Some teaching reflections from 2020

This year I’ve taught more Greek than ever before, and more in ways that reflect my own pedagogical commitments. And so here’s some reflections on teaching Greek and Latin from my experience with The Patrologist courses, aka SeumasU

Athenaze

I decided to whole-heartedly adopt Athenaze as a text this year and use it as a basis for most of my intro Greek classes. This allowed me the freedom, ironically, to lean into the text. Previously I’d tried teaching from the Italian version alone, but the difficulty in getting hold of that text makes it somewhat prohibitive for learners.

The more time I spend with Athenaze, the more I grow to both like it, and be sorely aware of its faults. In my view, it’s still the best, currently available, continuous narrative text for learners. That’s my main reason for using it. I typically ignore all the grammar sections and let my students read them if they wish; my video course has grammar videos that go over those points too. But in class, we read the text, and as far as possible, use in-Greek question and answer to go over it, clarify meaning, etc., so that the text itself is understood. Comprehension is my main focus, and the text is both the means and a certain end, because it’s a good story for the most part.

It’s also far more text than you get anywhere else, except for maybe JACT’s Reading Greek. Which brings me to Athenaze’s great fault – too much vocab, too steep a slope. Because, under the hood, Athenaze is still a grammar-translation driven textbook. It still very much expects you to be translating the readings, memorising paradigms, and rendering things into English. So every chapter has too many new words, and too much new information.

That is, however, why I don’t instead use Reading Greek, because it’s actually worse at that. Much more grammar-as-a-hammer, pages of glossed vocabulary, and an even steeper difficulty curve.

All this has given me 3 main takeaways:

  • We need lots and lots more easy to read Greek aimed at learners. Novellas, novelliolas. Stand alone narratives. Connected narratives. Interconnected narratives.
  • The most useful thing for students to do on their own time, generally speaking, is to read things they can understand, including text they have already read. Corollary: the most useful thing they can do with my time is in-Greek communicative discussion, and making-comprehensible things they can’t do themselves
  • The more I can get away from the textbook, any textbook, to comprehensible Greek activities, while also ensuring that learners continue to develop reading skills, the better.

 

Latin

I mostly teach Latin from LLSPI, with a combination of some visual resources, and in-Latin discussion and paraphrase, and personalisation of q&a. This is one area where it’s clear just how good Ørberg’s LLPSI really is. The more I work with it, the more I see how carefully it has been constructed, and the progress I have seen with students in understanding Latin as Latin, as well as producing Latin output, has been tremendously encouraging.

Post-Beginner Conversational Classes

These were a bit of an experiment for me this year. I knew I wanted to run some classes that were, “okay, you learnt Latin/Greek, but you’ve never spoken it, so how do you start doing that”. And I also knew that I wanted to incorporate some of the standard gambit of CI type techniques – things like Picture Talk, Movie Talk, Storyasking/TPRS, as well as giving students a vocabulary to talk about things that they wouldn’t necessarily have ever encountered (e.g. modern things, grammar itself, daily activities).

I do think these courses could become a bit more structured from my end of them. That’s a little hard for me because I really do enjoy flying improv with just enough preparation to not fall entirely flat on my face. But that’s something I’ll be thinking about – clearer structure and progression in these kinds of classes.

Post-Beginner Reading Classes

Teaching μὲν beginners is super fun and super helpful for me. It’s like building and thickening and strengthening my own foundation as a speaker and developing my skills as a teacher.

Post-beginners δὲ are a different source of teaching joy. This year I taught a Greek patristics reading group for a term, and a 4-term sequence of theological Latin through the ages (right up to and including the Reformation era). It’s incredibly interesting to read texts and discuss them with students in their language. This is something I’m excited to develop further next year and beyond, with offerings in Biblical, Classical, and Post-Classical texts in both languages.

RPGing

My last ‘type’ of course has been running table-top RPGs in Latin. I ran two groups this year, and both were a great deal of fun. They came about because (a) I really love RPGs, (b) I had a great time playing D&D at the Australianae Rusticationes, (c) I think they have a huge potential at all levels for language acquisition. I’ll have a bit more to say about these in an upcoming paper at ASCS42, which I will also release at that time. There will definitely be more Latin RPGing coming up in 2021, and maybe Greek?

 

All in all, it’s been a big year of teaching for me, and I’ve loved 95% of it. Teaching teaches me a lot, and it keeps me committed to learning as well – I am always working on my own Latin and Greek behind the scenes, as well as self-professional-development as a teacher. Looking forward to a veritable smorgasbord of ancient language teaching in 2021.

 

Biblingo, a review

Biblingo is an app developed for learning the biblical languages (very specifically, New Testament Koine Greek, Biblical Hebrew), through a (semi-)communicative approach. I became aware of Biblingo some time ago, and followed some of their pre-launch promotional material, but more recently decided to test it out with a 10-day free trial.

Firstly, a little background. I studied Hebrew at Seminary for 3 years, doing quite well along a traditional grammar-translation track. I didn’t succeed very well in keeping Hebrew up, and so my Hebrew slowly atrophied. I did teach a course on the exegesis of Amos in Hebrew (in Mongolian), but my ability to read Hebrew now is very weak. I have off and on considered taking some communicative courses to get some Hebrew back in a more robust and active mode, but it is relatively low down my everyday priority list of languages.

You can sign up for a 10-day free trial of Biblingo at present, and that’s what I did. It gives you access to the four current ‘modules’ – Language Learning, Flashcards, Alphabet, and Bible Reading. Beyond that, one needs a subscription, for either or both languages.

The user interface is very pleasant, and generally easy to navigate. You are given some short pop-up intro videos to get you oriented, and if you don’t know the alphabet at all, there are a number of video lessons to get you acquainted with it. Those video lessons are high quality presentations, but there are no exercises to familiarise you with using the alphabet. You also have a range of choices for pronunciation in each language.

The core of the engine at present appears to be the Language Learning module. This starts off as all locked, and you need to do and complete lessons in sequence, with 3 levels, a total of 26 units across them, and 4 sub-lessons per unit. Each lesson consists of a sequence of Vocabulary, Grammar, and ‘Final Act’ or application. I can’t say if the format and organisation changes in the more advanced units, as I haven’t gotten there.

Vocabulary begins by offering you 6ish new vocabulary items, each presented with a short video or an image, with an audio track. You can click for an english language gloss, but I did not use that feature, preferring to associate image directly with word/phrase. This did leave me unsure about exactly what some verbs were portraying, but my presumption is that I’ll figure that out as I progress.

The images and videos are high quality, and tied to an imagined biblical world, that is you will see people in biblical settings, with biblical type clothing, doing biblical type things. There’s no telephones and where is the bibliothèque here.

Vocabulary proceeds through several stages: presentation, then passive knowledge (selecting the right answer from multiple choices), and ‘active’ knowledge – inputting the correct word(s) by either typing or using a word bank.

Generally, I found the typing frustrating. Partly this was some technical issues with my own Hebrew keyboard layout, but the app would benefit greatly from some kind of typing tutorial mini-app. Similar to, say, duolingo, it doesn’t penalise you for minor misspellings, such as accents, but offers a gentle reminder and a chance to ‘practice’ (i.e. retype), though the retype is not compulsory. I did notice that even when I had entered the Hebrew correctly, I was sometimes told to ‘watch my spelling’, perhaps because of a mismatch in coding between the Hebrew keyboard I was using and what the app expected. I had no problems when testing out the Greek side.

I’m not quite convinced that simply entering vocabulary by typing it out quite counts as ‘active knowledge’. It certainly helps spelling. But it still seems to me a rather limited treatment of active recall. The Hebrew/Image presentation is a good improvement over bilingual flashcards, but we are still mostly being presented with isolated vocabulary items.

Grammar in each lesson takes the vocabulary you have learnt, and puts it into sentences, with one or at most two new grammar elements per unit. Again you are prompted with a video or image, and with a symbolic presentation. The symbolic presentation uses symbols to prompt for things like definite articles, 1/2/3rd person pronouns, and the like, which would not be obvious from the videos alone. This is quite a novel feature and well executed. The movement through this section is very similar to vocabulary – presentation, ‘passive’ recognition (multiple choice with 4, then 8 options), and then active input.

The last section is usually quite short, and asks for active response to a slightly harder prompt, requiring some creative application of materials you have learnt. It is a way of testing that you’ve understood and can apply in a more adaptive way, the things you have been shown with more restricted vocabulary.

Overall the sequence of lessons is well done, it moves in very bit-sized pieces, and gives lots of opportunity for exposure and integration. I will say that a few features and perhaps critique are in order though.

The multiple choice responses could be improved on the UI level by offering keyboard numbered responses. The typed-response was frustrating in not recognising some correct responses because of presumably some Hebrew Unicode representation issues, and so I ended up using the wordbank more, which makes those exercises a little more ‘passive’ again. There’s no way to skip lessons, or even to skip to, e.g., the latter part of a lesson that you’ve already completed, you have to work through each one in sequence, and if you ‘retake’ a lesson, you need to go through it from the start.

The inability to skip/unlock was frustrating, partly because I wanted to check out some more advanced Greek lessons. I have a very solid knowledge of Greek, but the linear direction of the app prevented me using features that would be useful for me, without investing significant time.

Two other modules are present in the app. Flashcard Deck is a spaced-recognition presentation of vocabulary that either (a) you have tackled in the lessons, (b) some preset decks (fairly limited at present), or (c) custom decks (to which you can add words from the Bible Reading module). The flashcard deck presents the vocabulary in ‘sets’, and asks you to work through the same association>passive>active sequence as are presented in the Language Learning module. For that reason it was also, in my view, a little tedious. I couldn’t indicate I ‘strongly’ knew a word for instance, and so I had no choice but to see it over and over, nor vice versa. I would like to have seen a more fine-grained version of ‘how well you know a word’ in the vocabulary engine.

