Despite growing interest from historians in the built environment, the use of architecture as evidence remains remarkably under-theorized. Where this issue has been discussed, the interpretation of buildings has often been likened to the process of reading, in which architecture can be understood by analogy to language: either as a code capable of use in communicating the architect’s intentions or more literally as a spoken or written language in its own right. After a historiographical survey, this essay, by contrast, proposes (...) that the appropriate metaphor is one of translation. More particularly, it draws on the work of Mikhail Bakhtin to suggest that architecture—and the interpretation of architecture—comprises a series of transpositions. As a building is planned, built, inhabited, and interpreted, so its meaning changes. The underlying logic of each medium shapes the way in which its message is created and understood. This suggests that the proper role of the historian is to trace these transpositions. Buildings, then, can be used as a historical source, but only if the historian takes account of the particular problems that they present. In short, architecture should not be studied for its meaning, but for its meanings. As historians we are always translating architecture: not reading its message, but exploring its multiple transpositions. (shrink)