n