The last module, and the most recently released, is Bible Reading. This is really meant for more advanced learners, it seems, so for this review I’ll talk a bit more about the Greek version. Essentially, you have the chance to browse the Bible at Book, Chapter, Verse level, and at each level it indicates what percentage of the words you know. You can thus sort and look for parts that you know more vocabulary, and which therefore should be easier to read.

Then, when you select a book, each word (or morphemes in the Hebrew version) is colour coded based on whether you know 0, learning-some, learning-all, or know-all sense of the word. Each word can be clicked, and you are given a ‘dictionary pop-up’ which divides the word into various senses, divided by semantic domain (this seems to draw on Louw-Nida). Pictures are given where available, but otherwise English glosses. Each word also has parsing information.

Navigating the Greek version of this, it was fairly cumbersome to try and ‘fast-track’ what I already know. So I couldn’t easily bring the app up-to-speed with what I knew, without again investing significant time in teaching it.

Nonetheless, especially for someone working from scratch, this is a good feature. Though it could be improved – it still presents the biblical text, and only the biblical text, as its reading input. So your ‘input’ options in biblingo are either isolated sentences in the language learning module, or raw biblical books.

Pros

The biblingo interface is smooth, and the underlying principles are generally sound. The de-emphasis of explicit grammar, the modulisation and indeed granularization of learning chunks, and the focus on image/video material for direct association of vocab, are all great features. So, too, is the use of a symbolic code for sentence structures.

I’ve now done a week of Hebrew, 15mins or more a day, and I feel like I understand everything i’ve been presented and can respond to it clearly, directly, and in Hebrew. That’s what you want in language learning, and that’s what I’m getting on this app.

Cons

The structure of learning is very linear, and each sub-unit has to be worked through in sequence, and as a block. There remains something of an emphasis on passive recognition, even in ‘active’ exercises. Comprehensible Input is limited to isolated sentences, and there is no truly ‘communicative’ component to Biblingo – I am interpreting meaning at an atomised level, but only for the purpose of language learning, not for any other purpose. I am also not actively outputting meaning-based language, and there’s no negotiating of meaning between parties, it’s all one way in that sense.

The fine-grained nature of ‘know/don’t know’ is not adequately, or at least adequately transparently, presented to the learner, so it feels like overkill on some words, underkill on others, and knowledge also appears to be mostly tied to vocabulary knowledge. For biblingo to be more data-driven and user-responsive, it needs to find ways to track learners’ exposure, and comprehension, of other features of language (e.g. syntax).

I would also suggest that biblingo simply doesn’t provide enough, and enough variety, of comprehensible input. The gap between ‘isolated sentences’ and ‘biblical texts’, is enough to mean that the jump from one to the other is too far. Learners need extensive exposure to CI, and that means broad and wide and varied input. This has long been a problem for historical languages, and is only starting to be addressed for Latin. It’s an ongoing problem for Greek, which I try to solve for my own students in a variety of creative ways (mostly involving finding new easy things for them to read).

Overall

Despite these criticisms, I find biblingo engaging enough as an app, and I am learning some genuine Hebrew without explicit grammar (which I mostly try to ignore even when it pops up), and without translation. That’s a very significant difference to any other app/program/asynchronous set of materials out there (though there are some other options, yes), and for which the biblingo team ought to be applauded. And, work on biblingo (to all appearances) is ongoing, so it may yet improve and offer more and richer ways to learn.

To find out more about Biblingo, see their website.

 

Same conversation, same text – a technique

One of the things I have been reflecting on recently, both in reading a fair bit, but also working on the composition of LGPSI, is how useful it is, and would be, to take the WAYK concept of “same conversation” and apply it at the text level.

Same conversation is a technique where you deliberately repeat the same conversation, on a familiar, habitual, repeated topic, practice, habit, occurrence. What might that look like for texts?

Well, take capitulum primum of Lingua Latina, it’s basically a geography lesson, that to some extent leverages of people’s ability to recognise major countries and cities, and of the visual prop of a map. That’s smart (good one, Ørberg!). LGPSI starts with the same basis, though less effectively (because fluvius, oppidum, insula do not work nicely in Greek, eheu).

But what if you did other geography ‘lessons’ after this. You’ve built some basic vocabulary and structures around these things, so you can ‘repeat’ this lesson with a text about somewhere or somewhen else. So, a map of the Mediterranean, but now it’s the 10th century. Or, a map of America. Or a map of the 1960s. And you can start to make things more complicated. Add a geographical feature or two. Build more complex relationships between entities. And so on.

This is the advantage of thinking about ‘reading sideways, but up a little’, you build on the same conversation, to revisit the same language, but a little bit more complicated, with different content.

Or, to be honest, you could come back to the exact same conversation/text, but later. E.g. what if there was a second volume of (e.g.) Familia Romana that was the same storyline and content as the first, but the language was all ‘scaled-up’ to reflect the fact that you’d read the whole book before.

These are some of the ideas floating around and looking for implementation in the broader vision of my LGPSI project. Multiple storylines in multiple times, gives the opportunity to do lots of ‘same conversation’ with different content, and growing complexity.

Reading… sideways?

One of the things I’ve become increasingly aware of as a strategy for developing greater comprehension and fluency, is what I’ve taken to calling ‘reading sideways’. In particular, this has crystallised for me lately while reflecting on my reading through the Lingua Latina Per Se Illutrata materials.

Because ‘reading forwards’ gets more and more difficult. Working through Ørberg’s main volume, Familia Romana, Ørberg introduces a whole new set of grammatical forms in a chapter, say a full set of present passive indicatives. And he uses the in context, with nice alternation, and really shows the learner how they differ and how they function. It’s quite masterful to watch. But, you read a chapter, and then if you are pushing forwards all the time, well for some learners it gets to be a bit of a slog.

Now, I always advocate to people to ‘read backwards’, i.e. any time you get stuck, you just back up as far as you can, and start reading forwards again (I know ‘backwards’ isn’t quite the right word here). By reverting to much easier material, you build up your ongoing exposure to the whole mass of language, and you get some repeated exposure to things that are new.

But this only serves you so well, because you know the material, you get familiar with the content, your brain isn’t trying so hard to understand the messages themselves.

This is why reading sideways is so valuable. And the two supplements Colloquia Personarum and Miraglia’s Fabulae Syrae do this really well. They are keyed to the chapters of FR, containing only modicums of new words and virtually no new structures, and they repeat the new material from the relevant chapter, but they tell entirely different stories. So you’re fresh material, but with the same ‘language’. This is ‘sideways’ – you’re not ‘progressing’ by adding more language, you’re progressing by seeing the same things over and over but in new and varied forms with new and varied stories.

This is, by the way, one of the reasons why my big-picture vision for LGPSI is to end up with multiple storylines. You can be reading multiple stories, all with similar progressions in vocabulary and grammar, mutually reinforcing the language you’re learning, but getting it in new formulations. But that’s a long way off. For now, look out for opportunities not to be reading ‘harder’ material, but just different material at the same level you’re already reading it.

Re-conceiving the middle voice for Greek and Latin students (VI)

Well, I hope by this point, I’ve done a few things:

  1. Convinced you that Greek has an active v. middle voice system, in which the middle domain covers a range of different categories, all generally united by ‘subject affectedness’
  2. Convinced you that ‘deponency’ is not a useful concept for explaining verbs that do not appear with active morphology.
  3. Given you a reasonable account of most of those semantic categories, thereby giving an explicit account of why their ‘middleness’ makes sense.
  4. Shown that Latin deponents, by and large, represent historic middles from a PIE origin, and can be accounted for mostly under the same types of categories.
  5. Show that the ‘passive’ forms in Greek are also a category of middle-morphology.

But what does this mean for teaching and learning? I wrote these posts partly to force myself to work through some of the research, partly to make it clearer in my own mind, but also to try and bridge some of the gap between linguistics and general classical-language education. In my experience, a lot of this simply never makes it down to teachers, let alone students, of these languages.

For the grammar-translation context

If you teach in a relatively ‘traditional’ mode of grammar presentation, and translation as exercise, I think there are certain things you can do.

Firstly, change your terminology. Start referring to Greek as having an active vs. middle (or mediopassive) voice system. Explain that ‘middle’ means a range of categories that represent some form of ‘subject affectedness’, and that as your students encounter various types of middle verbs, you’ll point out what those categories are. Don’t try to present them as I have here! Rather, on a case by case approach, simply say, “here’s a new verb ἔρχομαι. It’s middle in its forms, because verbs indicating translational body motion are often middle.”

Greek students need to be met with the idea early that Greek voice is quite different to the Active-Passive system in English, and isn’t best understood with reference to English, but learning to accept it on its own terms. My mantra here, as elsewhere, is meaning comes first, then translation.

I think it’s worth adopting the language of media tantum (‘middle only’) for verbs that lack active counterparts. But I also think it’s worth adopting the language of middle-preferred or middle-primary for verbs that mostly and ‘by default’ appear in the middle, treating their active counterparts as the secondary/subsidiary form (e.g. causatives).

As students meet more middle-only and middle-preferred verbs, you can begin to systematise some of the semantic categories, and give more general explanations for why these verbs tend to show middle morphology as an expression of middle semantics.

It’s still worth preparing them/inoculating them, against the ‘deponent’ explanation, by giving a short historical rationale for why this language was used, and why you don’t. I often say, “middle in meaning active in translation”, to highlight that active translations in English reflect English voice and semantics, not Greek.

