Lawrence Venuti, on page 13 of his book The Translator's Invisibility, gives the following definition of translation:
Translation is a process by which the chain of signifiers that constitutes the foreign text is replaced by a chain of signifiers in the translating language which the translator provides on the strength of an interpretation.
I like this definition, first, because it assumes the presence of meaning. After completing a master's degree in translation studies, I had been exposed to enough post-structuralism in translation studies and related fields to last a whole academic career. On my M.A. dissertation (a thesis is doctoral in the U.K.), one of my readers even remarked that my analysis of certain theories in relation to a specific text and translation was focused on the possibility of meaning. The remark was not at all negative and my approach worked just fine, if my final mark was any indicator. But the fact that the remark was made (implying that the possibility of meaning is not always assumed) left me bemused.
That said, I also like this definition, second, because it does not assume that meaning is always evident or singular. This is a very different matter from the existence of meaning. Any original has meaning, but any translation is dependent "on the strength of an interpretation" of the original.
I like this definition, third and finally, for the phrase "chain of signifiers." It sums up what could become a prolix attempt to explain what the different signifiers are (lexis, sentences, etc.) and how they relate (grammar, pragmatics, etc.).