The Natural Order Hypothesis: Why Grammar Is Acquired in a Predictable Sequence

Estimated reading time: 7 minutes

The natural order hypothesis holds that learners acquire the grammatical structures of a second language in a broadly predictable sequence, and that this sequence does not change simply because a teacher presents the structures in a different order. It is the most empirically grounded of Krashen’s five hypotheses, resting on the morpheme studies of the 1970s, and it carries an uncomfortable implication for curriculum design: a syllabus can control what is taught, but not what is acquired next.

Executive Summary

Research beginning with Dulay and Burt in the 1970s found that learners of English acquired certain grammatical morphemes in a similar order regardless of their first language, their age, or the sequence in which they were taught. Some structures appeared early and reliably; others resisted acquisition long after they had been explicitly taught, sometimes for years.

Krashen incorporated this finding as the natural order hypothesis. His argument is not that teaching is useless but that teaching order and acquisition order are different things. A structure taught in week three may not be acquired until a learner is developmentally ready for it, and no amount of drilling brings that readiness forward. For district leaders the consequence is that persistent errors in a specific structure are frequently a developmental signal rather than an instructional failure, and that a curriculum sequenced purely by grammatical complexity is optimising something the learner is not following.

Key Takeaways

• Acquisition follows a predictable sequence that is broadly consistent across learners.
• Teaching order does not determine acquisition order. The two are independent.
• Some structures resist acquisition long after being taught, and that is expected.
• The evidence base is the morpheme studies, which makes this the best-supported of the five hypotheses.
• A persistent error is often developmental, not a sign the lesson failed.

What Is the Natural Order Hypothesis?

The hypothesis states that the grammatical structures of a second language are acquired in a predictable order, and that this order is a property of the acquisition process rather than of the syllabus. Learners pass through the same broad stages whether or not anyone has taught them in that sequence.

Quick answer: the natural order hypothesis says learners acquire grammatical structures in a roughly fixed sequence that instruction cannot reorder. Teaching can supply input and awareness; it cannot make a learner acquire a structure before they are ready for it.

A familiar illustration is the contrast between the progressive marker in “she is running”, which tends to appear early and stably, and the third person singular in “she runs”, which is taught early almost everywhere and frequently remains unstable in a learner’s speech for years. The second is not harder to explain than the first. It is simply acquired later.

Where the Evidence Comes From

The natural order hypothesis is unusual among Krashen’s claims in that it summarises an independent body of empirical work rather than proposing a mechanism. In the 1970s, Heidi Dulay and Marina Burt studied children acquiring English from different first-language backgrounds and found strikingly similar acquisition orders for a set of grammatical morphemes. Subsequent morpheme studies extended the finding to adults and to other languages.

The finding has held up better than the interpretation. Later researchers have debated how fixed the order really is, how much first-language background shifts it, and whether the morpheme studies measured acquisition or accuracy in particular task conditions. What is not seriously disputed is the core observation: learners do not acquire structures in the order a textbook presents them.

Teaching Order Versus Acquisition Order

The failure mode in the last row is worth dwelling on, because it consumes real instructional time. When a taught structure does not appear in student output, the intuitive response is to teach it again, more thoroughly. If the natural order hypothesis is broadly right, that response addresses the wrong constraint. The learner has understood the rule and cannot yet produce it spontaneously, which is a different problem from not knowing it, and is closer to the territory of the monitor hypothesis.

What It Means for Instruction

1. Expect persistent errors in specific structures. Third person singular, articles, and plural markers commonly stay unstable long after teaching. This is normal.
2. Do not diagnose from a single structure. An error that everyone makes at a given stage says little about an individual student.
3. Keep supplying input above the current level. The hypothesis pairs with the input hypothesis: readiness advances through comprehensible input, not through repetition of the target form.
4. Sequence for communication, not for grammatical tidiness. Organising a syllabus around what students need to say is more defensible than organising it around perceived rule difficulty.
5. Judge progress by range, not by one form. A learner producing longer, more connected utterances is progressing even while a specific morpheme remains unstable.

District Benchmark: The Error Everyone Is Still Making

The hypothesis makes a prediction that is easy to check informally and that most EL teams find persuasive when they do.

Take the errors your teachers report most often across an EL population. In nearly every district the same small set appears: the third person singular, articles, plural agreement. Now ask when those structures were taught. In most curricula they were introduced in the first year, and many students in the third and fourth year still produce them inconsistently.

If the explanation were instructional, that pattern would vary between teachers and between schools, because instruction varies. It does not vary much. The same errors persist across classrooms with very different teaching quality, which is what a developmental constraint looks like and is not what an instructional failure looks like. A district that recognises this stops spending remediation time on the structures that will resolve on their own schedule, and redirects it to what actually moves: the volume of language a student produces and receives.

