Why foreign language teachers have to rethink their approach to grammar instruction

images (7)

In one of my latest posts I made some recommendations as to how grammar instruction should be implemented. One particular point made in that post seems to have resonated the most with my readers:

Never say ‘my students have learnt structure ‘X’ effectively’ unless you have evidence that they can perform it accurately under Real Operating Conditions.

What this statement alludes to is a common misconception amongst many L2 teachers that a given grammar rule has indeed been acquired by their learners if these can articulate it and/or apply it accurately in the context of gap-fill exercises, multiple choice quizzes, translations or written pieces. This assumption leads to a misguided approach to grammar teaching, i.e. one that:

  • Teaches grammar through means which merely impart intellectual knowledge, i.e. how the target grammar rule works (e.g. verb formation, contexts in which structure ‘X’ should be used and not used);
  • Involves the students in the application of the grammar rule in contexts in which working memory’s attentional systems have more time to monitor performance than real-time communication would normally allow them ;
  • Does not aim at high levels of routinization (i.e. automization) of the target structure application, i.e. the transformation of declarative knowledge into procedural knowledge. In other words, focuses on the conscious application of the target grammar rule, not on its automatic (and accurate) implementation;

Such an approach can have harmful consequences for learning, especially in the absence of systematic and well-planned recycling of the grammar structures taught (another common flaws of much grammar teaching). The reason being that in the absence of routinization the learners are likely to make errors in contexts where they experience cognitive overload, such as oral unstructured communicative practice, essay writing under time constraints, or any other circumstances in which their working memory’s attentional capabilities are drastically reduced (e.g. when under stress).

The recurrence of such errors may lead to their automatization and to the consolidation in Long term Memory of erroneous forms relative to a given grammar structure which may end up competing for retrieval with the correct structure. For example, a student who has learnt the Perfect tense formation rule “Auxiliary ETRE + PAST PARTICIPLE’ (for verbs like ‘ALLER’ or ‘SORTIR’) but has not had the time and practice to routinize it, may get it wrong several times – as it often happens – and say ‘J’ai allé’, for instance’ when performing under R.O.C (real operating conditions). If that mistake keeps happening over and over again and it is not treated effectively it may become automatized; when that happens, the student will end up storing in their brain two cognitive structures referring to ‘Aller’ in the Perfect Tense: ‘J’ai allé’ and ‘Je suis allé’(the correct form). When under communicative pressure or stress, the two forms will compete for retrieval and the more automatized structure will win. Notice that the automatized structure – not necessarily the correct one –  WILL win the retrieval race even though the student does consciously know the rule and will be able to self-correct the mistake once he is cued to its occurrence. This has huge consequences for teaching and learning.

Before delving further into the implications of the above point for L2 grammar instruction, let me quickly reiterate some key points made in previous posts about grammar acquisition and automatization

Automatization or Routinization –

Automatization (or routinization) means that the performance of an L2 grammar rule is applied without having to ‘think’, so to speak. In other words, the performance of the grammar rule bypasses consciousness and Working Memory. This is easy to understand, but how does one measure routinization, i.e. the extent to which a grammar rule is automatized?

To my knowledge, no studies so far have measured automatization in quantifiable terms. And for teaching purposes it is not necessary in my opinion to know how many milliseconds a native speaker takes to deploy a grammar structure. The kind of automatization that a teacher would want to detect in their students’ oral output will be dependent on many factors, such as the years of instruction, individual variables, the type of structure whose deployment one is assessing, the linguistic context in which the structures is being used (familiar or unfamiliar), etc. Hence, it is up to the teacher to decide, based on the specific context they operate in, how close to native speaker speed they would like their students’ performance to be and which criteria best serve their pedagogic purpose. I use the very simple scale below to assess the accurate automatization of a given structure. It allows me to assess speed and accuracy of deployment simultaneously. The categories are quite broad but allow me to get useful enough data.

Very fast     Highly accurate   Quite accurate    Fairly accurate   Inaccurate   Highly inaccurate

Fast            Highly accurate   Quite accurate    Fairly accurate   Inaccurate   Highly inaccurate

Fairly fast   Highly accurate   Quite accurate    Fairly accurate   Inaccurate   Highly inaccurate

Slow            Highly accurate   Quite accurate    Fairly accurate   Inaccurate   Highly inaccurate

Very slow    Highly accurate   Quite accurate    Fairly accurate   Inaccurate   Highly inaccurate

The tasks one uses to assess automatization and the task linguistic environment will have a bearing on the accuracy and speed of rule application. Hence, one has to stick to the same task/linguistic context for the whole year if one wants to map progression consistently. I tend to use oral translation tasks in which the language used is familiar so as to focus the student’s cognitive resources solely on the target structure.

The three routes to grammar acquisition (routinization)

According to my espoused theory of L2 acquisition, Skill Theory, grammar rule routinization in a typical classroom setting occurs along three routes:

(1) Procedural to Procedural : this route does not involve explicit grammar rule learning. To go back to the “je suis allee’ example, the learner will learn the perfect tense of ‘Aller’ without having to consciously understand the mechanics of its formation. They will just learn ‘je suis allé’ as a chunk, like you do with a lexical item. They will also learn the other perfect tense forms of ‘Aller’ in the same way. This approach allows for quick routinization but lack generative power because the students do not have a rule to generalize to other verbs / contexts.

