A method and apparatus provide for performing language translation by obtaining a selection of a language preference for presentation of text and/or speech on an electronic device. When data is received that includes content in a language other than the preferred language, a decision whether translation is required may be reached by analyzing the context of the language or metadata associated with the language, such as. HTML tags, or language data included in a particular protocol, e.g. HTTP headers. A selection may be made of billing level, determined at least in part by the languages involved, desired accuracy, latency and volume of translation. If necessary, the data may be routed to a translation host or service before final presentation.