For Latin, I think a similar approach can be taken, except that there really is an active vs. passive voice system, and so when historic-middle verbs begin to appear in your classes, it’s worth stopping and giving a short historical explanation, “These are verbs that indicate subject affectedness and often changes-of-state, they have passive morphology because they were originally middle. etc..”

Communicative Language Teaching

In some ways, it’s actually far easier to teach these in a CLT approach. You simply introduce them without comment. You don’t need to explain why orior is an -r formation, or ἔρχομαι has a middle ending, any more than you need to stop and explain verb endings. You just introduce them in ways that are comprehensible, and it’s only if students begin asking questions that you need to pause and give, e.g. some pop-up grammar.

That could be as simple as, “oh, some verbs use these endings instead, because of what they mean.” The ongoing exposure to their usage in regular conversation, and readings, will make clear enough how they are used.

For students that persist, the same approach as any grammar-curious student – a short explanation on the spot, then follow up with a more detailed explanation outside the communicative context, or a written-up version (perhaps not quite the version I’ve done, though you can try!)

My own take-aways

For myself, reading through Kemmer, Allan, Aubrey, and bits and pieces elsewhere has helped crystallise my understanding of voice systems in both Latin and Greek. I have a much clearer understanding of Latin deponency, and a stronger articulation of Greek’s middle system, including how the ‘middle’ forms and the η/θη forms carve up the middle domain. From here on I’ll be using terminology such as mediopassive voice, middle-only, middle-primary, and explaining the selection of voice based on the semantics of the lexical items in question. And, in CLT contexts, I’ll be worrying less about “used-to-be-called-deponents” as being oddities, and treating them more as a normal and regular feature of the language.

 

A pdf of this whole series is available, slightly edited and formatted. You are free to distribute it under a CC-BY-SA 4.0 licence.

I’d be very happy to hear from you, if you have corrections, suggestions, additions, or just generally in response to this series of posts!

Re-conceiving the middle voice for Greek and Latin students (V)

So, about Latin then 

All through this series (one, two, three, four) I’ve been careful to include Latin examples of the various semantic categories covered. That’s because I’ve often wondered about Latin – does it really have deponent verbs? The answer is, it’s complicated.

It’s complicated by two factors: firstly, how we define ‘deponency’, and secondly how we understand the Latin middle-only verbs. However, by the traditional description of “passive verb with active meaning”, and “verbs that ‘set aside’ (deponere) their active forms”, no. That’s not true.

But if we mean simply ‘defective paradigms’ or ‘form-function mismatch’, that does appear to be sometimes the case in Latin, because Latin is not Greek. Latin does have an active vs. passive morphosyntactic opposition, not an active vs. middle. So the passive only verbs in Latin are more anomalous. For the most part, they appear to be verbs that were historically middle in an earlier stage of Latin’s development from PIE, and so one can usually see that categorisation at work, though for some verbs it remains elusive (see earlier discussions on loquor for instance). However, Latin deponents often take active morphology for participle and gerundive forms.

Late Latin, however, may be a different situation. There you see verbs switching from active to ‘deponent’ (at least, true deponency!) or vice versa moving from passive-only to active morphology. Generally however the trend is for deponents to become active-morphology verbs. This might be linked to the loss of synthetic passives in place of novel analytic forms.

At the end of today’s post I give a lengthier list of Latin deponents and the kind of middle semantic category I see them as.

Semi-deponents

“Semi-deponents” are the label often used for verbs that appear to have regular active morphology in one tense-form, but switch to a middle (or passive, in Latin) form in another tense system.

Much as I dislike the term ‘deponent’, I am going to keep using ‘semi-deponent’ at least for this post. There are two categories of semi-deponents, as I see it. The first are words where the paradigm is in fact suppletive. That is, the stem used in one tense is altogether different from the stem used in another. The second, then, are words that do not involve suppletive formations.

Suppletives: ἔρχομαι and ἦλθον

Probably the most well-known Greek example of suppletion is ἔρχομαι. This verb, as most students encounter it anyway, means ‘to come’ (though, a read through LSJ will make you a bit wiser on that score), and its aorist is supplied by ἦλθον, its future in Attic is supplied by present-forms of εἶμι. I’m only going to talk about the present v. aorist alternation here.

Now, the fact that ἔρχομαι is middle-only we referenced back in our second post. It’s a type of translational-body-motion middle, and ἔρχω is found only as a barbarism or as a point of discussion by grammarians. One may translate it regularly as an intransitive active in English, but that’s beside the point, its Greek semantics are middle.

Why then is ἦλθον morphologically active? I would suggest the reason is this: the semantics of that stem encode different features.

Now, let’s do some analogising to see how this works and how you might explain it, to yourself or to students.

Firstly, just because many verbs of translational bodily motion are middle, doesn’t mean that they have to be middle. βαίνω and βαδίζω involve the same idea, but they are active in morphology.

Secondly, in English we have but a few words that involve suppletion. go/went is a nice example. We teach people that this is ‘irregular’, but really ‘go’ and ‘went’ are formed from two separate stems and the past tense of ‘go’ was replaced, with went, in about the 15th century as I understand it.

Thirdly, in English we also have words that occupy roughly the same semantic space. I’m going to use devour and eat as my example. In normal usage, eat can take an object, but it can be omitted (some would call this ‘ambitransitive’. So James eats the apple and James eats are both fine. devour is necessarily transitive. James devours the apple is fine, but *James devours is not normally acceptable (yes, I realise that there are some meanings of devours for which this seems okay, don’t write to me about it).

Now… (imagination caps on): imagine that in the 15th century we’d stopped using the present of devour and we’d also stopped using the past of eat. So we started to supply devoured as the past of eats.

James eats the apple

James devoured apple

James eats

*James devoured        (better>) James devoured the meal

So that we considered it incorrect English to express the past without supplying the object.

Do you see how this is a case where two roots can be used together with suppletion, but different syntactical entailments? I know it’s a little bit of a stretch, but I hope you got there.

This, I think, is the best way to conceptualise ἔρχομαι/ἦλθον. The present derives from PIE *h₁ergʰ- , the aorist from *h₁ludʰét, and the two roots encode different voice.

Non-Suppletives: the Greek ‘middle futures’

While the prior explanation of why some verbs are middle in some tense/aspect systems but not others works with suppletive verbs, it does not explain why some verbs are active in one system, but middle-only in another. In particular, a reasonably large number of Greek verbs become middle in the future. I confess, I found this puzzling, and while I have come across some answers, they are not entirely satisfying.

I first turned to R.J. Allan’s thesis on the middle voice. Again, I don’t have published book version, perhaps he had more to say in that than in the thesis. Nevertheless, at the start of chapter 4 on the future, he has this footnote:

Another interesting issue is the occurrence of middle future forms of – mostly intransitive – active presents (εἰμί – ἔσομαι). From a synchronical point of view, the middle inflection of these futures can be explained by their semantics. All verbs in question involve a physical or mental affectedness of the subject, e.g.. perception ἀκούσομαι, motion βήσομαι, receiving λήψομαι, change of state θανοῦμαι. Ἔσομαι appears to be the only exception. Historically, these middle futures may be explained as former desideratives. The middle voice, then, expressed the mental involvement of the subject. For further details, I refer to Rijksbaron (2002: 156).

Now, for the most part that makes some sense. (a) It’s common to say that the future system (and the subjunctive) developed out of a desiderative (< desire, for those who don’t love jargon) form at an earlier stage of the language (aka PIE), and (b) you can see that all (really, all?) the verbs involved in this form of semi-deponency fit into the semantic categories already established: subject affectedness, especially mental involvement, which is heightened in the desiderative, enough perhaps to ‘tip’ an active into a middle only.

Rijksbaron does treat this, on pages 156-57 of his The Syntax and Semantics of the Greek Verb (an excellent read, by the way). For his part, he classifies them as verbs “denoting essential functions of body and mind”, in categories related to sound, various types of excretion or extrusion, physical-and-mental grasping-and-taking, movement, and bodily affection. You can see how most of these are close to the ‘middle’ domain already.

He then says, “This phenomenon” that is, the middle only/dominance in the future) “is not easy to explain”. Following C.J. Ruijgh, he attributes it to the σε/σο suffix for these having an originally desiderative value, and thus also have a preference for middle endings. Thus, the diachronic development hypothesised is that the σε/σο suffix preferred middle endings first, and then was applied to corresponding active forms, but only when the active form would have a meaningful opposition to the existing middle form.

That, I have to say, is quite interesting, if only because of the way the middle form is prior and primary in the diachronic development. Is it true? Hard for me to say. Does it have some explanatory power for middle-only futures? Yes, it seems to.

Although, at the end of the day (and this post), it’s worth remembering that in trying to understand the middle (or any apparently unusual feature of a language), we’re trying to describe what is, and it’s not up to a language to give us some neat system that justifies its logic to us. Yes, often there is a logical explanation for why linguistic phenomenon X is X, but there doesn’t have to be some kind of “this is the way the language thinks about this thing.” Sometimes you just have to say, “well, it’s just like that”.

 

Non-Suppletives: the Latin perfects

There’s a third set of semi-deponents which I confess have resisted my attempts to find a good accounting of. These are the Latin perfects. They are few, being primarily audeō, fīdō, gaudeō, soleō, and their compounds, which switch to a periphrastic passive in the perfect system: ausus, fīsus, gāvīsus, solitus + sum.

I don’t have any answer for these. I’ve tried a few avenues of exploration, but have so far come up empty-handed. I’m very open to hearing from someone a historical-linguistic explanation for these!

