After reading and watching a few podcast on the subject yesterday, what was interesting to find out, is I guess there are quite a bit of professional Latinists, who do not know Latin fluently as we would envision it. What I mean by this, is when we think of reading fluently, we think about what most of us do with English, i.e. pick up a book and very rarely have to look something up. What the podcast was saying, is that even many Latinists in the academic arena, do not have this type of fluency, but can only read Latin with a lexicon besides them. Nor, do they ever really pursue the same fluency we would have in English.
This is often said, but is misleading. Obviously there has been a catastrophic decline in the classics and so in the standards. But there are plenty of competent readers of Latin in academia, and academia gives you the tools to read Latin well. There are several factors that give the result that many can't:
- Many academics don't need to: many classicists are actually specialists in other aspects of the classical world, and their Latin doesn't need to be brilliant. It should be, but it isn't.
- Classical Latin texts are often extraordinarily hard. What we have of Classical Latin is often picked from the most literary and abstruse authors that a culture that loved a bit of literary abstruseness had. So Quintilian the Roman educator and orator, described Varro's writing as really obscure and hard to read for natives. Imagine that there were only twenty books surviving from the twentieth Century English, and one of them was James Joyce - and you had to try and read that. And then try and read some fragments from Spenser and Chaucer that survived on the back of an envelope. Often classicists that admit they can't read Latin well aren't saying 'I can't read Caesar/Cicero/Augustine' or whatever; they are saying 'when I read Lucretius I still need a dictionary'. That's not the same, and is pretty similar to a good second language speaker of English trying to read Joyce and Chaucer.
- There are many words that occur literally once in all classical Latin, despite a pretty large corpus - how are you supposed to learn those? Even the dictionaries can't necessarily be sure what they mean.
- Many of the best Latinists aren't classicists. If you have to read Medieval or Renaissance texts, there aren't any translations or good editions. You are on your own. These texts are often a bit simpler than the above, but still... So there are plenty good Latinists around.
And it has to be said that if your goal is to read theology, the task of learning Latin is much easier and really doesn't need any clever thought. Getting to the point where you can read Cicero like a novel is perfectly manageable, if hard work; getting to the point where you can read Aquinas is easier.
This baffled me a bit, because it doesn't seem fun at all to try to read something and have to stop every other word to look it up. Of course, I can see that in the beginning, but not from people who are considered professionals in Latin.
No, but obviously there's a sliding scale here - what about once a page? once a chapter?
This got me thinking, how could I use the tools we have to try to remedy this?
You really don't need to. Plenty of tools exist.
One of the things I thought of, which, I don't know if it will work, is utilizing AI, and a Latin word-list/definition (about 38,000 unique words) and try and produce a 20 vol. - 400pg-500pg (per vol.) science-fiction series in Latin, that utilizes all of those words at least 3 times each in the series. I am trying to stay away from overtly religious themes, but tried to base it off of a mixture between The Matrix, Alita Battle Angel, Interstellar, and Altered Carbon.
I know it is going to be real rough going at first, because I will need to look up so many words. But one of the people were saying, that while there are different methods to picking up the grammar rules, one of the main short-comings is restricted vocabulary intake, so that, while one may do good with the words in any given textbook; when they come to an actual Latin book, whose vocabulary has not been restricted, it is back to the slow-trod of Lexicon translation, rather than leisurely reading.
Now, I don't know if something like this will work.
Please please please don't! Firstly, there are solutions to learning this word list that are far more effective than reading those 20 volumes (we'll get to that) but secondly,
AI Latin is not good. Like, terrible. If you ask our basic models to produce ten independent sentences, you nearly always get substantive and problematic errors that would mislead anyone new to Latin. I do use AI to produce sentences with words in context for vocab learning, very regularly, and I have to weed out a lot of errors!
It is not even slightly ready to produce longform text. Some have tried to produce Latin readers with AI, and they have been
awful. Even finding all the errors is hard for a good Latinist, because many of them are so unexpected. All you'll do is wreck your sense of Latin grammar. I've tried, many times, and sure, it can write a short story in Latin... as long as you don't mind lots of awful mistakes, a puerile level of simplicity, and no sense of Latin style. It doesn't prepare you for real Latin at all, because the sentence structure is not like real Latin. The challenges of real Latin are often about clause construction and verb order and so on, and none of that is present in AI Latin, even if you ask for it.
Will the answer be different in five years? Maybe. But not next year.
Back to the bigger question: how best to actually learn Latin. Obviously Efraim is a capable language learner who knows multiple languages, and plenty of others here are too, and there are better Latinists on this board than me too.
But
reading is not the best way to learn vocabulary. It is necessary (in large quantities) to learn it
well, but it is not efficient as a learning method, especially for less common words. Learning via extensive reading only was fashionable in the 80s and remains so online, but research has shown it is far from optimal.
If you are interested in language learning, it's really worth reading some of Paul Nation's work, particularly his short (free) article 'What do you need to know to learn a foreign language?' and especially his table on how much reading is necessary to learn vocab by reading only (p55). In order to learn a word reliably through reading, you need to be exposed to it around 12 times. That's easy with common words - you need to read around 200,000 words to learn the most common 1000 words in English, for instance - 7 minutes a day for a year will do, if you read at 150 words per minute (not that learners can do that!. But to learn the top 9000 words, you need to read 3,000,000 words, 1 hour 40 minutes a day, due to decreasing frequency. The increase is of course not linear, so this is far worse for 10,11, and 12 thousand. And again, learners
can't read at this speed, at least not without a very carefully designed program of graded readers of the sort that basically only exists for English/Spanish/maybe French learners. There are a lot of new graded readers in Latin, and they are great, but it's nothing like this, and some day you are going to have to open Caesar and you aren't going to be going that fast. In actual reality you aren't getting to these levels without studying full time or for a number of years...
9000 words is enough to read a language without constant resort to the dictionary... but you probably need 12,000 to read comfortably. Most natives have 20,000+; literate natives with a good university education, 35,000+.
Back to the classicists who can't read Latin, this is what has happened: they have learned, but not read nearly enough, nor continued to do so over years. And it's hard to get enough exposure to many of the harder texts to absorb all their words - if a particular Latin word only occurs once in one text, do you have read that text 12+ times to learn that word???
Nation does very strongly encourage extensive reading, but to learn vocab well demonstrates it should be combined with vocab flashcards, which are several times better in terms of time investment. And he wrote before tools like
Anki leveraged
spaced repetition that uses
Ebbinghaus' forgetting curve. This makes flashcards more effective but takes 1/10th of the time. Add it all together and a deck of words ordered by frequency learned via Anki, combined with a rigorous reading, is orders of magnitude more effective than just reading.
You also want to target words that actually occur in what you want to read. Once you've got to 8000 words or so, most of the new words you encounter are domain specific - relevant to theology/history/military things/whatever. Using a general list of Latin terms isn't ideal beyond this point if you want to learn theology, as you'll be learning obscure vegetables and bits of military kit and Greek loanwords for Persian clothing and whatever, rather than the words you need.
How best to tackle vocabulary learning, then? Assuming you have been through a good textbook first (or at the same time) My method has been to use a
frequency list deck for Anki with the first 10,000 ish words, and then move to adding words drawn from particular books I'm working on so that I learn relevant words. For many common classical works, you can extract lists from
The Haverford Bridge and add them to Anki - including some Augustine etc. There are lists for Aquinas floating around too, which is helpful for technical theology in the 17th Century as well as medievals. Then it gets harder - you can add words manually as you go, when you look them up in the dictionary, meaning you only ever look them up once and actually learn them. Some day I'd love to put Muller's
Dictionary of Greek and Latin Theological Terms into Anki.
Obviously this is a waste of time if not combined with reading and/or listening in reasonable quantity as well, but for those of us who are busy, it's worth it.
