I was responding to a post the other day about why Latin is still taught the way it is (grammar-translation) when modern languages are not. It's worth considering the chronic inertia in most classical language pedagogy. I will talk mostly about Latin, but what I say is equally applicable to Ancient Greek, and to some extent Biblical Hebrew.
Reason 1 - Historical short-sightedness
The grammar-translation method is often considered to be the 'traditional' method. But historical research into language learning pedagogy in Europe suggests that this method really only traces back 300 years or so, to the age when active Latin usage declined rapidly, and there was a shift to relieve students of the burden of Latin composition and active production when it was no longer required in academic contexts. This scaling back of Latin's practical usage meant that methods changed to focus on explicit grammar, since this was now the 'goal' of Latin instruction.
Realising that grammar-translation is actually a relative new-comer to pedagogical practices for classical languages is a key factor in dethroning it. For most of the past 2000 years, Latin was taught communicatively.
Reason 2 - How we were taught is, by default, how we teach
In the pressure of teaching, the easiest mode to default to is that by which we learnt. This is true of lecturing as it is of language instruction, and most teachers of classical languages both (a) were taught by G-T, (b) were among the "4 percent" who thrived on it. This notion of the 4% is quite useful - it's not an accurate number by the way. But the number of students who truly 'take' to grammar and enjoy grammatical analysis is relatively few. It's probably higher once you get to upper levels or college classes, since they are self-selecting. But these are the kinds of people who go on to become teachers of classical languages, and so they default to what they know, and then are baffled by why the 96% doesn't get it.
Reason 3 - False assumptions
The purposes of classical language instruction are different than modern languages. We can and ought to concede this. The latter aims at communicative proficiency for life experiences, whether work, social, leisure, or other. The former aims at competency at reading and interpreting ancient written texts.
The mistake is assuming that grammar-translation produces the goals of classical language education better than other approaches. My argument is and remains that communicative methods produce faster and better readers of ancient texts than grammar translation methods.
Furthermore, in the current academic climate, one of the goals of CL instruction is the ability to discourse about the text in grammatical detail. This is almost always done not in the target language, but in English or another modern language. It's contrasted with being able to order your latte at the non existent Koine Café.
But what if you felt equally comfortable ordering that Latte in Koine, as you did discussing the imperfective force of a certain verb also in Koine? There is no a priori reason grammatical analysis and discussion couldn't happen in the classical language itself.
Reason 4 - The myth of hard work
Most of those who succeed at classical languages enough to become teachers got there by hard work, with hard copy books, and looking things up in the lexicon, and brute force memorisation. It's a solitary form of torture that is character-producing and results in a shared "been there, suffered that" mentality. It also reflects in the attitude that looks at students and expects them to suffer the same - solitary hours spent in rote learning and analysis, to produce the same result.
But almost all our pedagogical research on Second Language Acquisition suggests that paradigm and vocabulary memorisation is woefully inefficient. That's why it's so painful and arduous - it requires so much more effort to achieve its outcomes.
It's not that we need lazy students, we just need to stop equating 'brute force' with 'good learning technique'. A communicative method would shift the focus to learning in community, would value in-context learning, and would devalue 'produce right form from memory' in favour of 'acquire right forms by repeated exposure'.
Reason 5 - The system reinforces it
Most classical language teachers got there by hard work, through a classical language curriculum that never intersected with modern languages, except maybe comps for higher degrees. They never learn Second Language Acquisition theory, never did courses in Language pedagogy, and are employed by departments that have little interest in those fields. They are given courses to teach that rely on grammar-translation methods, and are locked into them in their early teaching careers. The ability to turn from a system that reproduces the same problems, and create a genuine communicative alternative, is not open to most institutional teachers, *even if they wanted to*.