Why the Cult of the One True Textbook Has to Stop

The following is a transcript of my long-form video essay, “Why the Cult of the One True Textbook Has to Stop”.

In short, if we truly care about improving language pedagogy, we have to stop worshiping Ørberg.

Salvēte omnēs, ego sum Magistra Hurt. I am one of the most vocal critics of the Nature Method textbooks series among Comprehensible Input Latin teachers. So imagine my surprise when Ayan Academy contacted me to create a sponsored video directing people to where, if you want,  you can buy yourself deluxe custom bound copies of Nature Method textbooks. This video is sponsored, and at the end I will provide an affiliate link for you to get a 5% discount off your purchase should you wish. (For those reading this on the blog, it’s this link with the discount code FOUNDINANTIQUITY5)

But the sponsorship or affiliate link does not change my strongly held opinion of the topic at hand. Ayan Academy knows this, and is on board with me giving my honest opinion, which is fortunate because I would have done it anyway. If I sound like I’m roasting this book too harshly, I want you to keep listening to the end because I think you’ll be surprised by my conclusion.

So, let’s get stuck into the Nature Method.

Part 1: A moderately tall person in a land of dwarves

Dēlia ancilla pulchrior est quam Syra.

– Hans Ørberg

In a land of dwarves, a moderately tall person seems like a giant. Language textbooks are a very slow-moving genre, generally far behind what the research says is most effective in language acquisition. They are like the dwarves in this analogy.

To begin with, a lot of them are not even vaguely based on the principles of Comprehensible Input, but on skill-building grammar translation that has been proven to be the least effective approach for many many decades, which nevetheless sticks around because it is the easiest thing to make unit tests and vocab and grammar quizzes from.

But even excluding those, among the “Reading Method” textbooks, those that at least try to support fluency through Comprehensible Input, the pacing and the general learner experience often leaves much to be desired. Examples include textbooks like the Cambridge Latin Course (CLC), Oxford Latin Course (OLC), Ecce Romani, and Suburani, among others. All these reader textbooks, when treated as if they are complete courses, little universes unto themselves, do not by themselves offer enough input for the learner. Paul Nation in a 2014 article estimates that to acquire a vocabulary of about 5000 words a student needs read about 2 million words, and to acquire 9000 words they need about 11 million words of extensive reading.[1] Suffice to say, the amount of words a student needs to read to gain fluency in the language is somewhere on the scale of millions. But textbooks contain much less than that: Lingua Latina Per Se Illustrata Familia Romana, for example, contains around 35,000 words, and the CLC in its first 4 out of 5 books contains around 36,500 words. As a result, all reader textbooks move at a faster pace through the language than the learner moves in their acquisition, and when learners are suddenly confronted with authentic texts after finishing these textbooks, they are still a million or more words short of what they need to succeed in reading fluently.

So with that, the learner of these reader textbooks is either forced to reread every chapter tens of  times before proceeding to authentic material, or to struggle slowly through authentic texts with frequent recourse to the dictionary and a lot of puzzling over explicit grammar rules. Or they are to read large amounts of intermediate material that bridges the gap as they slowly work themselves out of the Intermediate Plateau. This is not usually explained to people when they are marketed a “complete course” in the language.

Lingua Latina Per Se Illustrata – LLPSI – is the moderately tall person, among vaguely tall persons, in the land of dwarves.

It is arguably the best currently available single resource for an adult self-learner to study Latin from scratch with. But it is not by a long shot the best textbook that could theoretically be written for that audience or other audiences. It is “good enough” at what it does, but where I take issue is the way people like to rave about it and hype it up to be something it isn’t. Some people just seem to think there couldn’t possibly be a better resource, and they are dead wrong. This is not by a long shot an ideal course, and that should affect how we talk about it and how we treat it.

A page from the 1957 version of LLPSI

LLPSI was first published in full in 1957 under a different title, Lingua Latīna secundum nātūrae ratiōnem explicāta. Ørberg continued to revise it throughout his lifetime, publishing new versions in 1983 and 1991. But here I disagree with Nancy Llwellyn’s assertion that the LLPSI method was continuously refined over 50 years of a man’s lifetime: sure it was edited and polished, but with all the tweaks and adjustments and refinements, if you compare the 1957 version with the current one, you can immediately see it is fundamentally the same text as it was 65 years ago, in terms of its method, its philosophy, and its course structure. The pēnsa (the excercises) are exactly the same as they were in the 50s. The core method in this book is as it was 65 years ago. This method is not the product of five decades of refinement; it is essentially the product of a small team of Direct-Method-inspired authors, producing textbooks according to a shared mould in the mid twentieth century. The meat of it, and the rules for its learning method, come from a point in time.

Part 2: Survivorship bias and internet echo chambers

Discipulus quī nec stultus nec piger est, sed prūdēns atque industrius, multās rēs ā magistrō discere potest.

– Hans Ørberg

Now Nancy Llwellyn certainly does not claim that LLPSI is without its faults, as she outlines many things to be aware of when using LLPSI in her speech, In Or Out of Ørberg. On those points we agree.

But there are many, many people on the internet – usually not teachers, but self-learners – who, typing reviews in their echo chambers, collectively shout from the rooftops that LLPSI is the theoretical pinnacle of language learning, and the canonical method of using it is to be uncritically followed without deviation from the author Ørberg’s intent. They do not restrain themselves to stating how useful it was to them as a textbook, but instead raise it on a pedestal as the magic silver bullet of language learning. This is extremely naive because it implies that six decades of scientific research into Second Language Acquisition (SLA) somehow has no way of improving language teaching, because the formula had already been magically cracked by these half dozen or so individuals in the mid 20th century.