Latin middle-only verbs categorised

abitror to think Cognitive, Mental Process
cōnor to try, attempt indirect reflexive (cf. ἐργάζομαι – e.g. self-exertion for benefit)
hortor to encourage, urge Emotive Speech
moror to delay Body motion?
mīror to wonder at Perception, or cognitive
testor to witness Emotive speech
polliceor to promise Emotive Speech
videor to seem Sp-Pr, or Passive-Middle
vereor to fear Mental Process, Emotion
mereor to deserve, earn Indirect
loquor to speak << derived from colloquor ??
colloquor to converse Reciprocal
patior to suffer Passive-Middle
queror to complain Emotive speech
proficīscor to set out, depart Translational body motion
aggredior to approach, attach Translational body motion
congredior to meet, come together Collect. Motion M.
ēgredior to go out, disembark Translational body motion
prōgredior to advance Translational body motion
sequor to follow < PIE chaining-middle ?
ūtor to use, make use of Indirect Reflexive
morior to die Sp-Pr
nāscor to be born, be found Sp-Pr
revertor to go back, return translation body motion? or
orior to rise, arise Sp-Pr (but also, change of body posture?)
potior to get possession of Indirect Middle
opperior to await, wait for ?
ordior to begin ? cf. ἄρχομαι
osculor to kiss naturally reciprocal
conflictor to fight naturally reciprocal
amplector to embrace naturally reciprocal
luctor to wrestle naturally reciprocal
altercor to wrangle naturally reciprocal
copulor to join, be linked naturally reciprocal; stative
misceor to assemble, unite naturally collective
congregor to gather, assemble naturally collective
colligor to gather naturally collective
venor to chase < PIE chaining middle.
consolor to take consolation Mental Event: Emotion
delector to delight in Mental Event: Emotion
misereor to pity Mental Event: Emotion
illacrimor to weep over Emotive Vocalisation
fateor to confess Speech Act
meditor to ponder, meditate Simple Cognitive
interpreter to interpret Simple Cognitive
comminiscor to think up, devise Simple Cognitive
conspicior to perceive, descry Perception
odōror to smell Perception
obliviscor to forget Complex cognitive
polliceor to promise commissive/intentive (complex mental)
scindor to split (intr) spontaneous event
tremblor to tremble spontaneous event, non-volitional movement

So, you want to log hours of language activity

How can you practically do this? How will I be doing this?

This is a follow-on post from this one.

Firstly, let me address an issue raised by my wife, which is comparing my Latin experience with Mongolian is apples and oranges. Because time spent in Mongolia was overwhelmingly communicative exposure, not ‘book time’. That’s precisely my point, in terms of that’s why it took a couple of years not decades to become competent. It’s both quantity and quality of input.

Of course, it’s hard to measure everything, and I don’t want to create paralysis by analysis. So here’s how I’ll be tracking language time in 2019:

I have a spreadsheet set up with the following fields:
Date, Language, Category, Quality, Time, Notes

Category is a drop-down box of Audio, Video, Reading, Communicative, and Grammar. That is, what type of activity did I do during the time.

Quality is a subjective rating from A to E, or 1-5, as you prefer. So “A” is something like a communicative session where we were in the target hour 90%+ of the time. B is an Audio/Video/Reading block in which I stayed mentally in the target language and processed at a decent speed. Etc.. You need to work this out for yourself a little. Grammar is always E. maybe even F.

When I say ‘count hours’, I mean A-quality hours. So I plan to also use the Quality rating to produce “weighted hours”. e.g. B is worth 80% of A time.

Time is measured in minutes.

Notes is where I record what I actually did during that time period. e.g., listening to Latin podcasts, reading some easy Greek, working on a text with a student.

I think this, or a similar system, is easy enough for anyone to implement. I’ll be posting some monthly reports and analyses myself. So at the end of January you’ll see what a month of this looks like.

Re-conceiving the middle voice for Greek and Latin students (IV)

We’ve spoken so far about a range of various ‘middle-domain’ events and actions, and in this post I focus in on our final group, related to Spontaneous Processes and Passive-Middles

(See posts onetwo, and three to get up to speed. All this work derives primarily from Suzanne Kemmer and Rutger Allan, with some others thrown in for fun)

Spontaneous Process events and the Passive-Middle

Kemmer moves on in her monograph to cover a range of ‘other’ categories that “impinge” on the middle (142), the categories of the “spontaneous events” and the “passive middle”.

A spontaneous event is one in which “the entity undergoing the change [of state] is the chief nominal participant” and “in which no Agent entity receives coding.” That is, there’s a subject, who undergoes a change of state, and no agent is mentioned. There may or may not be a ‘conceivable’ agent.

Allan notes that a spontaneous process thus differs from the passive-middle, because the latter implies an unspecified agent, the SP tends to exclude one. SP often have active counterpart verbs in Greek that are causative.

This includes various possible subcategories. E.g. biological processes: dying, growing, aging; physiochemical: melting, freezing, rusting; changes in properties: ‘greening’, ‘squaring’ (okay, I made that one up, but presumably there exists a language where you can regularly form a verb from the adjective ‘square’ and express the change of state that results in a square).

Examples:
ἀπόλλυμαι – die
αὐξάνομαι – grow
γίγνομαι – come into being
τρέφομαι – grow up
φύομαι – grow
καίομαι – burn
τήκομαι – melt
λευκαίνομαι – become/grow white
θέρομαι  – become warm
φαίνομαι  – appear (Allan includes verbs of appearing & disappearing)

Allan also outlines reasons to adopt the “Spontaneous Process” label in preference to anticausative or similar (44) in that the middle form here is not secondary and derivative to a causal version or an active.

For Latin, we find

morior – to die
scindor – to tear, split
nascor – be born
orior – arise    (possible, certainly in its existential sense rather than posture)

Kemmer explores why these are middle in terms of the choice between portraying such an event as having an external Initiator (not always possible), or otherwise selecting the Patient as the chief participant and thus the event as ‘autonomous’.(145) In the case that there is an external Initiator, that participant is deemphasised. So “I die” might be an autonomous spontaneous process, or I might be killed by an angry proponent of grammar-translation, but even in the latter case, I have deemphasised the agent of the action.

Some types of non-volitional movement (e.g. shaking, cf. tremblor) may also be treated under this category; as well as generic ‘happen’ type verbs (γίγνομαι).

Kemmer goes on to consider passive-middles, in which an agent is understood to exist, but is deemphasised, whereas Allan treats that category prior to the SM. Allan also helpfully notes that several verbs in the SP category can be considered to occur with or without an external agent, and so categorisation is difficult in absolute terms. Nonetheless, in both cases, the Patient, and the change-of-state event, are the highlighted or dominant features of the presentation.

That basically concludes all the semantic categories treated in Kemmer, and mostly in Allan. And, I hope by now, you have some sense of how most Greek middle verbs readily fall into these categories.

The η/θη “passives” in diachronic and semantic focus

In the rest of today’s post I want to highlight the work of Allan, and Rachel Aubrey, in considering the η / θη aorist and future “passive” forms.

Allan

Allan’s work is broader, considering various morphological realisations of η vs θη as well as sigmatic aorists vs ‘passives’. In particularly, he does several interesting things: (1) considers the semantic distrubtion of alternatives, i.e. which categories of middle usage are found with which markers, (2) the morphological distribution, i.e. which stem-formations take which endings, (3) the diachronic development, i.e. how these forms shift over time from Homer onwards.

In Homer, he finds that SP favours η, passive or body motion favours θη. In classical, the distribution of all formation types shifts heavily from η to θη.

Over time, the collection of ‘middle-type’ categories, including the ‘passive-middle’, which are represented by the η/θη types, expands across the middle domain. Notably, Allan finds that the sigmatic aorist middle does not occur for Spontaneous Process or Passive-Middles in Homer, (111) but rather for animate subjects with volitional actions (111). There is overlap for mental process, collective motion, and body motion (112). By the classical period, the sigmatic is being used primarily for (indirect and direct reflexive, perception, mental activity, speech act, and reciprocal action), the θη type for (passive, spontaneous process, mental process, body motion, and collective motion) (117).

Why does this matter? Well, (1) the θη types are not a morphologically encoded passive as opposed to the sigmatic middle. Rather, the passive is a semantic subcategory. (2) that means there are simply two middle morphological forms in Greek, with the θη forms emerging from a “spontaneous process + passive-middle” core, but over time expanding to include other middle-type categories. (3) diachronically, the θη forms come to devour the sigmatic middles altogether; (4) in some verbs, contrastive sigmatic-middle vs. θη middle has semantic contrast, between different types of middle usage.

Aubrey

Rachel Aubrey, as I understand it, has been at work on a Masters’ thesis on the middle voice in Koine for some time, and I’m looking forward to it. She gives some anticipation of it in her article ‘Motivated Categories, Middle Voice, and Passive Morphology’[1]. She begins by highlighting the ways in which (θ)η forms break their expectations in not expressing passive syntax (566) and the fundamental problem with called (θ)η forms ‘deponents’ in expressing an ‘active’ meaning and ignoring the middle (567). On Aubrey’s construction (θ)η entered the language as expressing change of state, then extended to prototypical passives, and onward from there (571-2). Aubrey expresses it well in considering a semantic continuum, and that the sigmatic middles tended to express more agent-like or agent-active events, the (θ)η types as more patient-like. (573)

The -(θ)η- form is better understood as sharing a division of labor in the middle domain with the sigmatic middle forms than as an exclusively passive marker with defective, deponent exceptions. It marks the same set of middle event types subsumed within the semantic middle domain with respect to the other middle-passive morphology in the present and perfect paradigms.[2]

She also reaches further back, to PIE, in seeing the origin of (θ)η in the state-predicate marker *-eh1– which grammaticalizes into an aspect-voice inflectional marker (578-9).