(2) Declarative to Procedural: this route is the most commonly used by L2 teachers. The target rule is taught explicitly before being practised. For instance, the teacher will teach how to form the Perfect tense of ‘Aller’ and of all the other verbs requiring ‘Etre’ as an auxiliary, ensuring that the students understand the underlying pattern. Automatization takes longer to occur along this route as the learner will have to automatize every single step in the application of the rule.

(3) Mix of (1) and (2): in this approach, which is my favourite, one uses the procedural-to-procedural way first, then, once the learners have had sufficient exposure to the target rule, the rule is brought into their conscious awareness and is explicitly taught – connecting the dots a posteriori so to speak. Using the ‘Je suis allee’ example: the teacher would first provide lots of exposure to the verbs requiring ‘Etre’ in the Perfect Tense teaching them as ‘chunks’ in context; then, after they feel that they have been routinized, the teacher will provide them/ask them to infer the rule(s). This approach combines the strengths of both approaches but requires more effort in planning and resourcing than the other two – and also more creativity, as one has to design activities which bring about opportunities to practise all the target verb forms before the onset of the declarative stage.

Implications for teaching and learnin

1.Aim at grammar rule routinization. Whatever the approach one elects to undertake, the crucial issue remains the same: the learners must automatize the target structure before one can safely assume that it has been internalized by the students. So, first and foremost, one must provide grammar practice conducive to automatization (i.e. fast and accurate rule application); secondly assessments of the kind outlined above gives us a fairly good idea as to where our students are in terms of accurate automatization and cognitive control.

As far as the learning activities which foster automatization are concerned, the same recommendations I have made for fluency development in previous posts apply here. The learners need lots of practice in applying grammar rules under time constraints in linguistic contexts which become increasingly more challenging. So after an initial stage which includes the usual gap-fills, ending manipulations (e.g. by using my verb trainer at www.language-gym.com) and written translations in the absence of any time constraint and communicative pressure, two stages should ensue: firstly, a stage in which the rule application occurs in writing under time constraints (e.g through L1 to L2 translations on MWBs ); secondly, a phase in which the grammar rule is applied in speaking in response to a stimulus (e.g. oral translation, picture task, conversation, etc.).

The key issue is that teachers aim consciously at automatization and plan for it by ensuring there is extensive practice in routinization and recycling of the target structures.

2. Do not jump steps and plan for horizontal progression. If grammar structure ‘Y’ requires the routinization of structure ‘X’ as a prerequisite for its successful uptake, then structure ‘X’ will have to be drilled in over and over again until it has been acquired. Hence, for ‘Je suis allee’ to be taught successfully (using route 2) the learners will have to have routinized the formation of the verb ‘Etre’ as well as past-participle formation. This is not often done; teachers do usually ensure that the students know the formation of the present of ‘Avoir’ and ‘Etre’ and of the Past Participle before embarking in the learning of the Perfect tense but do not ensure that they have been automized. In many cases the teacher explains Past-Participle-formation rules in the same lesson in which they teach and practise Perfect tense formation. This poses an unnecessarily heavy cognitive load on the learner. Thus it is crucial that teachers plan for horizontal progression by ensuring that extensive practice in target structure application is provided to learners until automatization has reached the desirable level (e.g. Fast application / quite accurate).

3. Assess grammar rule routinization – not understanding/knowedge of . Assessment of the kind envisaged above ought to be carried out to ascertain that the target grammar rules have indeed been automatized, especially when they refer to key structures. To ‘know’ / ‘understand’ how a grammar rule works does not equate with having acquired it; so do not use assessment tasks that merely test knowledge and understanding of the target rule

4. Teacher response to error on specific grammar structures should be based on their level of automatization. Teachers often get frustrated at seeing errors recurring over and over again despite many grammar explanations and corrections. However, neither the explanations nor the corrections are conducive to automatization. Teacher response to errors relative to a given structure should aim at provide extensive practice in that structure’s deployment leading to its accurate automatization. Only at that stage will the errors in the application of that structure cease to occur or at least drastically diminish. Hence, simply correcting errors with the Perfect Tense of ‘aller’ when they write ‘J’ai alle’ in an essay at the very early stages of routinization of the ‘Etre verbs’ rule may constitute a helpful reminder, but is hardly likely to eradicate the error.


Accurate automatization of the target grammar rule being the ultimate goal of any explicit grammar instruction, foreign language teachers may have to rethink the way they teach grammar and assess its internalization as well as their response to learner grammar errors. Foreign language learners must receive extensive practice in target grammar rule application under time constraints and R.O.C. (real operating conditions).

Teachers can only claim that their learners have acquired a specific grammar structures when they can deploy it fast and accurately across a fairly wide range of familiar and unfamiliar contexts. Hence, the assessment of grammar uptake must be carried out not through typical traditional means (e.g. gap-fills, grammar rule explanations or written translations) but through tasks which elicit from them a fast response in cognitively demanding contexts.

Curriculum designers must take into consideration the fact that grammar rule acquisition requires extensive practice. Hence, sufficient time must be allocated to grammar teaching – if it is one of the course’s priorities – and frequent opportunities for recycling must be carefully planned for.

Finally, let us not forget that accurate automatization of grammar rules contributes to fluency enhancement. This is a further reason for fostering its attainment as much as possible in our daily practice.