What can I say? If it worked for you, it worked for 100% of your personal sample size of 1.

Because when you are a self-learner, your sample size is 1. Maybe you could expand that to about a dozen if all your friends are also in on it and are trying it out with you. But friends tend to have similar interests; like attracts like. Even if you expand your sample size to the select people in your friendship circles, you aren’t getting a random sampling. And you will only hear from people for whom it worked.

The result of this survivorship bias is that if anyone were to make up a new language learning method today, no matter how silly, and promoted it widely enough, I guarantee they will find some dedicated pockets of people who will swear by it on this planet of 7 billion.

When you’re a language teacher, your sample size is considerably larger and more random. This year I’m teaching 192 students across six year levels, in ten classes most of them quite full, an average of 19.2 students in each class. I’m in my fifth year of teaching at this school, and the numbers are roughly the same each year, so it’s fairly safe to say I’ve taught a few hundred students in my day. I’ve taught students with learning disabilities, behavioural problems, absolute beginners slotted without preparation into second-year classes, students with chronic illnesses who are absent half the year, gifted students who get everything done in two seconds, English Second Language students, dyslexics, people with language processing problems, ADHD and autism, you name it.

The self-selecting group of self-learners rave about following the rules of the Nature Method.

The school teachers who have taught with these materials or similar, are a lot more realistic and not quite as Messianic about it, and heavily adapt it. Or at least, the good teachers adapt it when they use it; as I will later explain, I have witnessed what happens when you do not.

So from the perspective of a Latin teacher who has been in the trenches so to speak for five years in this school, or six years teaching in total, nearly ten years if you count my tutoring days, and one who has been more recently looking into modern SLA research for evidence-based practices apply in the classroom and in self-learning, let’s talk about the pedagogical validity of the Nature Method in 2022.

Where did the Nature Method come from, and what has happened since 1957 in the realm of best-practices language teaching?

As a broad overview, the Nature Method came about as a written adaptation of the very orally focused Direct Method of the early 20th century, which shares some major similarities as well as differences with communicative language pedagogy today.

The Direct Method involved a lot of high-input practices among other things, but it is not fundamentally based on the Comprehensible Input hypothesis, because the hypothesis had not yet been proposed.

In our zeal to promote the way of CI, we have to resist making up a kind of hagiography, a fanciful story that casts the Direct Method in general and Ørberg in particular in the role of a saint fighting the good fight against the darkness of Grammar-Translation. The Direct method was not perfect in every way, and the way we are teaching now almost certainly is not perfect in every way; this is why we need to critically evaluate our teaching and look to evidence-based practices rather than simply inheriting and replicating another line of grand tradition. In short, if we truly care about language pedagogy, we have to stop worshiping Ørberg.

Part 3: Some casualties of the LLPSI cult

“ō Marce!” inquit, “hōc modō nihil tē docēre possum! tam stultus ac tam piger es quam asinus!”

– Hans Ørberg

Admire the man if you wish, appreciate his work, be thankful for Ørberg’s legacy. But worshiping the doctrines of a mere man – whether living or dead – is an embarassment to the pursuit of truth, and it reflects poorly on our discipline.  

It certainly turned me away: when I left high school and first started my life in classics, I was a Grammar-Translation (G-T) supporter for over 10 years. Then last year around June I did a sudden 180° turn and converted to Comprehensible Input (CI) based methods, tore down all my G-T stuff and started rebuilding everything from scratch. You know what had kept me away from taking CI seriously for 10 years? The slavish adherence of the acolytes of Ørberg to their favourite textbook, LLPSI. I had heard boundless praise for this textbook when I first started Latin tutoring, so naturally I bought it thinking it would be great. I tried it out with my tutoring students, and started reading things with them directly from the book. It was a flop: my students and I really couldn’t get the stories to hold our interest for long, and I also didn’t know all the techniques for making reading interesting and interactive that I’ve been more recently learning (it’s not like the textbook comes with any instructions on how to create the necessary variety of reading activities to go with it).

At the same time, a young teacher in a nearby school, let’s call him “Tim”, radically transformed his middle school Latin curriculum to be based on LLPSI, and I was given a string of tutoring students in the middle to low achieving bands who had been taught by this guy but had been left far behind by his approach. (I later found out that “Tim” had not been teaching equitably, but had been favouring the high-achieving students and simply allowing the middle and low ones to slip. However, at the time I was tutoring, all I could find out about him was that he had learned Latin through LLPSI just like those people on the internet, he had fallen in love with the approach and thought it was the best thing since sliced bread; he had condemned G-T with a passion; he had then radically transformed the middle school curriculum with LLPSI at the centre, thinking with great certainty it was the best thing to do, but it clearly hadn’t worked for everyone because if students dropped behind in his class, they dropped FAR behind.)

So, stung by all these small- and large-scale failures, I switched right back to what I had been taught and what seemed to work better for everyone: a hybrid of G-T and the reading method. To me LLPSI had been so over hyped as a “fun”, “nearly effortless” method for learning Latin that for it to be boring and draining and often incomprehensible in reality was an utter betrayal: no one on the internet at the time seemed to have any words for this book except unreserved absolute praise and a total blindness to what a compelling narrative is, and none of them would back down on the most minor point of criticism for their master, Ørberg.

Over those years I hardened my heart against CI because all the people promoting CI were clearly just a group of mutually-brainwashed people lifting up a 1950s textbook as if it was the inspired, God-breathed word of the Latin language itself. I wonder how many other potential allies to our CI based methods have been driven off by the blind and stubborn defensiveness and reverence that the Living Latin proponents have for Ørberg? You can’t just take responsibility for the successes of this method: you also have to own your failures.