Why Readiness Still Depends on Volume

It would be a misreading of the natural order hypothesis to conclude that nothing can be done and the sequence simply unfolds. What moves a learner along the sequence is exposure and use, which is where the Speaking Time Gap enters: the shortfall between the responsive spoken practice an English Learner needs and what one teacher can supply across a full class.

The math, framed as developmental time. If the order of acquisition is fixed but the pace is not, then the only lever a school holds is how quickly a learner accumulates the input and interaction that advance readiness. Two students with identical developmental sequences will arrive at the same structure at very different times depending on how much usable language passed through them in between. A student getting a few minutes of genuine spoken interaction a day is advancing along the same sequence as a student getting an hour, just far more slowly.

That reframes the persistent error usefully. The structure is not going to be reached early by teaching it harder. It is going to be reached sooner by a learner who is doing more with the language in the meantime, which makes the quantity of interaction the variable worth managing.

Common Mistakes District Leaders Make

1. Reteaching a structure the learner is not ready for, and reading the continued error as a teaching failure.
2. Building a syllabus on grammatical complexity and assuming acquisition will follow it.
3. Flagging developmental errors as evidence of a disability, when the same error appears across the whole cohort.
4. Measuring progress by a single target form rather than by the range and length of what a student produces.
5. Concluding that instruction does not matter. It supplies the input that advances readiness; it just cannot reorder the sequence.

Recommended Actions

Immediate (this month): Ask your EL team which errors they see most, and check whether those structures were taught years earlier. The overlap is usually striking.

Medium-term (this year): Shift progress monitoring away from single target forms toward measures of range, length, and complexity of student output.

Long-term (strategy): Increase the volume of comprehensible input and genuine interaction, since pace, not order, is the part a district can influence.

Questions District Leaders Should Ask

• Which errors do our teachers report most, and when were those structures taught?
• Do the same errors persist across teachers and schools, or do they vary?
• Are we referring students for evaluation based on errors the whole cohort makes?
• Does our progress monitoring track range of output or only accuracy on target forms?
• How much genuine interaction does each English Learner get per day?

Frequently Asked Questions

What is the natural order hypothesis?

The natural order hypothesis is Krashen’s claim that learners acquire the grammatical structures of a second language in a predictable sequence, and that this sequence is not determined by the order in which structures are taught.

What evidence supports it?

The morpheme studies of the 1970s, beginning with Dulay and Burt, found similar acquisition orders for grammatical morphemes across learners from different first-language backgrounds. Later work has debated how fixed the order is, but the core observation that acquisition order differs from teaching order is not seriously disputed.

Does it mean grammar teaching is pointless?

No. It means teaching cannot reorder acquisition. Instruction still supplies input, raises awareness, and builds the explicit knowledge that functions as an editor, but it does not make a learner acquire a structure before they are developmentally ready.

Why do students keep making the same errors after being taught?

Because the structure has not yet been acquired, even though the rule is known. Forms such as the third person singular are taught early and commonly remain unstable for years, which the natural order hypothesis predicts.

How does it relate to Krashen’s other hypotheses?

It is one of five in the Monitor Model, alongside the acquisition-learning distinction, the monitor hypothesis, the input hypothesis, and the affective filter hypothesis. The natural order hypothesis describes the sequence; the input hypothesis describes what moves a learner along it.

Related Resources

• Stephen Krashen’s Theory of Second Language Acquisition
• The Monitor Hypothesis
• The Input Hypothesis
• The Affective Filter
• Stages of Second Language Acquisition
• Complete Guide to Second Language Acquisition

Conclusion

The natural order hypothesis is the most testable of Krashen’s five, and the one most likely to change how a district reads its own data. The errors that persist across every classroom in a system are not a verdict on instruction; they are the sequence doing what it does. What a school controls is not the order but the pace, and pace runs on how much language a student actually receives and produces along the way.

Sources: Colorín Colorado, Language Acquisition: An Overview; Reading Rockets, Second Language Acquisition.

The Part a District Can Actually Move

There is something clarifying about a hypothesis that tells you what you cannot change. The sequence is not negotiable, and no amount of reteaching brings a structure forward. What is left is the pace, and pace is a function of how much usable language a student receives and produces.

That narrows the problem considerably, and it lands where every other line of this research lands. Input can be delivered to a whole class. Production cannot; it happens one exchange at a time, and it is the part of the equation a full classroom rations most severely.

This is why some districts add interactive practice that runs alongside instruction rather than competing with it for the teacher’s minutes. Telo AI is one example, adapting to each learner’s level, in English plus Spanish and French for bilingual programmes.

See how districts increase the volume that developmental pace depends on: https://mytelo.ai/how-telo-works/

The order will take care of itself. How long a student spends at each point in it will not, and that is the number a district can still change.