Institutio Oratoria
Quintilian
Quintilian. Institutio Oratoria, Volume 1-4. Butler, Harold Edgeworth, translator. Cambridge, Mass; London: Harvard University Press, William Heinemann Ltd., 1920-1922.
For analogy was not sent down from heaven at the creation of mankind to frame the rules of language, but was discovered after they began to speak and to note the terminations of words used in speech. It is therefore based not on reason but on example, nor is it a law of language, but rather a practice which is observed, being in fact the offspring of usage.
Some scholars, however, are so perverse and obstinate in their passion for analogy, that they say audaciter in preference to audacter, the form preferred by all orators, and emicavit for emicuit, and conire for coire. We may permit them to say audivisse, scivisse, tribunale and faciliter, nor will we deprive them of frugalis as an alternative for frugi:
for from what else can frugalitas be formed? They may also be allowed to point out that phrases such as centum milia nummum and fidem deum [*](i.e. nummum and deum should, strictly speaking, be accus. singular. ) involve a
Augustus again in his letters to Gaius Caesar corrects him for preferring calidus to caldus, not on the ground that the former is not Latin, but because it is unpleasing and as he himself puts it in Greek περίεργον (affected).
Some hold that this is just a question of ὀρθοέπεια or correctness of speech, a subject to which I am far from being indifferent. For what can be more necessary than that we should speak correctly? Nay, I even think that, as far as possible, we should cling to correct forms and resist all tendencies to change. But to attempt to retain forms long obsolete and extinct is sheer impertinence and ostentatious pedantry.