The Written Word is Not the Spoken Word in Any Language
From Kenneth Jackson’s (1909–1991), Language and History in Early Britain, (Edinburgh: University Press, 1953):
In the Lowland Zone [of Britain] all education and writing, and such Roman literature as was composed, would be in Latin. But here a caution is necessary. As already remarked, Haverfield[1] made much of the fact that graffiti scratched by tilemakers in the process of their work were written in Latin…. He concluded from this that the urban lower classes spoke Latin and not British. This may or may not have been the case, but the evidence in question will not prove it. It should always be borne in mind that British was not a written language, and that the only language of writing was Latin; it would not occur to anyone to write in British, nor would they know how to do so. One tends to forget that to write down in an alphabet the sounds of a speech (even though it is one’s own) which one has never been taught to write is a very considerable intellectual feat. In Roman Britain those who had enough education to know the alphabet had enough to know some Latin, and those who had none did not write at all. We, who learn to write almost as soon as we learn to speak, are so much penetrated with the idea of writing our spoken language that we cannot easily dissociate the two; and are usually not aware of the fact that it is quite possible for one and the same man to speak a language which he cannot write and (when he as to) write a language which he cannot easily, or does not habitually, speak. (pp. 99–100)
[1] That is Francis Haverfield (1860–1919), The Romanization of Roman Britain (Oxford University Press, 1906).
UPDATE: Just saw this and thought it semi-relevant: