So, I recently saw this blog post, and I am offering something of a public rejoinder here. I don’t know Carla Hurt, and I actually have no particular issue with the specific intervention suggested here, but there’s a systemic misunderstanding of CI, Krashen, and Ørberg in this post that is reflective of a more widespread misunderstanding of those three things.
So, firstly, the comprehensible input hypothesis. Hurt offers it in this format:
“a learner should be introduced to the each feature of the language incrementally, by receiving input that contains their previous level of competence plus the next feature (i + 1).”
That, however, is not the comprehensible input hypothesis. The CI hypothesis is that learners have a mental representation of language, as well as extra-linguistic information (i), and when presented with input that contains (i+1) where 1 represents ‘the next amount of information beyond i, which is comprehensible to the learner, then they will acquire that next increment.
It is specifically not a claim about explicit language learning, about general competence in a language, and most importantly here it is not a claim about the order of introducing features of a language. The Natural Order hypothesis, which has good research support, tells us that learners acquire features of a language in a predictable and unskippable order, but that is not an order for teaching, and Krashen would explicitly reject that idea.
Rather, Krashen, and those who have built upon his work or otherwise implemented CI based strategies in their teaching, understand language acquisition to be a primarily implicit process in which learners subconsciously acquire i+1 without being explicitly taught what the +1 ever is at any stage, and nor should you try to control the +1. Meaningful messages in communicative contexts is the driving force of CI-based methods, not incremental introduction of new language features.
Understanding this shows why Hurt’s title is wide of the mark – the issues she draws attention to is not a failure of CI as a principle.
And yet, understanding CI in this way also helps to understand why Lingua Latina per se Illustrata is not a CI-based book or method. Ørberg wrote LLPSI along the lines of the direct or natural method, which has some affinities to CI, and it is true that “Latin CI folk” really do love Ørberg, but LLPSI is not a CI based book.
You can see that in the very fact that Ørberg introduces new grammatical features, step by step, building on what the learner should know, and then expects the learner to explicitly grasp those features, and learn them based on a few exposures. Ørberg’s book(s) are not designed to provide lots and lots of comprehensible input, they are designed to teach through a grammatical sequence. For CI folk, LLPSI’s great strength is that it does indeed provide a considerable amount of comprehensible input, with at least a reasonably compelling story. But mistaking LLPSI for CI or vice versa is a very common mistake in the Latin teaching world.
To repeat: a CI based approach is not built upon specifically attempting to sequence the grammar (syntax and morphology) that a learner is exposed to. Nor does it have any expectation that students will notice, or develop an explicit, awareness of those features. It’s built on comprehending messages whose purpose is primarily communication, not primarily pedagogical.
So Hurt’s particular issue is the target feature of accusative endings. And her lament is,
“The error comes in believing that if a student successfully translates a sentence which happens to contain the target grammar feature, they necessarily understood the target grammar feature.”
Which I agree with. Correct translation is no guarantee that a learner has understood a grammar feature. It isn’t even necessarily an indicator that a student has comprehended the sentence. Personally, I am more interested in whether they have comprehended it than whether they can translate it, but that’s a separate question.
Hurt points out that piling up SOV examples doesn’t help, and this is where she uses LLPSI as her example text. And it’s true – learners will assume the first noun is the subject, and won’t necessarily process the inflectional endings. That, though, has nothing to do with CI or LLPSI, but rather the First Noun Principle (FNP). The FNP means that learners (across the board, cross linguistically) process input by assuming that the first noun or pronoun they encounter is likely the subject. Similar to this is the Lexical Preference Principle – learners will tend to process lexical items (e.g. a temporal adverb, heri) rather than an inflectional ending (e.g. perfect tense, –vit). And it works, of course, until it doesn’t (Latin isn’t even strictly SOV anyway, so it is bound to break down).
Hurt’s response is:
“My favoured approach to teaching students to really heed the accusative case is to give them many examples of a sentence type I call “SOV-OV” and work on them in an environment that gives instant feedback:”
Which is pretty close to VanPatten’s work on Processing Instruction, which involves just these types of specifically structuring input to remove the dependency on, e.g., those above principles. For example, provide tensed sentences without lexical items marking time, and then ask for tense-based responses. Similarly, providing specific instruction that disables the First Noun Principle will cause learners to process other features, such as inflectional noun endings.
However, it’s not even (necessarily) clear that this is a case of the failure of Comprehensible Input. Processing Instruction does appear to work, but I wonder whether this emphasis gets quite close to a form of the Noticing Hypothesis developed by Schmidt, a hypothesis that has been critiqued partly for having no particularly clear underlying basis, nor being testable.
As I stated at the outset, I think the type of intervention Carla Hurt recommends or uses here is actually quite valuable, but I do so because of what it does: it asks learners to process input by relying on a feature of the language they need to acquire (inflectional endings). However, the broader misunderstandings of CI, Krashen, and Ørberg, need to be challenged because we can’t keep having debates where we misunderstand and misrepresent others’ views and approaches.