Part 4: The Direct Method

Nunc tempus est numerōs discere. Prīmum dīc numerōs ā decem ūsque ad centum!

– Hans Ørberg

So, where did LLPSI come from if it was not handed down from on high? As I alluded to above, LLPSI is a product of the Nature Method, and the Nature Method is itself a written adaptation of the mostly spoken Direct Method. Let’s start by explaining what the Direct Method really entailed and move from there to see how the Nature Method Institute and LLPSI came to be shaped the way they were, while evaluating LLPSI and other Nature Method books in the light of modern SLA-based practices.

The Direct Method of the early 20th century was not exactly the same at all times and in all places. Teachers conceived of it as a broad collection of language teaching practices that roughly fell under the umbrella term: ‘the Direct Method’. In 1917, Mark Skidmore wrote an article about the Direct Method in the Modern Language Journal. He wanted to define what people meant by the method, what was essential to it, and what was more variable in its practice. He sent a survey to 140 language teachers about it, and 74 of them replied; in this sample 19 said they were currently using the Direct Method, 16 said they were not, and the rest said they were adapting various features, ranging from “largely” to “trying to use it”.[2]

From these responses, Skidmore concluded that these were the main features of the Direct Method:

“there is almost unanimity on the following six points:

  1. good pronunciation (phonetics)
  2. extensive oral work
  3. inductive teaching of grammar
  4. real reading, not mere eye-reading, as the basis of the instruction
  5. translation is reduced to a minimum
  6. much use of “free composition”.[3]

Regarding grammar, Skidmore provides a quote from Professor Myers which better describes the role of explicit grammar in the Direct Method: “The direct method as we are using it here is not an inductive grammar method taught in the foreign idiom, but primarily a language course, in which the students are taught to see and imitate the expression of certain ideas in the foreign idiom, with sufficient training and drill in the elements of grammar to strengthen their sense of certainty and security in using the language.”[4] That is to say, it’s not a French grammar lesson taught in French, and not even primarily a grammar lesson taught inductively, but a language course in which students are expected to mimic and reproduce what they hear to express ideas with the help of drilling, until they are accurate in their output.

Skidmore also described some more specific activities employed in the Direct Method, that he cautioned were not universal:

“The use of gestures and pictures, suiting the action to the word, singing, reciting in unison, the entire exclusion of the vernacular are not universally, nor even generally, insisted upon as indispensable.”[5]

On the final point, language immersion, only 8 out of the 74 respondents said that the Direct Method means “the entire exclusion of the vernacular from the class room.”[6] But then, as now, there was a vocal minority who insisted that 100% immersion was the only way to go. Skidmore provides an anonymous quote, “If by the direct method is meant a method that excludes the vernacular as much as possible, I favor its adoption anywhere and everywhere.”[7] I can’t help but feel that the sentiment being expressed here is that total immersion, no matter how poorly implemented, is unquestionably superior to all other possible teaching methods.

Speaking realistically, teachers tend to observe serious attrition in immersion programs. Bob Patrick, a pioneer and advocate of communicative Latin teaching, commented on the subject, “Immersion almost always insures that some will become so lost that they will give up.” Lance Piantaggini, also a communicative Latin teacher, shared that he had “seen adults cry at a conventiculum. Two different people, actually.” A conventiculum is a Latin convention – often a multi-day event – in which everyone only speaks Latin, and they attend conferences in Latin. Attendees to conventicula typically sign a formal contract swearing that they will not use a language other than Latin for the duration of the conventiculum. Lance continues, “These are people who paid for the experience. That’s not a typical/regular/normal – whatever you want to call it – learner, and even *they* didn’t make it.”

As good as it is to maximise use of the target language in the classroom, that doesn’t have to mean total exclusion of the native language at all costs. Even in the 1917 survey of the Direct Method, the majority of teachers did not think the Direct Method essentially required banning first language (L1) use in the classroom.

So let’s focus on Skidmore’s unanimous six points:

Firstly, there was a primary emphasis on pronunciation. It is interesting that pronunciation tops the lists of essential features in the Direct Method; one of the respondents to Skidmore’s survey remarked that the method “presupposes phonetic drill”. Modern communicative attitudes about pronunciation are often more ambivalent about enforcing pronunciation, except where it would hinder communciation. As an example, Randall Buth, who promotes the use of spoken Ancient Greek, does not insist on correcting students for sub-phonemic distinctions, but only where distinctions could carry a potential difference of meaning in the language concerned. The main difference between the Direct Method practitioners and today’s communicative language teachers is that today’s teachers are more concerned with communicative competence, whereas the Direct Method teachers were concerned with accuracy of output. Our priorities and measures of success, in pronunciation as well as other areas, have moved away from demanding complete accuracy and towards supporting proficiency.

The reason for this general shift towards prioritising proficiency before accuracy is because, if you want to end up with both, you will get both faster if you pursue proficiency before accuracy than the other way round. I have observed it in my classes: the ones who hyper-focus on details of relatively minor importance to the detriment of their overall understanding tend to progress slower in acquisition than their peers (but get everything right in grammar tests). The last remaining languages in which teachers continue to insist on valuing accuracy as a special virtue in itself are the dead languages – languages in which it is acceptable for respected and tenured professors to be terrible at reading fluently. Ultimately, though, this is not an either-or situation: if you care to think of Latin as a human language and not just a mark on an exam or an ornament on a certificate, you will want both proficiency and accuracy. And you can gain this most efficiently when you aim for increasing proficiency ahead of increasing accuracy.