Both Allan and Aubrey suggest that the (θ)η emerged in the aorist (and future) but not present because (θ)η is associated with telicity (i.e. the event has a conceptual ‘end-point’ which is reached), which the imperfective aspect forms (e.g. the present tense), do not.

Aubrey’s article goes on to give a robust prototypical explanation of semantic categories, participant roles, and the like, focused on the (θ)η forms. It concludes robustly with a reconsideration of the (θ)η forms along the same lines as Kemmer and Allan point us to.

Changing our categorization of -(θ)η- from the analogous English counterpart (passive) to a typologically attested middle form alters our view of Greek voice. Instead of seeing it as a passive marker with defective active outliers in an active-passive system, -(θ)η- is rightly treated as marking the less-transitive middle events—including passives—within a larger transitivity continuum in an active-middle system. The middle share of the space divides the labor across two morphological forms in the aorist and future compared to one in the present and perfect.[3]

And with that, I leave you for this post. In coming posts I will discuss some varieties of semi-deponency, talk about Latin’s voice system in more depth, and conclude with some application for reading and for pedagogy.

 

[1] Rachel Aubrey, “Motivated Categories, Middle Voice, and Passive Morphology,” in The Greek Verb Revisited: A Fresh Approach for Biblical Exegesis, ed. Steven E. Runge and Christopher J. Fresch (Bellingham, WA: Lexham Press, 2016), 563.

[2] Rachel Aubrey, “Motivated Categories, Middle Voice, and Passive Morphology,” in The Greek Verb Revisited: A Fresh Approach for Biblical Exegesis, ed. Steven E. Runge and Christopher J. Fresch (Bellingham, WA: Lexham Press, 2016), 575.

[3] Rachel Aubrey, “Motivated Categories, Middle Voice, and Passive Morphology,” in The Greek Verb Revisited: A Fresh Approach for Biblical Exegesis, ed. Steven E. Runge and Christopher J. Fresch (Bellingham, WA: Lexham Press, 2016), 620.

Logging hours in a language, not ‘years’

How many years have you been studying Greek?

18 years. Which is useless information. One of the things that has lodged deep in my brain from listening to Bill VanPatten et al., is that measuring ‘years’ is not actually very useful. And here’s a simple comparison. After 1 year in Mongolia I started teaching, not entirely in Mongolian, but certainly in a predominantly Mongolian classroom. By 18 months, I was running at least some of my classes entirely in Mongolian, and at about 2.5 years in, I not only ran my classes 100% in Mongolian, but I preached several times in Mongolian. That was the height of my Mongolian proficiency, and it was all achieved within 3 years.

So, what’s better than counting years? Hours.

I suppose one could count minutes but that’s going to be hard. Hours are more manageable, and here’s where this post is going:

In 2019 I’m going to actually log the hours I spend on language. I already log working hours on several projects, so I’m used to logging my time, I’m going to take that and expand it to logging every block of time I spend on Greek, Latin, and Gaelic, over the year. I’ll probably set up some sub-divisions (reading, grammar, communicative, teaching, learning) as well.

I think this will be a useful experiment for myself, and I am hoping the observation effect will push me to spend some more time on these languages too. And, here is an invitation to you – try logging your own hours for a month or two in 2019, and see what you discover about your own language habits.

Re-conceiving the middle voice for Greek and Latin students (III)

We’ve spoken so far about direct reflexives, and grooming actions, and three categories of body-action middles, including changes in body posture, non-translational motion, and translational motion.

(See posts one and two to get up to speed. All this work derives primarily from Suzanne Kemmer and Rutger Allan, with some others thrown in for fun)

In this post I’m going to talk through (i) Indirect reflexives, (ii) naturally reciprocal events, including collecting and chaining events, and the (iii) cognitive middle. In a 4th post, we’ll cover the important category of change of states and spontaneous processes, and a few odds and ends. A 5th post will treat some categories of semi-deponency.

 

Indirect Reflexives

A prototypical transitive event involves A (Agent) acting on B (Patient). While a direct reflexive occurs when B is also A.

Similarly, a prototypical indirect event involves A (Agent) acting on B (Patient), where C is a recipient or beneficiary. This becomes an indirect reflexive where C is, in fact, A. Kemmer treats this in pages 74-81. It becomes an indirect middle where the action involves is normally, usually, or typically undertaken with A as the beneficiary or recipient. E.g., just as direct reflexives tend to be marked with a reflexive marker (‘oneself’) unless customary/habitual/usual, so too indirect reflexives. She writes, “the indirect middle situation type comprises actions that one normally or necessarily  performs for one’s own benefit”.

Greek:

αἱρεῖσθαι         to choose
κτάσθαι           to acquire for oneself

εὔχεσθαι          to pray

Latin:

liceor              to acquire by bidding
apiscor            to get, acquire
potior              to get possession of

Allan has a fuller list of Greek verbs, which importantly includes δέχομαι, ἐργάζομαι, ἰάομαι, and ῥύομαι (at least in Homer). The ‘healing/preserving’ verbs may originally, he suggests, have had a sense of self-benefit built into the action.

In all these cases, the middle (and it’s very often a middle-only verb in view), indicates without any particular emphasis, that the subject stands as beneficiary and/or recipient.

Naturally reciprocal events

These are events which, by the very nature of the action itself, involve reciprocity. That is, whatever A is doing to B, B is doing to A. This draws on Lichtenberk (1985), and then expanded by Kemmer (p96-9101) to cover collective and chaining type events as well (see below).

Personally, I find the ‘fighting’ verbs a great example of this. In Greek, verbs like μάχομαι, ἐρίζομαι; but the category is not limited to fighting, as evidenced by the inclusion of διαλέγομαι and similar in this category.

Kemmer also considers actions such as meeting, joining, touching, kissing, etc., and the way that a difference in meaning may appear between expressing these with middle marking, versus explicit reciprocal marking. E.g. “They lovers kissed” vs. “The lovers kissed each other” (see page 111-4 for an in-depth discussion of the distinguishability of kisses)

Into this category fall Latin verbs such as:

osculor            to kiss
conflictor        to fight
amplector        to embrace
luctor              to wrestle
altercor           to wrangle
copulor           to be joined

Collective

A collective event differs in that where a naturally reciprocal event involves A > B, B > A, in the collective event, the action as a whole “is carried jointly be the participants involved”, and yet not individually (that is, not a distributed action), but as a group (i.e. the participants have low distinguishabilty from each other). Allan focuses in on particularly collective motion (2.1.5), primarily gathering and dispersing. λύομαι, interestingly, falls into this category, as do similar verbs of dissolution, as does (συλ)λέγομαι .

For Latin, examples such as misceor, congregor, colligor may be adduced.

This is an appropriate place to stop and discuss the troublesome Latin verb loquor. There’s no straightforward category for loquor to fall into. It doesn’t appear to be an emotive speech act (a category coming up…), nor is it easy to categorise as a speech act in which the Agent is normally the indirect beneficiary. It may be derived from a naturally reciprocal verb.

Thus Latin loquo-r, instead of containing an arbitrary instance of -r, can be explained as descended from an old verb of the naturally reciprocal type, possible meaning ‘converse (with each other)’. [Seumas: colloquor?] If this semantic reconstruction is correct (it must be noted there is no direct historical evidence for it) then the Latin verb at some point lost the sense of mutuality and began to occur with singular subjects with the meaning ‘speak’. A similar hypothesis could be invoked for the Latin deponent fo-r ‘speak’. (Kemmer, 108)

Of course, a reconstruction without any evidence is speculative, but it does account for a rather odd middle-only form in Latin.

Chaining

I find this a fascinating subcategory. Instead of a relationship of where A stands to B as B stands to A, or a collective in which individual participants acts as a whole, this is the situation where A stands to B, as B stands to C, as C stands to D. There are not that many verbs (or situation-events!) that typically encode this idea. But one that consistently does so across Indo-European languages is ‘follow’. Perhaps unsurprisingly, this is sequor in Latin, and ἕπομαι in Greek, both media tantum forms. Even Old Irish maintains this as a middle form. Sihler, we may as well note, thinks its PIE root meant “keeps in sight” and is related to hunting (New comparative grammar of Greek and Latin, 449). Kemmer sees a different hunting connection, in that venor may have a similar ‘chain’ type semantics.

 

The cognitive middle

We now come to a rather broad category, which appears to depart from the kinds of ‘middle domain’ situations that can be easily related to the direct and indirect reflexives. What is it about these that tends to create middle-dominant or middle-only verb usage?

Kemmer begins with simple events, and the type of event going on. We have an Experiencer (i.e. the animate being having the mental experience), and a Stimulus (which brings about the mental event). The Stimulus may not be expressed, encoded, or it may be internal to the Experiencer. In any event, the entity involved as Experiencer is decisively the Endpoint for the event, and to a greater or lesser degree, they are (often) the Initiator. That is, it very often tends to be an event that the Experiencer initiates (hence middle), or else passively experiences (a kind of mental passive event, which Greek would encode as morphologically middle anyway).

This meta-category includes emotion events (conolor, delector, misereor, vereor), emotive speech-acts (queror, testor, ὀλοφύρομαι, μέμφομαι, αἰτιάομαι, ἀρνέομαι).

We might note here that Allan splits emotive speech acts from the cognition middle, and treats speech acts as their own category altogether, very often with the Subject as beneficiary or recipient, hence a form of indirect middle. Under that umbrella, he includes verbs of promising, commanding, asking, answering, and begging (e.g. εὔχομαι, ὑπισχνέομαι, ἐντέλλομαι, ἀποκρίνομαι, πυνθάνομαι, δέομαι). But neither are these absent from Latin, e.g. frustror.

