There has always been a difference in diction between speech and literature. At Rome, Cicero and Caesar, and of course the poets, did not, in writing, use quite the language of the people. All the words of daily speech were not taken into the literary or classical vocabulary, which had often quite other words of its own. Moreover the writers, in forming their prose and verse and constructing their compositions, were affected deeply by their study of Greek literature.[227] If Caesar, Cicero, Virgil, and their friends spoke differently from the Roman shopkeepers, there was a still greater difference between their writings and the parlance of the town. No one need be told that it was the spoken, and not the classical Latin, which in Italy, Spain, Provence, and Northern France developed into Italian, Spanish, Provençal, and French. On the other hand, the descent of written mediaeval Latin from the classical diction or the popular speech, or both, is not so clear, or at least not so simple. It cannot be said that mediaeval Latin came straight from the classical; and manifestly it cannot have sprung from the popular spoken Latin, like the Romance tongues, without other influence or admixture; because then, instead of remaining Latin, it would have become Romance; which it did not. Evidently mediaeval Latin, the literary and to some extent the spoken medium of educated men in the Middle Ages, must have carried classic strains, or have kept itself Latin by the study of Latin grammar and a conscious adherence to a veritable, if not classical, Latin diction. The mediaeval reading of the Classics, and the earnest and constant study of Latin grammar spoken of in the previous chapter, were the chief means by which mediaeval Latin maintained its Latinity. Nevertheless, while it kept the word forms and inflections of classical Latin, with most of the classical vocabulary, it also took up an indefinite supplement of words from the spoken Latin of the late imperial or patristic period.