Secondly, the Direct Method was said to contain extensive oral work. Oral activities could often include the teacher holding up a picture and describing it in the TL, or acting out a verb and describing it; the teacher and students would then take turns describing the pictures or actions, racking up lots of repetitions orally. To me this sounds very similar to TPR, Total Physical Response, a comunicative method in which teachers command students around the classroom in the TL and often describe and interact with the physical space, pictures, and props. The difference here is in the role of oral production: it sounds like the Direct Method had a goal of getting students to repeat the language aloud in oral drilling, whereas TPR spends a lot more time, especially initially, in getting students to respond and interact with the commands non-verbally, before they are comfortable enough to respond and interact verbally. The reason why TPR teachers don’t demand verbal output from students until they are ready is based on Krashen’s Comprehensible Input Hypothesis (that language acquisition is driven by understanding input, not producing output) and his Affective Filter Hypothesis (that the unconscious processes of language acquisition are significantly impeded if the student feels stressed or anxious).

The Direct Method’s emphasis on the oral meant that textbooks were not actually necessary to the success of the method – a feature shared by both the Direct Method and modern TPR again. One of the possible definitions of the Method considered but rejected by Skidmore included the comment, “the conversational work may  [italics supplied by Skidmore] be supplemented by text book or note book.” [8]

Thirdly, grammar was taught “inductively”. Inductive means that the student sees an example, and works out the rule themselves, rather than being told upfront what the rule is. The previous quote from Prof Myers that the Direct Method is not an inductive Grammar method but a language teaching method has to be put in the context of saying, when grammar was taught, it was taught inductively, even if grammar was not focused on in isolation. But we have to be aware that inductive learning is not the same as acquisition. Induction is student-led rule-making; the student is tasked with discovering explicit grammar rules. By contrast, acquisition is the nearly unconscious process of getting so familiar with a grammar feature that the learner becomes internally, instinctively able to know when the feature is used correctly or incorrectly, and eventually produces the feature without exerting mental effort to edit how they speak or write.

On the use of reading in the Direct Method, Skidmore calls it “real reading, not mere eye-reading” and I honestly don’t know what he means by eye-reading – probably reading without understanding, or mere reading aloud. But the role of reading stories with understanding is emphasised in the Direct Method, and I can’t help but compare this with TPRS, Teaching Proficiency through Reading and Storytelling, a set of practices in which the teacher cooperatively writes and reads stories with their students. TPRS  is often set up to follow on from oral TPR practices. The Direct Method’s combination of oral practice followed by story-reading appears quite similar in character to the TPR-TPRS route that some communicative language teachers are taking today.

The opening story in Carolus et Maria, 1933.

But the character of the reading material of Direct Method texts differs quite significantly from the character of CI texts written today. You’ll notice that as soon as you pick up material from the early 20th century like Carolus et Maria, Maxey and Fay’s A New Latin Primer, Latin for Today, Cornelia, etc. The learner texts were not primarily written to present interesting content through the language, but to teach the language through rather plain but comprehensible material. Skidmore reports a quote from Professor Holbrook objecting to the use of literary material in the classroom, which reads, “The student should never be allowed to forget that what he is primarily attempting to learn is the French language.”[9] He then adds metaphorically: “one cannot become a botanist or a gardener by merely loving flowers, and usually it is those persons who know most about them who love them best. Similarly, the beauty of a linguistic construction, its fitness, should be most apparent to him who understands it best.”[10] Interestingly, knowing about the language here is touted as the key to appreciating literature. It is the language that is the object of study, not the meaning expressed in the language. The content is more or less supposed to serve as filler while the student concentrates on language learning.

That is in direct contrast to best practice communicative language teaching today – to encourage acquisition and not mere inductive rule-learning, students ought to be consciously focused on the message expressed by the Comprehensible Input, and indeed interested in understanding it, as it is through their attention to the content that their subconscious langauge systems can best make form-meaning connections. Stephen Krashen strongly advocates not for dull texts or even just “interesting” input, but “compelling” input. He elaborates, “Compelling means that the input is so interesting that you forget that it is in another language.”[11] By contrast, Direct Method practitioners seem to have deliberately avoided making their materials too compelling or interesting; the foreword to Cornelia puella Americana states, the learner “should not be called upon to deal with situations outside his own experience or to acquire knowledge through the new medium”.[12] The things that pass for stories in the Direct Method seems to focus on being orderly, expected, and obvious. Suspense and surprise, as critical as they often are for constructing compelling narratives, are avoided in the writings of the Direct Method.

Point number five: translation is reduced to a minimum. Notice the wording: it’s not that translation is an unforgiveable sin, or that a person instructed in the Direct Method could never become a professional translator. But in the language class, for pedagogical purposes, translation is “reduced to a minimum”. SLA research since that time has consistently criticised the grammar-translation method’s excessive dependence on translation for language learning.[13] It’s not to say that one couldn’t possibly learn a language through translating; it’s that the effect of oral interaction, extensive reading, and other high input activities is so much greater, that it is a relative waste of time to spend very long on translating compared to the more profitable activities. In this point, the Direct Method and SLA-informed best practices are in agreement.

Finally, point six of the essential components of the Direct Method as reported by Skidmore was “free composition”:  Skidmore does not define this term, but the wording suggest free writing, where the student writes whatever they want to. If I have understood him right, this is definitely one of the practices in modern communicative language teaching; students are given a timer and asked to write as much as they can within 5 minutes or so, and they can say anything. It’s a way to increase fluency of output, at the pace set by the students themselves. The products can then be collected into a writing portfolio and used as evidence of progress.

