Reflections on the practice of reading, GTM vs ComAp, mental translation, and other things

This post is a response to a twitter thread by David Schroder, responding to another twitter thread by biblingo, responding to another twitter thread by me. Twitter threads three-deep. Anyway, David is a current student of mine, and biblingo are friends-in-languages, so this is written in a spirit of dialogue and reflection.
For full disclosure, I first learnt Greek and Latin (and Hebrew) via GTM. I was a master of explicit grammar, and rote-memorised vocab relentlessly (down to about 2 or 3 occurrences for the New Testament). And let me say that GTM ‘works’. If you hear me regularly dunking on GTM, that might surprise you, but GTM works. But it doesn’t work well or efficiently.
See, I got to a point of reading the NT fairly well, because I poured hours and hours into it. I really doubt that many other people in my seminary cohort reached that point, not for years later. So, let’s just accept that I’m an outlier. This isn’t meant to be boasting, I don’t know that I’m “gifted” in languages, I’m just a smart person who put inordinate amounts of time into Greek. I would regularly leave Greek grammar exams after 15 mins with everything complete.
But let’s consider: (a) what does it mean for GTM to ‘work’, (b) why does GTM ‘work’, and (c) is that a good reason to uphold it as a general practice.
I take it that one of the main intended outcomes of GTM is that students can read ancient texts with comprehension. When I say, “read”, I mean “mentally process the words on the page, without resorting to translation in order to understand, in a process analogous if not quite identical to native-language reading”. I don’t mean “producing a translation”.
That’s not the only imagined-intended outcome of GTM, many would add in a lot of things like grammatical analysis, exegetical process, etc.. But I take it that many GTM practitioners believe that with enough translation-to-understand, the process becomes faster, more automated, and eventually becomes internalised, hence ‘reading’.
I don’t think GTM works like that, when it works. The field of SLA seems fairly unified in saying that acquisition comes (primarily) via input, and input is exposure to comprehensible messages in the target language. That’s not that controversial, even if certain elements of Krashen, VanPatten, etc., like to be pushed-back on by SLA researchers. You don’t need to join a Krashen bandwagon to back that input is the primary driver of acquisition.
So what GTM does do, is it makes a text comprehensible, which means that the Greek is functioning as input. The ratio of work to comprehensible input is incredibly high though, which means it’s a very inefficient way of getting input. Say you spend half an hour translating Eph 1:3-14, which is 217 words. As opposed to reading or listening to readily comprehensible material at your level, at a very very conservative 70 wpm (I’m basing this on a reasonable reading speed in an L2, and/or very slow and controlled speech). That’s a difference of input by a factor of 10.
One of the main “slowers” of reading texts (apart from the paucity of suitable material for learners) is just unknown vocab. This is why reader’s editions, student texts, etc., are incredibly useful. It is helpful to be able to glance to the bottom of the page or across to a facing text, get a quick L1 gloss, and then return to the text. You don’t need to have L1 immersion. Creating something like LLPSI is an incredibly difficult task. Whatever keeps learners reading in the L2 is worth doing, and you can sacrifice some ‘purity’ about keeping it in the L2 to do so.
I mentioned above that I brute-forced my way to a large NT Greek vocabulary. People like me who (now) suggest that flashcards aren’t that great, that needs some qualification. Spaced Repetition memory work does work. But again, we need to talk about what ‘working’ means. SRS works really well at memorising discrete chunks of information and holding them in your memory. For vocabulary work that is really quite limited to recognising an L2 word and being able to retrieve a couple of glosses in your L1. That does work, in the sense that you can learn a lot of words this way, but vocabulary knowledge is not a binary of know/don’t know, and it’s not as simple as “I know this word, it means X gloss in my L1”. Vocabulary knowledge is multifaceted, and its networked. It’s knowing, I dare say feeling the vibe of a word in the target language, as part of a web of meanings across the rest of the language-system. That comes via building up a long-term extensive mental representation of the language.
So, yes, ANKI or whatever is beneficial. And honestly, acquiring vocabulary from context in reading is better, but it’s also harder, because you generally have to be reading texts in which you understand 95% or more (98% is good) of the words already, which we simply don’t have enough texts designed like that.
I’m going to turn to the question of mental translation for a moment. I’ve written about most of the above in various posts here, but I’m not sure I’ve ever talked about this issue specifically. I want to say it’s okay if your brain can’t help translate on the fly while you’re reading. In fact, to some extent I think this is unavoidable. What I tell students is to neither encourage nor fight this. We want to encourage our brain to think in the target language. That is entirely doable, especially if we start by communicating in the language, building up prompt/response, q/a, dialogue in which L2 is given, and L2 is given back. For me it’s a bit like a ‘switch’, which I’ve written about here
So we want to encourage the brain to operate in the L2, but it’s probably going to keep on giving you some running translation in the background. That’s fine. Don’t fight it. Fighting it is actually encouraging it, like not thinking about purple elephants. I don’t have any particular theory or research for this point, this one’s entirely personal anecdata. But give it a go, don’t worry about it, don’t fight it, don’t encourage it, just keep bringing the mind back to operating in the language.
To return to whether GTM is worth it. My particular perspective is that a proper communicative approach can achieve genuine acquisition, in a shorter time period overall, with more fun, less toil, greater inclusion, and more people along for the ride. Nor do I think, if it’s thoughtfully done, do you need to ‘sacrifice’ what “Grammar” promises. Linguistic analysis, complex discussions, etc., can still be done. I would delay them, do them separately in English, or boldly train students to do them in the L2, but they can still have a place, and I think they would have a richer place among students who have developed an intuitive sense of, e.g., the feel of ἐφ’ ᾧ, not evaluating a list of options they’ve found in a commentary.