Simple acts of cognition fall here too (meditor, interpreter, comminiscor, ἡγέομαι, βουλεύομαι, σταθμάομαι, and many more Greek verbs in Allan (p47)).

So too do perception verbs, especially (per Allan) where the subject is volitional in perceiving. ἀκουάζομαι, γεύομαι, θεάομαι, σκέπτομαι, though not necessarily, αἰσθάνομαι, ὀσφραίνομαι. Cf. also conspicior, odoror.

Complex mental events are those where there is a dependent event, normally expressed as a proposition encoded in a dependent clause (Kemmer, 137). e.g. English, “I forgot that I put my keys in my pocket”. There are two events here “I put me keys in my pocket” – the dependent event, and “I forgot X” – the primary mental event. Cognition type events (rather than emotion or perception, Kemmer p138) are most likely to be middle here.

Kemmer’s analysis differs from Allan’s in carving up the terrain of these various types of events. For example, what Allan takes as speech acts of promising, Kemmer treats as ‘commissive’ events in which the mental source initiates a dependent event. e.g. “I promise to learn how to use the middle voice properly” has a dependent event, “learn to use the middle voice properly” which I, as Mental Source, undertake to put into place. Hence, polliceor in Latin, ὑπισχνέομαι in Greek.

What’s key, in all these cognition middles, I’d say – and by way of concluding today’s rather expansive post – is that the Subject stands as Experiencer and so is “subject affected”. That, and the corresponding commonality that the Subject is very often the Source or Initiator of the event, is why these events are typically marked as ‘middle’. And that’s what this series is mostly about – unraveling the ‘logic’ of what types of events fall in a broad conception of ‘the middle domain’, so that you have a better grasp of that domain’s “realm” and all that falls in it.

Re-conceiving the middle voice for Greek and Latin students (II)

In today’s post, following on from the last, I’m going to walk through some initial semantic categories which Kemmer presents as mostly or typically used for middle-type situations. I had hoped to get through more categories, but it seems that will need a couple more posts.
I’ll provide some Greek and Latin examples, drawn from Kemmer, and from R.J. Allen’s work on Greek.

 

Kemmer starts with the Direct Reflexive. This is an event where one participant does the action to themselves.

e.g. Mike stabbed himself.

In English, we need to mark this with a reflexive form – Mike stabs implies that he stabs someone else (as a habit, probably).

In Greek (and Latin), actions that are normally performed on someone else (er, ‘stabbing’) take a reflexive marker. Allen gives this example:

ῥίπτει αὑτὸν εἰς τὴν θάλατταν (Dem 32.6)[1]   He throws himself into the sea.

But there are actions that are normally carried out on oneself that appear in the middle. These are “body action middles” including several sub-categories.

The first category are ‘grooming’ actions. Actions like dressing, bathing, shaving, decorating, etc.. Here we see typical middle-marking.

ornor               to adorn (oneself)
perluor            to bathe

κοσμοῦμαι      to adorn
λοῦμαι             to bathe

In all these, the participant is both acting upon themselves, but acting upon themselves using themselves. The sense in which you can distinguish ‘agent’ and ‘patient’ is low. For this reason, too, these verbs often lack an active. When you do find an active voice form, its usage is often contrastive – of course I might usually bathe myself, but I can bathe someone else. λοῦω v λοῦμαι shows that kind of contrast.

A second category involves various types of bodily movement. Between Kemmer and Allen you can see three sub-categories of this: change in body posture, non-translational motion, and translational motion.

  1. Changes in body posture involve actions like sitting up, standing, reclining. Again, here the agent is acting upon themselves, but in such a close unity that the difference between ‘agent’ and ‘patient’ is minimal, the event is unitary.
  2. Non-translational motion involves movement of the body but not along any ‘path’. Things like turning, twisting, bending, bowing, nodding, etc..
  3. Translational motion involves moving the body through space along a path.

κλίνομαι          lie down (CiBP)

στρέφομαι       turn around (nTrans-motion)

πορεύομαι       to go (translational-motion)

Where there is an active form it often has a causative meaning:

ἵσταμαι            to stand up/still           > ἵστημι           to cause to stand

στρέφομαι                                              > στρέφω        to turn (something else)

πορεύομαι                                              > πορεύω to cause to go

What’s common about all these verbs is that encoded in the verb is the sense that the Initiator and Endpoint are the same, with low distinguishability, and generally these actions do not involve an external endpoint.

There are less examples of these in Latin, though I think some appear to fall into this category. For instance, the –gredior compounds, e.g. progredior, as well as proficiscor. orior (to rise) seems to me a borderline instance, though it may also fall into the “spontaneous process” category.

This is a good place to stop and talk about deponency for a minute. Is πορεύομαι ‘deponent’? Not by traditional definitions because an active form exists. But the middle form is so prevalent that it may as well be learnt as the primary form. It hasn’t ‘assumed’ an active meaning that has been left vacant by a defective paradigm. Within the ‘logic’ of the Greek language, it’s a typically middle form.

Whether any particular verb is middle only or not depends, in part, just on attestation. If we had no instances of πορεύω, we’d conclude it was deponent. But that’s not really true, is it? ἔρχομαι seems deponent because we don’t have active instances of it. Except for grammarians saying things like:

οὐδεὶς γὰρ λέγει ἔρχω ἢ εὔχω ἢ πέτω ἢ δέχω ἢ ὀρχῶ καὶ τὰ λοιπά, ἐπειδὴ τὸ σημαινόμενον κωλύει.

For no one says “I ‘go’, or I ‘pray’, or I ‘fly’, or I ‘receive’, or I ‘depart’ and the rest, since the meaning prohibits it. (Georgius Choeroboscus, Prolegomena et scholia in Theodosii Alexandrini canones isagogicos de flexione verborum.[2])

The meaning of ἔρχομαι prohibits an active meaning. That’s why there’s no active, not because an active form has disappeared and the middle has picked up the meaning, but because the meaning of the verb is itself middle in its meaning, and an active does not make sense. One supposes that speakers could have coined an active version of this verb with a causative sense, but they didn’t. That’s why this, and similar verbs, are better termed media tantum, ‘middle-only’, rather than deponent. They lack a morphologically realised active form.

So, what does this mean for deponency and terminology? It seems best to lay to rest the term ‘deponent’ if we mean “a verb that has no active morphology but uses a middle or passive form with active meaning”. That’s not accurate to these verbs, especially if conceived of as verbs that actually ‘set aside’ their active forms. It’s far better to conceive of these verbs as media tantum, i.e. verbs that only have middle forms, never active. And then, you have verbs that are primarily used in the middle, where the active is less common precisely because what the active is expressing is less usual. Some verbs prohibit an active form, others disprefer an active form.

Lastly, at least for today, cautioning students that ‘active in meaning’ is not the same as ‘active in English translation’. Because English does not exhibit a ‘middle-voice’ system, in translation the best options will often be an English intransitive active, or an English reflexive. That’s about the best way to render something in English, not about what the Greek means.

In the next post in this series, I will cover some more semantic categories, and then talk more specifically about the Latin middle-only verbs.

[1] R.J. Allen, “The Middle Voice in Ancient Greek. A study in Polysemy”, PhD Thesis. 2002. 65. There is a published version of this but frankly I don’t have a copy.

[2] A. Hilgard, Grammatici Graeci, vol. 4.2, Leipzig: Teubner, 1894 (repr. Hildesheim: Olms, 1965): 19: 27-28.

Re-conceiving the middle voice for Greek and Latin students (I)

I’ve long been of the view that there’s no such thing as deponency in Greek, just verbs that are ‘middle’ and don’t have an active. But recently I was asked (twice) about deponency in Latin, and it got me reading again, which is a great thing. Personally, I’d been helped a great deal by R.J. Allen’s work on the Middle in Greek, and the elaboration of semantic categories. But to deal with Latin, I needed to do more.

And that meant reading the key work by Suzanne Kemmer, The Middle Voice, which “approaches the middle voice from the perspective of typology and language universals research” (1).

In this post and some subsequent ones, I’m going to do my best to translate Kemmer into some more accessible language, and at some point I’ll specifically talk through Latin deponents.

The problem with Greek voice

All my problems started, or continue to start with Greek students, and Greek grammars, especially pesky NT Greek ones. NT Greek courses do a particularly bad job at the nuances of Greek, I find. So, students are very often taught, or at least end up with, a view of Greek that is:

Active:            I hit Mike.

Passive:           Mike is hit by me.

Middle:           “something vaguely in the middle where I am benefited by hitting Mike”

This is usually a very English (vel sim) -driven view, in that the middle is an awkward third-voice squished between the Active and Passive.

Two things, in my own long growth in Greek knowledge, helped get over this. Learning that (a) Greek has an active-middle voice system, and that “passive” is a semantic, not a morphological realisation. e.g. there are no forms that actually mark “passive” in Greek, there are just two sets of middle-voice in the Aorist and Future. (b) that middleness is basically some form of “subject affectedness” (the core of Lyons’ definition, see Kemmer, 1-4).

So, with Greek we are dealing with active v. middle. But even before we get to Greek, let alone Latin, Kemmer helps lay out a broader semantic understanding of events and their participants which provides a lot of the basis for understanding how the “middle” works across various languages.

Kemmer uses a cross-language approach to map out the kinds of semantics associated generally with the middle voice. That is, what kind of meanings tend to be expressed with middle-type structures. She also lays out a bit of a map of how languages mark the middle.

One thing that I found exceedingly helpful was how Kemmer laid out a spectrum of events (at least their conceptualisation).

  1. One-participant
  2. Middle
  3. Reflexive
  4. Two-participant.

If we take the last category first, this is the classic situation where X does something to Y. Let’s use a made-up verb for a little while, grazhonks.