However, “free composition” may not have been always as universal as Skidmore suggests here; an article from 1949 in the ELT journal  advocates for rather constrained composition activities, stating “until a fairly advanced stage is reached, the exercises should be restricted to the completion and conversion types. Exercises in which the learner is to compose material wholly his own are best deferred.”[14] This is suggests that within the larger umbrella of the Direct Method, there were teachers who set very constrained writing production activities.

Those types of highly constrained writing activities (eg. the conversion activity of “turn every singular word plural and every plural word singular”) are fundamentally not communicative language activities: no one is expressing meaning, the utterance is not being addressed to an audience who will hear the message, and there is no possibility of a meaningful response. There is no “information gap” being bridged between interlocutors. These types of fill-in-the-blank or change-the-word-endings activities do not by their nature encourage the connection of meaning and form, but of rules and form. And the 1949 paper makes that explicit: “conversion exercises have the advantage of combining thought (about the grammar mechanism or whatever it may be) and association (eg. between the present perfect tense and the certain class of adverbial)”.[15] That is to say, the exercises are intended for developing explicit command of the grammar through rules.  Plain and simple.

It is quite surprising to hear from the 1917 article that free composition was a nearly unanimously agreed feature of the Direct Method, when it is strongly advised against in the 1949 article, but this shows the variety of practices and mindsets within the Direct Method.

As you can see, the Direct Method was not applied the same way in all places; a lot of it came down to the teacher in the classroom. Many teachers did not publish their materials, and so their methods died with them. Many others did publish what they developed, and a lot this is excitingly now in the public domain.[16] The products of this movement serve as potentially quite useful Comprehensible iInput, even if they are not very compelling. Try reading them at least; your mileage may vary.

Part 5: From the Direct Method to the Nature Method: What was adapted, and what was lost?

scrībite hanc sententiam: homō oculōs et nāsum habet.

– Hans Ørberg

So now we move over to the Nature Method and its adaptation in LLPSI. The first textbook of this distinctive style was an English self-learning course published by the Nature Method Institute in 1939 by the name of English by the Nature Method. In 1954 the Institute also published a French Course, Le Français par la «méthode nature». By then Hans Ørberg had joined the Nature Method Institute as a permanent employee from 1952, and was tasked with making a Latin course which replicated the same method as the other two books. LLPSI was finally published in full in 1957.

So to be clear, Ørberg did not largely invent the method, but adapted it (with a great deal of creative decision-making, to be fair) to Latin’s situation. The base format however had been pioneered nearly 20 years prior in 1939 with English by the Nature Method.

Georges Bonnard in the preface to that book outlines the goals of the course, saying that it is “aimed at providing those who wish to learn English and are denied the help of an ordinary teacher, with a text-book that might, in little over a year, bring them to the point where reading English books and conversation in English may be, or at least begin to be, actually possible.”[17] The absence of a teacher is assumed from the beginning; the course is not designed to be used in conjunction with a teacher, but to make up for the lack of a teacher. This would naturally explain the lack of guidance in how teachers might adapt the material to suit the needs of a classroom environment. [edit: but for the record, there is currently an official teacher guide for LLPSI online.]

As to the demographics of the target audience, Bonnard hints that the promoter Arthur M. Jensen’s “main concern has been with young people in the world of business” stating that it is there that “he will mostly find young men and women who feel the need of some knowledge of English and have never had the opportunity of getting it. But he has taken care not to give undue importance to their requirements, so that his course may be used with just as much profit by whoever desires to learn English by himself.”[18] It was not aimed at an intellectually elite audience, which is absolutely great, but it was also not aimed at children or teenagers in schools. It was written for young professionals with a self-determined motivation to learn the English language.

So, Jensons’ writing team (his daughter and colleague; Jensen himself did not write most of the content of the course) were producing an English course for a motivated, adult young professional who due to their life circumstances, didn’t happen have a language teacher. And the model they looked to adapt for a teacherless textbook was naturally, the Direct Method. 

Let us now examine what made the transition from the Direct Method to these books, and what did not.

I have identified seven major features they brought in from the Direct Method:

