Keyboard entries by users are tracked and filtered to reduce ambiguity in translating the entries to characters and predicting text based on the entries in a multiple keyboard environment. Key tables such as an equivalence table mapping characters to a base character value, a digit mapping table mapping characters to a digit between 0 and 9 are used to determine possible matches in text prediction. A letter accumulator UI is employed along with a number accumulator UI to provide users feedback on tracked letters.