A reflexive event occurs when X grazhonks X. Here grazhonks is the event, and its Iniator and Endpoint are the same. So a language will mark this reflexively, if possible.

At the other end of the scale is the one-participant event. Either there is no Y, or Y is deleted. So, “X grazhonks” or “X bedtweeops”, where no external endpoint is, or can be, in view.

The middle then, as a constellation of subject-affected actions, lies between the reflexive and one-participant, in that an event occurs in which X acts on or with regard to X, but X is conceptualised as one participant, not “two” like in the reflexive.

What ends up in ‘the middle voice’ depends in part upon how a language marks different voice options. Is there a reflexive form, is there a middle form separate or related to a reflexive form? Greek, for the most part, has both an expansive middle system that is separate from reflexive, and which encompasses the passive. Latin, I would say, has an active-passive opposition, but the -r forms also mark some middle uses, especially among historic deponents.

In my next post I’ll walk through the semantic categories that tend to be expressed in the middle or with reflexives, and why that might be so. But let me finish with a nice English example that illustrates some of those 4 categories.

α) I hide
β) I hide the book
γ) I hide the book first, then I hide myself
δ) I was hidden by the rebel alliance.

α is an intransitive active, being used to express a middle-type meaning – an event in which initiator and endpoint are conceived as a single entity. But hide also works as a transitive verb, as in β. We also use it reflexively, as in γ, but primarily when we want to provide contrast (this matches with Kemmer’s observations about how languages with both reflexive and middle markings tend to use them in relation to each other). Lastly, to express the passive in English, we must switch to a passive construction, the intransitive-active-as-middle won’t cut it.

Reading all of LLPSI (extended remix edition) in 2019

I like plans. And I like ambitious ones.

In 2019 I intend to read, or re-read, all the various Lingua Latina per se Illustrata books, including the Ørberg supplements and the not as official Ørbergesque supplements. I’ve read some of this material in the past, certainly the core books, but not that much in the supplements.

Here’s a list:

Familia Romana
Colloquia Personarum
Fabulae Syrae
Fabellae
Epitome Sacrae Historiae
Amphitryo
De Bello Gallico
Sermones Romani
Roma Aeterna
Aeneid
Ars Amatoria
Elegiae (Tibullus)
De Rerum Natura
Bucolica (Vergil)
Cena Trimalchionis
Catalina

That’s a lot of reading. I still have to work out how to get hold of DRN and the Bucolica. They are  Accademia Vivarium Novum editions, and among other things I’d like to order through Amazon.it. Alas, Amazon’s policy of not shipping from any international site to Australia makes that impossible. I’ve been exploring other options, but no success yet.

Anyway, regardless. Perhaps you’d like to read along with me? I’m always up for various types of shared reading.

And, as always, a great way to start a ‘new year’s resolution’ is to start it in November…

Learning to live with your Monitor, aka dealing with error correction

The Monitor Hypothesis is part of Stephen Krashen’s theory of Second Language Acquisition. The hypothesis is that the ‘monitor’ acts to apply conscious, explicit, learned grammar to ‘edit’ your output. The Monitor only does so when (a) you have enough time, (b) you focus on form/correctness, (c) you know a(n explicit rule) to apply. (you could know an implicit rule and apply it to, to be fair)

In Gaelic there is a structure called ‘the inverted nominal’. When your sentence begins with some kind of modal or modal-like construction, e.g. “I want, I need, I like, I dislike” etc.., then a direct object of the verb will precede it (the verb in question takes a form called the verbal noun).

Tha mi ag iarraidh cèic ithe – I want cake-to-eat.

Feumaidh tu bainne òl – You’ll need to drink milk.

 

I’m very familiar with the grammar rules that govern inverted nominals, I could explain them to you over and over. But when I’m speaking ex tempore and at pace, I often get them wrong. It doesn’t help that there are other verbal noun constructions that don’t invert. I suspect that in terms of order-of-acquisition, this one takes a while.

But this, of course, does not and has not stopped a teacher or two along my many-years Gaelic learning journey from both (a) explicitly correcting me, (b) marvelling, ‘How can you know this rule so well but you keep getting it wrong in speech? More practice needed!’

Now, I can tell you exactly why the second statement occurs. It’s not that I need more skill-practice, though it does help my monitor to do explicit skill practice. It’s that I need more and more comprehensible input. I need to hear those inverted nominal structures again, and again, and again, until they get deeply acquired, and not just explicitly learnt. ‘Cause I already learnt them, right? And any time you test me on them, with enough time, a focus on form, then I’ll apply the rules and get it ‘right’. But get my speaking at speed, and they’ll occasionally come out wrong.

 

As for dealing with teachers that like to error correct, even when you knew you said the wrong thing the moment it left your lips! (a not uncommon occurrence)? I’ve learnt to let it go. It depends on context, of course. In some circumstances, you could ask/tell a teacher/tutor to not correct you. In others, that ‘upward’ instruction/management might not be appropriate. Learning a language is relational, and this is one relationship you must navigate. So, as best you can, don’t take error correction to heart – it’s usually well meant, and if you can not get down about it, it probably won’t hurt you!

On neglecting, or choosing not to learn, new languages

I always marvel when scholar X talks about ‘picking up a new language’ like it’s nothing. Or even like it’s something. Perhaps I’m actually bad at languages. (I don’t believe that people are good or bad at languages, aka language aptitude).

For myself, I made a conscious decision to not continue investing in more languages. I’ve written previously about my experiences, learning (to one degree or another) some Japanese, Spanish, Latin, Greek, Hebrew, Mongolian, and Scottish Gaelic, and superficial dabblings in French and German.

I’ve reached a point in life where I know that I do not have the time, either week by week, or long term, to truly learn French of German to a useful point. I have largely abandoned them. My Hebrew is… rusty. 3 years of grammar and exegesis at seminary were indeed useful, but the apex of my Hebrew ability is gone.

And yet, I do not mourn these, except insofar as I mourn the opportunity lost of many good things in this life. But my choice is not a passive one, it’s a very intentional and active one.

It’s the choice to pursue few languages deeper. I want to know Greek, Latin, and Gàidhlig really, really well. ‘Superior Speaker’ well. ‘Read any text with relative ease’ well. Converse with comfort well. And that takes a lot more focus, dedication, and narrowing, than ‘learning’ 15 languages would, or worse, 15 grammar + dictionary usage abilities.

I’ve been at these three a long time now. And not always efficiently. Well, not always optimally. The longer I’m in this game, the better I understand the game itself, getting better at learning languages, and learning these three better.

Epistola ad Praeconem Latinum (editum a Arcadio Avellano), scripta missaque a Albin Putzker, 1895

Amicus noster, in pipiatione nuper hoc misit, atque hic quoque ponendum esse dixit.

Here’s a wonderful letter from Albin Putzker, in 1895. plus ça changeplus c’est la même chose. You can see the original in situ. The footnotes, nisi fallor, are Arcadius Avellanus offering his own thoughts.

 

Univ. of Cal, Berkeley, Oct., 7th, 1895.

 

Your last “Praeco” was particularly good. How can one read with feeling and emotion Latin master-pieces and Latin poetry, if one has not that instinctive knowledge of the language which comes alone from the power of speaking it?–If you read through translations, the best element is lost[1].–The claim that the study languages for the “mental training” is all talk; it is not true[2].–We study languages chiefly to realize, to feel the beauties of the great thoughts, as put by the best minds; in that should the training consist[3].–Let us have life, not more dissection of corpses; living language for living thought. –Of course, mere conversation as such no body advocates. If teachers could speak Latin, the objection would not be made[4], and the teaching would assume a different character; routine would make room for living interest. –Would that we had the right kind of simple, interesting, strictly progressive reading material for the lower schools, such as we have in German and in French[5]. There is much room for improvement in this respect, and a great Latinist could render great service on these lines. – Your article on poor Ulrichs was read by me with deep sympathy; I should like to learn much more about him. Could you not rite more about his life?[6]

With the best wishes for your success,

I am, most truly,

Albin Putzker

 

 

[1]Praeterea, versiones prostant pro 50 libellis.

[2] Bene mones; est mendacium.

[3] Veritas aurea. Quod, per Deos Immortales posset esse absurdius, quam Tullium Hratiumque de ablativis, de “hidden quantities,” de radicibus Sanscriticis, aliisque vercordiis disserendo profanare & desecrare? Classicos sic polluens est quasi sus in hortum elegantissimum irreptus, qui flores devorat, rubos conculcat, gramen depascit, atque proboscides cuncta susque-deque vertit ac convellit.

[4] Sane, minime gentium! O si scirent, quam diverse arguerent!

[5] Habebimus in Tusculo!

[6] Faciemus proxime.

Vocabulary learning

Lately I’ve been reading, among other things, Joe Barcroft’s Vocabulary in Languge Teaching, one of the e-modules in the Routledge E-Modules on Contemporary Language Teaching. And it’s been helpful in thinking through some issues about vocabulary acquisition.

Like many, I studied quite a few historical languages in a very traditional mode. And, I excelled at it – I walked out of 2hr Greek exams in 20 minutes, used space repetition to memorise New Testament vocabulary down to 3 occurrences, and other such feats. But, I also came to think that this approach is basically non-productive of acquisition, and a huge mis-investment of time. By the time I came to study Mongolian, I rarely if ever spent any time explicitly learning glosses for Mongolian words.

And partly that’s due to a shift in how I conceived of lexical items. “Vocabulary” does not mean learning that ἄνθρωπος means “man” or “person” or “human being”. In fact, it doesn’t mean any of those things. Those are, at best, sometimes-appropriate translations of ἄνθρωπος. (And so, when I mark essays, I regularly correct students for mistaking ‘translation’ for ‘meaning’).