  1. Exclusive use of the Target Language. English by the Nature Method is written completely in English, with no recourse to any other language. Likewise, LLPSI is fully in Latin, teaching Latin-through-Latin. This reflected some of the practices of the Direct Method where TL use was maximised; but it is important to restate that the total banning of L1 was not necessarily practiced by the majority of Direct Method teachers, and “immersion” practices are not mandated or necessarily beneficial for communicative language teaching today.
  2. Pronunciation was stressed in its importance. This is challenging for a written textbook, but English by the Nature Method emphasised pronunciation by providing IPA transcriptions under words. LLPSI doesn’t appear to have as much overt emphasis on pronunciation, but it does print macrons.
  3. Grammar was taught inductively. Each chapter introduced a new target feature, a grammar lesson of the day, which the learner was meant to consciously discover through working out the meaning of the sentences. This is true for all the nature method textbooks. It essentially meant that grammar was withheld until its grand entrance in a chapter all about that feature. The result is sometimes some very odd circumlocution was employed to avoid using grammar that hadn’t been formally introduced, even if the context would have strongly suggested it. For example, in Capitulum X of LLPSI, Aemilia sees her son being carried, and asks “Why can’t he walk?” In the story, the boy has just fallen out of a tree. The natural expectation is to say “he fell out of a tree”, repeating the same words as before but in the past tense. But the perfect tense had not been formally introduced. So instead Julius says “Quintus can’t walk, because he is not a bird and does not have wings! He who wants to fly but cannot, falls to the ground!” This mindset of sheltering grammar at all costs and insisting on students mastering rules one at a time leads to missed opportunities where a fairly innocuous unexplained grammar feature could have been woven in when the context demanded it, in which case it would have made the storytelling clearer and ironically more comprehensible.
  4. Reading as the basis of instruction. The Nature Method books inherited the style of storytelling which had been present in the Direct Method for decades: very obvious descriptions of pretty safe subject matter where for long stretches of time, mundane realities are described without incident. That is not to say there aren’t plots in these Nature Method books; it’s just that the stories were clearly not designed for entertainment or to function as interesting let alone compelling stories, but for being the filler material which the student can dissect for discovering rules in the language.
  5. The use of pictures and visual aids. The illustrations in the Nature Method textbooks are all very practical, clear, and relevant to the story at hand. Teachers have always wished there were yet more visuals of course – if there could be a drawing for every meaningful utterance, we would be very delighted. As it stands, the illustrations that the text provides are very meaningful and practical aids for language learning.
  6. Translation is reduced to a minimum. The Nature Method textbooks indeed do not encourage translation, but reading.
  7. The incorporation of constrained and somewhat-unconstrained composition exercises. English by the Nature Method had Exercise A, which was a fill-in-the-missing-word activity, followed by Exercise B, which was an answer-the-question activity, the questions being fairly closed questions with predictable answers. For the first few chapters, Exercise B is further constrained by providing a certain number of dashes corresponding to the number of words expected in the answer. Exercise A, the fill-in-the-missing-word activity is not terribly useful because it is possible fill in the missing word without understanding what was being said. As for Exercise B, answering a question is not as free as “free composition” said to have been essential in the 1917 paper on the Direct Method, but at least it involves the processing of meaningful messages and the task of expressing a meaningful answer. The exercises are arranged in order of most-constrained to least-constrained: this is consistent with the advice of the 1949 paper. LLPSI has three sets of Exercises: Pēnsum A is a fill-in-the-endings exercise, Pēnsum B is a fill-in-the-word exercise, and Pēnsum C is respond in Latin to Latin questions of a fairly closed nature.

    As with English by the Nature Method, LLPSI orders the exercises (pēnsa) from most-constrained to least-constrained. The exercises are supposed to be integral to the method; in practice they act as gatekeepers to the rest of the content: the learner is expected to produce perfectly accurate output of everything they have encountered up to that point before moving forward. In the light of modern SLA research, this is actually an unrealistic expectation for acquistion: if input is the basis for acquisition, and acquisition is a gradual process, your output accuracy must reasonably be expected to lag behind your comprehension.

    When faced with the pēnsa, the learner essentially has two options: repeatly reread the chapter as many times as necessary to memorise – and not necessarily internalise – the forms of that chapter, or consciously apply explicit grammar rules to get through. If the learner does neither of those things and just skips the pēnsa, the authors can then blame the laziness of the student if they fail in any way. Failing at the pēnsa means you apparently weren’t ready to move on.

    I tend to believe that the pēnsa are more beneficial to the reputation of the course than to the learning experience, because making accurate output a kind of arbitrary gatekeeper to receiving more input is not exactly in the best interests of the learner, if what really drives acquisition is sheer volume of comprehensible input. And if you’re only allowed to progress if you’re perfect, only the perfect will stick with the course, and the rest will drop out and you won’t hear from them again.

    Add to that, as Nancy Llwellyn points out, the pēnsa as they are ordered in the book demand language competence in the reverse order of its actual acquisition: in reality, learners acquire large chunks and whole words before they acquire smaller details like the endings of words.

And so these seven features of the Direct Method, both good and bad, were adapted to the print textbook medium of the Nature Method textbooks. But what was lost in the transfer between the classroom practice of the Direct Method and the teach-yourself-textbook of the Nature Method?

Extensive oral practice. The greatest loss between the two formats is the loss of extensive oral practice which had formed the bedrock of the Direct Method. A book cannot talk with you, point to things, describe them, ask you questions, laugh, give you commands, receive commands, or any of the myriad rapid-fire oral interactions that a teacher can do with you in person (or which could be filmed and shared via video, like my introductory Latin TPR video series). In my experience with large and diverse classes, building a bank of concrete vocabulary through oral activities, especially TPR, makes reading and storytelling a lot more accessible to students when you get round to it. Actually what I noticed this year is that the students who turned out to be particularly problematic in other ways, behaviourally or academically, tended to do well on TPR. People who doubted their language abilities tended to respond well to TPR too. TPR in particular and comprehensible oral activities in general are in my experience far and away the best start you can get in a language: low pressure, high levels of input, quick feedback cycles, lots of interest and variety.

But it is not something a paper book can easily convey, in the absence of a teacher. Even if you try to read transcripts of TPR or oral conversations in books – and there are books that do that[19] – the book can never capture the charisma of the teacher in the moment, and the reality of seeing something true right in front of you being described. In comparison, starting with a map of the Roman Empire followed by the description of a fictional Roman aristocratic family doesn’t have the same power to hold students’ attentions and prepare them for more interesting stories.

On the topic of oral language teaching, it was actually oral methods which converted me to Comprehensible Input. When I saw videos of Aleph with Beth, a course that teaches Biblical Hebrew through TPR, comprehensible input, oral questioning, and eventually storytelling, I was blown away by how much more effective and enjoyable it was at getting Hebrew to stick in me compared to the grammar method, or inductive method. Aleph with Beth videos are such an outstanding resource in the CI teaching space, and so gentle in their learning curve, that I think they put Nature Method textbooks to shame. And in no small part this is because they can capture and reproduce real human interactions in speech with live props in a way that a print textbook simply cannot. If Aleph with Beth had been made in Latin, I believe it would be worth ten LLPSIs.

