Barbara Newman, in her LRB review of Medieval Grammar and Rhetoric: Language Arts and Literary Theory, AD 300-1475, edited by Rita Copeland and Ineke Sluiter, has a nice passage on the medieval idea of grammar:
Grammar, the foundation of the medieval curriculum, covered far more ground than moderns understand … under that rubric. It extended from memorizing phonemes through morphology, syntax, figures of speech, and the avoidance of solecisms all the way to ethics, poetics and literary criticism. … ‘Grammatica’ herself was personified as a matron, sometimes warm and nurturing but more often wielding the birch. In a bizarre but enduringly poplar textbook, The Marriage of Philology and Mercury, Martianus Capella portrayed her as a dental hygienist with sharp metal instruments: ‘by gentle rubbing she gradually cleaned dirty teeth and ailments of the tongue’ picked up through careless speech. By the late Middle Ages grammar had developed at its furthest reach into a theoretical discipline, called ‘modist’ or ‘speculative grammar’, that treated the modes of signification. Akin to such contemporary fields as linguistics, semiotics and the philosophy of language, it anticipates the theory of universal grammar, more controversially developed by Chomsky.
She goes on to compare medieval grammarians teaching Latin, nobody’s mother tongue, to “the fresh-faced college graduates who today fan out across the globe, teaching English as a Second Language in Tokyo, Budapest or Dubai. But … Latin as a textual language always took priority over speech, even though a more colloquial Latin remained the sole lingua franca by which Germans and Irishmen, Parisians and Bohemians could communicate. … When early grammarians felt the need to compare a Latin construction with examples from another language, they chose Greek (though they seldom really knew it) rather than their own vernaculars.” She has fun with Donatus, who “explained in his elementary textbook that Latin has four genders: masculine, feminine, neuter and common. His sole example of common gender is the noun sacerdos (priest), as every Catholic priest in training learned for a thousand years — but I have yet to find the form haec sacerdos ‘this (female) priest’ in a medieval Latin text.”
I’m not crazy about one bit, though. After talking about “the medieval conviction that language is a comprehensive, fully rational system, in which any part may be logically derived from the whole,” she goes on to say “Not so distant is the belief underlying genuine historical linguistics, a creation of the 19th century. Like its sibling disciplines of the same era (history of religions, comparative mythology, folklore studies), it posits that, to learn the deepest meaning of such phenomena, one must trace them back to their most ancient historical roots. In principio veritas.” This is rhetorical flimflam. The distance between premodern ideas of mystical coherence (as exemplified by Isidore‘s Etymologiae, the jumping-off point for her discussion of this subject) and modern historical linguistics, with its scientific demands for verifiable laws rooted in facts on the ground rather than ideas in the imagination or traditions repeated unquestioningly for millennia, is very great indeed, and it does no service to anyone but wild-eyed postmodernists to fudge it.