Rather, lexical items exist in our mental lexicon as items with a whole set of associated data. We have a core meaning for most of these terms, e.g. table tends to mean “piece of furniture with flat top surface and one or more legs, useful for putting things on”; but we also store alternative, derivative, metaphorical, extended meanings; e.g. table as, say, a set of figures in columns and rows. We also store collocations and phrases, table talk, turn the tables, etc.. And we store things like arguments, i.e. that put requires an agent, patient, location. And we store relations between words, including synonyms, antonyms, homonyms (a different kind of relationship, but one very necessary for puns!) near-synonyms and how the differ, etc.. And we store connotations, and so on…

And this is why giving students a list of words and one or two key glosses, not only is incredibly boring and dull, but is misleading about the nature of vocabulary learning. It suggests that learning vocabulary is simply about mapping a 1 to X set of correspondences between L2 words and L1 words. That, itself, is false. It may be a useful starting point (though I think this is debatable), but to the extent that it reinforces this “laundry list of glosses” notion of what vocabulary is, it misleads students (and to the extent this myth persists among teachers, demonstrates their misledness).

How then do and should we learn vocabulary? Well, unsurprisingly we learn it by input. Repeated exposure to meaning-bearing instances of the novel lexical items, in communicative contexts. The more, the better. And, in fact, this is how you solidify not just a ‘core gloss’, but the variety of meanings, nuances, connotations, collocation, proverbial sayings, etc..

To return to my impetus for his post, Barcroft, he distinguishes between three components of vocabulary (form, meaning, mapping; that is “what the word sounds like, what it means, and the connection”) and suggests that various activities prioritise each of these components, i.e. attention “processing resources” can be devoted more to form, to meaning, or to mapping, but at the detriment of the other two components. I think this is where two of his modules most interesting points (to me) occurred: (1) that using multiple talkers to repeat input of novel items (keeping other values constant), saw increases of 38-64% in target word learning. That’s quite an effect size! (2) that word copying, i.e. copying the target words, actually has a detrimental effect on L2 word form learning.

Barcroft himself articulates (in a separate 2012 book, and briefly in this module), an Input-Based Incremental approach.  Which, I’d largely endorse. It’s core is promoting frequent, and repeated, input of novel words in meaning bearing comprehensible input. Limiting output, especially in the early stages, and promoting L2-specific meanings and usage over time.

I confess, in my own learning I have become very laissez-faire about vocabulary acquisition. I just figure that more input over time means more exposure to vocabulary, and I’ll learn whatever is frequent and relevant. Which is true, but it’s an entirely incidental and haphazard approach. I myself could be more intentional in structuring my own studies, and in lesson planning.

But, at the end of the day, the one thing I would want everyone to go away with, is that “vocabulary” never equals “rote learning a set of L1 correspondences”. It’s anything but that.

 

 

Between two types of translation

I’ve said in the past that it’s just a bit wild to give beginning students translation as a task to do, because translation is a high-order skill, not a low-order one. In this post, I want to explore two different types of translation practice, and how they sit at opposite ends of the spectrum.

Firstly, there’s translation to convey meaning. I haven’t really found a satisfactory label for what I mean here. But let me explain. Suppose you’re a fully proficient bilingual, say in English and Portuguese. And you work professionally as a translator (written texts) or interpreter (real-time spoken communication). Translation in this case is about conveying messages that occur in one of these languages, in the other (source > target), as accurately as possible (and, if in real time, with speed). It involves a high degree of language proficiency, and cultural and domain-specific knowledge.

Even if one of these languages is an L2, and adult-acquired language, and not an L1, a native-acquired language, it’s still a very demanding process. And the level of language proficiency, and both general cultural and domain specific knowledge required to translate well, is high. It’s one reason translators tend to only operate into their L1s (i.e. with an L2 source but an L1 target, not so much vice versa).

Translation of this sort relies upon the translator understanding the source text as is, in the language, and so depends upon a very well developed representation of that language in their mind.

Secondly, there’s “translating to comprehend”. This is what goes on in historical language classes. You get presented with a text, you apply your external knowledge of grammar, and your external knowledge of vocabulary, and you output a translation in your L1. You didn’t understand the text when it was presented in the L2, it was beyond your linguistic competency to process and understand that message. Translation in this case is an external, extrinsic process where you conduct a grammatical, or linguistic, analysis, to produce an L1 version, and thus render the original L2 message comprehensible to you.

That’s not just orders of magnitude different, it’s arguably a different category of process going on. And that’s why I don’t consider that second process, “translating to get meaning”, to be a form of reading. If that’s what you’re primarily doing, you’re not reading, you’re operating on texts beyond your linguistic competency. Which, is not the end of the world. Especially if you’re in a traditional-type program. Just remember that the process of creating a translation of a text is actually a mechanical process that renders input, comprehensible. And you are going to need a lot of comprehensible input if you are going to actually acquire that language, not just learn about it (and learn to practice grammar/translation as an externalised skill).

Quantity and rate of comprehensible input, revisited

Recently, I crunched some personal numbers. Using Italian Athenaze, a text I’m reading for CI purposes, I clocked myself reading about a page of text (approx 130 Greek words) at 50 secs a page, which calculated out at 156 wpm. That’s pretty decent, I would say, and especially since I expect Greek to be more ‘word dense’ than English. For me, this represents reading (i) faster than I could possibly parse, (ii) with some vocalisation and sub-vocalisation (so it could be faster), (iii) at a level appropriate to my understanding.

I would probably make a vocab note at an average rate of 1-1.5 per page, and this includes words I’m unfamiliar with, or constructions I’m semi-familiar with but think are not entirely transaparent in context and a note will help me later. So that’s a comfortable 98% vocabulary/comprehension factor.

I think this supports the kinds of numbers I came up with in that earlier post, though I’d want to downgrade reading speed estimates for most people. It also highlights some problems…

You can’t give beginning students 4 hours of reading a week. There’s no way a student can cope with 280 minutes, say even at 100wpm, 28,000 words, when their effective vocabulary is so low. And there’s not enough in the way of sheltered-vocabulary texts. It’s just not feasible to have them learn 2 words every 100, meet those words several times over, and then write your next story with 102 words + 2 new ones.

So early level students are going to have to do more intensive reading than extensive reading. That’s fine, they should. Because they’re at a level that even simple communicative messages in the TL are full of overwhelmingly ‘unknown’ information.

And even at more intermediate levels, well, quantity of suitable text is always going to be a problem. It’s a little hard to estimate exactly how many pages/words of Greek text It. Athenaze has, but if you really did read 4 hours a week in a college level course, *and* had the implicit acquisition required, you could read the whole 2 volumes within a few weeks. This is not actually possible and not actually feasible.

However, I point out these problems to continue saying, “more reading is better, more reading is possible, and lack of suitable texts, more than anything else, is the external factor holding us back here. But anything is better than nothing. We don’t need to shoot for the stars. 10mins a day, 70 a week, 7,000 words of comprehensible extensive reading a week, that’s a great start for anybody, whatever their level.

Is it possible to think in a dead language? (Yes)

This is probably the most sub-blog post I’ve written, as let’s be honest that it’s a response to a twitter conversation built of a previous post, a twitter conversation in which I appeared more like a bystander.

Here’s the counter-view: you can never truly learn to ‘think’ in Latin because Latin is dead and our access to it is only via translation.

I think this is wrong, but I think it’s wrong in important and interesting ways.

Let me ask a different question: is it possible to think in a living second language?

Gosh, I hope so, because I do this all the time. As do millions of other people. I speak moderately well in Gaelic, and I spoke very well (though now slightly rusty) in Mongolian, and when I speak those languages I think in them. I’m not translating, not (even) mentally. So just anecdotally (and based on a fairly broad anecdata of second language speakers), plenty of people operate in an L2 without ‘translating’.

So what’s different between a living language and a dead one?

Here’s what I’m ‘hearing’ as the difference put forward: there are no native Latin speakers so our only access to Latin is through (a very, very large) corpus of texts. So we never “think” Latin except as we’ve learnt to approximate it by having its meaning translated into (L1).

And this is why this is wrong: there’s no real linguistic difference between this and a living language that’s L2, because in both cases I’m a non-native learner that’s acquiring all this language from an external ‘corpus’, whether written or spoken around me. Sure, the ‘corpus’ of Mongolian I was exposed to was being spoken by native speakers around me constantly for several years, but it was just as ‘external’ as a written corpus of Latin texts.

It is true, that we can never speak Latin as a native 1st cent. BC Roman did, but that’s not what I’m advocating, or arguing for. Neither will I ever speak Mongolian as an L1 native speaker. But neither do I speak English as you do. My English is an idiolect, formed by my linguistic and socio-historical experiences as an Australian of a certain age, gender, geography, demographic, etc. etc.. In this sense, no two people share the same English, they have only Englishes. There is no “English” (unless some platonic “Form of the English”!).

If the answer to whether you can learn to think Latin is “no, because it’s all externalised“, then this is true of all L2s, and so true but trivially true, because no L2 speaker learns to speak an L2 as an L1. That doesn’t equate to all L2 being ‘translation’ or ‘learnt via translation’, that’s a false equivalency.

Let me end by saying, without any offense intended, that if you’re a monoglot whose only experience of a language is an historical one focused on translation, then believing it’s possible to “think in Greek/Latin/ancient Hebrew/etc.” may be difficult, but it’s possible, even at the simplest level. Just this past week I’ve conversed with several people, at different “levels” of complexity (mine and theirs) in a range of languages. This is the norm for most of the world who are proficient in an L2, or the vast number who have more than a single L1.