(Side question, would you like it if I tried dubbing Aleph with Beth straight into Latin? I know that’s totally not the proper way to structure an introductory language course, as Aleph with Beth is carefully built around Hebrew grammar, but you know, just suggesting it if you were interested.)

Part 6: Does LLPSI beat everything else?

pāstor humī sē prōiciēns dominum ōrat nē sē verberet: “nōlī mē verberāre, ere! Nihil fēcī!”

– Hans Ørberg

So in summary, how do Nature Method textbooks such as LLPSI stand up in 2022? 

It doesn’t beat a real teacher, by which I mean someone like Justin Slocum-Bailey, who can captivate a class with a spontaneously generated spoken Latin role-play scenario within his second lesson of Latin with them, and spend the entire time with their attention on him as he speaks and interacts 90%+ in Latin with them. It does not hold a candle to a live teacher doing wide varieties of oral activities and interactive story-telling.

But, as a resource competing among resources, is it, as Nancy Llwellyn said, “simply 80 times better than anything else”?

I’d say it is better than at least most textbooks in some important ways, but worse in other important ways. Taken as a whole it could be a better single resource than all the alternatives. However I qualify this by saying this only works when students – in both self-learning and classroom contexts – are somehow kept interested in the content either with external help from the teacher or fom sheer internal drive.

However, I don’t think that it’s much of a high praise to say that LLPSI is perhaps better than other textbooks overall. The sad fact of the matter is that there aren’t a whole lot of very useful truly CI-based textbooks published as complete courses. Teaching lags a few decades behind the science, and it seems textbooks lag much further behind the vanguard of teachers.

The reading method from the 1980s and onwards produced a lot of courses such as the Cambridge Latin Course (CLC), the Oxford Latin Course (OLC), Ecce Romani, Suburani, Latin through Ovid, and others, which present Latin through stories, but they all introduce vocabulary at a high rate per story, and rely heavily on glossing, to such an extent that a large portion of students are often not really reading the stories but stringing together glosses or spending most of their time flipping through the dictionary at the back of the book. It’s not really reading when every five or so words you have to stop to look up a new word you’ve never seen before. Unless the teacher heavily modifies the materials, for a student to get the experience and benefit of extensive reading in these courses, they have to reread previously understood stories. And as much as I tell students to do that on their own time for their own benefit, I really don’t think it happens unless the teacher specifically carries out an in-class activity that makes them reread (such as the many post-reading activities listed here).

The one advantage of these reading method books, however, is that they tend to be more interesting as stories than LLPSI and the Direct Method materials, particularly the CLC. However sometimes I feel the authors of reading method textbooks try a little too hard to tell a story a certain way, and you see a sudden concentration of glossed vocab on the side when they do that, which very often ruins the pacing of the story when a student attempts to “read” it. Crafting a story which is both compelling and comprehensible takes special effort beyond just thinking “it’s a an adaptation of a myth and the myth says XYZ, so now I will say XYZ and gloss it.”

There is also a pretty nice newcomer on the textbook scene, Via Latina, which is entirely written in Latin with Latin glosses, similar to LLPSI. Its learning curve however is a bit steeper than LLPSI and it seems to presuppose a classroom context in which the teacher will be assisting students with getting extra vocabulary practice, perhaps through oral questioning, which makes it less immediately useful to self-learners, but still useful. The written exercises in Via Latina are far better than LLPSI and any other course I’ve seen in the textbook space. They provide a lot more connection between Latin words and their meaning, as well as greater variety.

In addition to complete textbook courses, however, we also now have a large number of Latin novellas for extensive reading and incorporation into TPRS based courses. Novellas tend to tell lively stories with as little new vocabulary as possible, thus providing the gentlest and potentially most entertaining way to read large volumes of Latin for fluency. But they also are not without some flaws. It is a very recent and rapid development in Latin teaching. As of 2022 the genre has only been around for 8 years, but there are already over a hundred novellas published. Most of these have been written in North America by Latin teachers working overtime, and are voluntarily edited and proofread by other Latin teachers in their overtime, with the result that the products are not edited to the same standard as massive big-budget textbooks. There are other questions around the Latinitas of some novellas which I hope to address more fully in a later video. LLPSI as a single resource beats out a lot of other single resources, but that doesn’t mean a subset of the best novellas couldn’t rise to the surface and eventually form a reading list that beats LLPSI out of the ring.

But ultimately, trying to find a “winner” is not the right question for us. Because what really matters most is the sheer amount of massive comprehensible input you receive in the language. For this reason, I will go on record with a rather firm endorsement of LLPSI: if it were beaten either in the present or future by not just one but two superior “complete” introductory courses, I would still recommend learners give the book a try and see if it works for them, as additional Comprehensible Input among everything else they are doing. For its sheer amount of comprehensible content, LLPSI doesn’t have to be “perfect” be recommended, or even “anywhere close to perfect”. It is “good enough” – so darn “good enough” – that it wins even if it loses.

Part 7: Final thoughts

Dominus rīdēns eum currentem aspicit, tum ad vīllam revertitur. Etsī dominus sevērus exīstimātur, tamen inhūmānus nōn est.

– Hans Ørberg

So, when all the dust has settled, I do actually end up endorsing and recommending LLPSI as a resource. Use it to learn and teach the language, by all means, just don’t uncritically accept the book in its specific format or the author’s methodology as the best possible way to learn a language.

We all have a tendency to look fondly on the first Latin textbook that really got us over the hump towards reading Latin. For a lot of people in the living Latin movement, that book was LLPSI, and their natural emotional response is to remember all the good times, and forget the bad. When you teach from a textbook, you’re forced to relive the bad parts over and over and over, believe me! So I get a lot more of an allergic reaction to the bad parts of a textbook than someone who just remembers it as their Latin textbook from years ago.

Writing this has also forced me to reflect on why I react so abrasively to LLPSI. I have long observed that when I read the stories, I feel an irrational surge of rage building in my blood, which I cannot quite explain, but nevertheless cannot shake. Perhaps part of it is the writing style: I’ve been sprinkling in sentences from the book to adorn the section headings here, and there are so, so many examples of things that I would never say in my own voice. Things like “stupid children can’t learn” or (said by a husband) “You are less skinny than you were back then,” or (said by a child) “how tall is the wall of the military camp?” And it has a very pro-slavery, pro-corporal punishment bent to it. But I don’t have this strong a reaction to the awkwardness, corporal punishment, or slavery references in other textbooks like the OLC.

I think a large part of my unresolved rage issue towards LLPSI is actually that I had quietly suppressed the story about “Tim” and the casualties of his failed curriculum reform. For years I was tutoring students who had lost a lot of Latin progress because he had not met their needs with LLPSI. I formed long-standing teacher-student relationships with these students whom he had overlooked. They helped me refine my Latin tutoring techniques, and I was very pleased to see them grow and thrive. But “Tim”, through either conscious or unconscious negligence, had not intervened with them. Instead he had covered up their failure with rigged tests that were impossible to fail (eg. a test with 20 items in it, marked out of 100, where an incorrect response counted as a single mark off from 100). He soon left the teaching profession after finishing one year of teaching with LLPSI, and pursued an alternate career. For years I had refrained from talking about this former teacher because at that time I was very early in my Latin career, and the world of classics is pretty small. If I badmouthed him in front of the wrong people, or tarnished the reputation of the prestigious school which had hired him, I worried it would damage my future career. But I think the injustice of what he had done with the curriculum had been eating me with a hidden flame; every time I see someone on the internet praise LLPSI to the high heavens it triggers that cognitive dissonance. I can’t help but think of this LLPSI fanatic who wreaked havoc with that methodology in a classroom. This is where you end up if you just hand people LLPSI and tell them to “get good” or get out, because it’s Ørberg’s way or the highway.

To my relief now, it seems enough time has passed – I haven’t run into him in nearly ten years – that I think I can finally tell the story in detail safely and put it to rest. For those of you in my area who can easily put two and two together and work out who I’m talking about, I ask for grace if I have misremembered or confused details of this old story. I have not been trying to think about it for a while.

Interestingly, I have yet to feel even the slightest bit angry while reading the Italian entry in the Nature Method series, l’Italiano secondo il metodo natura. This further suggests that it is my personal history with the fanaticism of LLPSI and not fundamentally the resource itself which makes me cringe so strongly against LLPSI. I’m not sure if I will ever bring myself to enjoy LLPSI, but on the basis of my research, I cannot deny its sheer utility for Comprehensible Input based language learning.

Hard cover Nature Method textbooks offered at Ayan Academy – click here to get 5% off your order.

Now if you’re in the mood to get a deluxe, custom, hardbound copy of any of the Nature Method textbooks mentioned in this talk (they don’t stock LLPSI but they stock the other languages), hop on over to the link shown on screen, and use the discount code FOUNDINANTIQUITY5 to get a 5% discount on your purchase. I can personally vouch that the Italian one, l’Italiano secondo il metodo natura, is a particularly good member of this family. I never restricted myself to only reading that resource for Italian, but I found interleaving it with other Italian input sources continues to be of great benefit to my language acquisition.

Anyways, valēte omnēs!


  • Carreres, A. “Strange bedfellows: Translation and language teaching. The teaching of translation into L2 in modern languages degrees: Uses and limitations.” In Sixth Symposium on Translation, Terminology and Interpretation in Cuba and Canada. Canadian Translators, Terminologists and Interpreters Council, (December 2006).
  • Hornby, A. S. “Diret Method Composition Exercises (I)” in ELT Journal, 4 no. 1 (October 1949): 22-27.
  • Jensen, Arthur M. English by the Nature Method. 1939.
  • Maxey, Mima. Cornelia Puella Americana. Chicago: University of Chicago Press, 1933.
  • Nation, Paul. “How much input do you need to learn the most frequent 9,000 words?” in Reading in a Foreign Lanugage 26, no. 2 (October 2014): 1-16.
  • Skidmore, Mark. “The Direct Method” in The Modern Language Journal 1 no. 6 (March 1917): 215-225.

[1] Nation, How much input do you need to learn the most frequent 9000 words?, 5.

[2] Skidmore, The Direct Method, 221.

[3] ibid, 217.

[4] ibidem

[5] ibidem

[6] ibidem

[7] ibid, 220.

[8] ibid, 216.

[9] ibid, 219.

[10] ibidem.

[11] Krashen, The Compelling (not just interesting) Input Hypothesis, 1.

[12] Maxey, Cornelia, vii.

[13] for instance, see Carreres, Strange bedfellows: Translation and language teaching.

[14] Hornby, 22.

[15] Ibidem.

[16] See Carolus et Maria, Maxey and Fay’s New Latin Primer, and Cornelia for examples of Direct Method Latin textbooks in the public domain on Wikisource.

[17] Bonnard (1951), in the Preface to English by the Nature Method, 5-6.

[18] Ibid, 6.

[19] See the Polis Institute’s Forum for Latin, and Saffire & Freis’s Ancient Greek Alive for an Ancient Greek example of TPR written in textbook form.


One response to “Why the Cult of the One True Textbook Has to Stop”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: