Ten misconceptions about EPI (Part 1): on grammar and phonics

Why this post?

In the last decade or so, a large number of language educators around the globe have embraced my approach, EPI (aka Extensive Processing Instruction). Many teachers and schools have done so ‘wholesale’, applying it to all aspects of their MFL curriculum; others only partially, incorporating some aspects of my approach e.g. my use of sentence builders, L.A.M. (Listening-As-Modelling), Narrow Reading techniques, Retrieval practice tasks, Chunking-aloud games, Fluency-training activities, Universals, L.I.F.T, etc. Of those who have espoused the approach, some have adopted it in its purest form, adhering to the letter of EPI. Others have ‘hybridized’ it, as it were, combining their existing approach or other approaches with mine (e.g. NCELP + EPI) creatively adapting EPI to their learning context.

Despite its popularity and the many blog posts and articles written on EPI, some published in prominent specialised journals such as Applied Linguistics, there exist gross misconceptions about the approach, some of which I was reminded of during several conversations with language educators at a recent conference in Sheffield, the ALL ‘Language World’. There I realised, to my horror, that even some prominent MFL influencers I talked to on the day had some serious misconceptions about the approach and a superficial understanding of it. They seemed to believe that EPI is about teaching random sentences embedded in substitution tables and getting the students to parrot them until they learn them by rote without any input about grammar, SSC (spelling-to-sound correspondance) and the meaning of individual words. In other words, EPI would consist of merely drilling in a set of unanalysed chunks ad nauseam through a range of ludic activities without a coherent instructional plan or rationale. A very prominent lady said: I don’t understand how it is different from what we used to do it the 80s.

Ten misconceptions about EPI

Here are ten of the most common misconceptions about EPI, supposedly the main reasons why language teachers, according to my detractors, should stay away from ‘contification’:

1. In E.P.I. we don’t teach grammar

2. In E.P.I. we don’t teach phonics

3. In E.P.I. we don’t teach the meaning of single words. The method is about learning unanalysed chunks by rote. Meaningless parroting

4. In E.P.I. talking is merely about read-aloud games a la ‘Sentence stealer’ and oral translation drills such as ‘No snakes no ladders’

5. E.P.I is slow. The coverage is unambitious

6. E.P.I. is about memorising paragraphs with parallel texts

7. In E.P.I. listening and reading comprehensions are banned

8. E.P.I. cannot be used with high-ability classes

9. In E.P.I. we don’t teach culture

10. E.P.I. doesn’t prepare for National Examinations

The reasons for these misconceptions are manyfold. One reason is that quite a few of the blog posts/webinars on EPI floating on the web were written/delivered by people who provided their own understanding or adaptation of my approach and in some cases have never attended any of my trainings (!). Another reason refers to the deliberate attempt by certain people and entities to misrepresent EPI as anti-grammar, anti-phonics, anti-culture, anti-retrieval practice, anti-creativity, and anti-  everything else that teachers hold as the untouchable pilllars of language teaching and learning – the aim: to put them off the approach.

Debunking the myths about EPI

In a series of posts – hopefully only two – I intend to debunk every single one of the above myths about EPI. For reasons of space, in the current post I will concern myself only with the top three on the above list.

#1. In E.P.I. grammar is not taught

Nothing could be farther from the truth. Grammar is taught at many points in a typical MARS EARS sequence, both explicitly and implicitly. The learning of grammar, however, unlike what happens traditionally, is not an end in itself; rather it is intended as a way to increase the generative power of the target chunks, i.e.: it aims at enabling the learners to skillfully manipulate the constructions modelled in a given unit of work to suit a communicative purpose. Moreover, it is subordinate to communication and lexis.

So, for instance, if in sub-unit 1 (“Talking about what I did yesterday’) of a unit of work on “Talking about the recent past”, the students have learnt the construction Time marker (e.g. yesterday) + I + perfect tense form + noun or/prepositional phrase, we may decide to teach the full conjugation of the verb ‘Avoir’, or parts of it, so as to enhance the communicative power of the target chunk (i.e. making it applicable to a greater range of agents). Thus, the grammar content is derived opportunistically from the target communicative context and not imposed top-down by the curriculum designer as a list of items to tick off. Figure 1, below, illustrate this point.

Figure 1: the selection of the linguistic content of a typical unit of work

Grammar teaching starts explicitly in the initial Modelling/Awareness-raising phases (the M & A in MARS), in which the teacher sensitizes the learners to one or more of the morpho-syntactic features which underpin the target construction. At this stage, we are not talking of a full-fledged grammar explanation but rather of ‘pop-up grammar’, a short capsule of language awareness, which draws attention to specific features in the input the teacher wants the learner to notice.

During the Receptive phase too (the R in MARS), grammar is learnt, implicitly, through input-flooding, input-enhancement (see figure 2 below), thorough processing and repeated processing at every level of grain (phonemic, syllabic, lexical, morpho-syntactic, pragmatic and semantic) of the target structure. The tasks used in this phase aim at inducing the so-called ‘structural (or syntactic) priming effect” first documented by Bock (1986).

Figure 2: definition of Sructural or Syntactic priming

Input-flooding and input-enhancement work in synergy in an attempt to draw the learner attention to the target linguistic features. In the example in figure 3 below, for instance, the text is flooded with French perfect tense and imperfect forms which have been highlighted in yellow and blue.

Figure 4: An example of input enhancement

Grammar is also learnt explicitly, through activities which deliberately direct the learner’s attention to the target features (see figures 5 to 8 below). For instance, in a lesson on ‘Talking about last week-end’ in French, the auxiliares AVOIR and ETRE may be gapped from a text which needs to be filled in by the students as part of a partial dictation. Or, as part of a ‘Faulty transcription’ task, the students may be asked to correct wrong instances of the perfect tense occuring in a text similar to the one used in the partial dictation. In a ‘Tracking’ listening activity, the students may be asked to note down how many instances of AVOIR and ETRE they hear in an aural text flooded with perfect tenses read aloud at moderate pace.

Figure 5: sample ‘Spot the pattern’ activity

Figure 6: Sample Parsing-grid dictation
Figure 7: sample sorting and parsing tasks

Figure 8 : ‘Word hunt’ a sample grammar-focused reading activity

Editing activities (e.g. ‘Editing Carousel’ or ‘Best writing’) may also be staged where working individually or in groups, the students would try to spot and correct errors. Whilst the activities above focus on form, focus-on meaning activities will be staged too, including semantic processing activities based on the principles of (1) elaboration  (2) distinctiveness (3) appropriateness to retrieval and application (4) relevance to personal experience.

Figure 9: the four principles underlying deep-processing tasks

During the Structured production phase (the S in MARS), pairwork retrieval practice and communicative activities elicit production of the target structure. Another short pop-up grammar session may take place at this stage to activate prior knowledge, clarify misconceptions and prime the students for the use of the target structure during this phase, which would ideally last two lessons. After each substantive retrieval practice or communicative task, short bouts of pop-up grammar may be staged to point out common mistakes the teacher noticed during the performance of the tasks.

In the Expansion or Explanation phase (the E in EARS), the target grammar feature is explained more thoroughly. Since, by this stage, the target structure has been processed many times over and the teacher will have already directed the students’ attention to the target feature several times through activation of prior knowledge, corrective feedback and informal assessment, often s/he will be able to teach it through guided-discovery techniques or even through full-fledged inductive tasks. In this phase, grammar and translation drills may be staged, where the learning condition is task-essentialness, i.e.: the target feature is necessary in order to complete the tasks-at-hand (e.g. a gap-fill task requiring the retrieval of an appropriate perfect tense form). It is worth noting that, with lower ability groups, with whom grammatical accuracy is not a priority, one may skip this phase and devote this lesson to consolidation of the target lexis. With higher ability students, a transfer-practice component maybe added in, whereby the structure-at-hand is practised in linguistic contexts and/or tasks different from the one(s) in which it was initially learnt. For instance, if adjectival agreement is being taught in the context of describing clothes using colours, the transfer-practice task could involve applying the agreement rules in other vocabulary domains (e.g. the one learnt in previous units on describing classroom objects, people animals, etc.).

In the Autonomous recall stage (the A in EARS), I always recommend embedding a short grammar assessment task, e.g. an Editing task involving identifying and fixing errors pertaining to the target structures or a translation or othe retrieval-practice task in which the target structure is task-essential. The aim is to test whether the students have grasped how the target structure ‘works’ declaratively and can apply it in familiar tasks.

In the Routinization phase (the R in EARS), form-focused tasks such as ‘Fast & Furious’ or ‘Fixy Echo’ will explicitly focus the students on the target grammar feature, whilst meaning-focused tasks such as ‘The 4,3,2 technique’, ‘Market Place’, ‘Speed Dating’, ‘Five’, ‘All for one one for all’ etc. may be used to elicit the rapid retrieval of the target feature in a communicative context.

Finally, in the Spontaneity phase (the S in EARS), unplanned communicative tasks may be staged in a bid to elicit the deployment of the target structure amongst others (e.g. under timed conditions, the students may be shown a story-board and asked to describe it in the past tense or they may be interviewed on what they did last week, etc.)

As you can see from the above, grammar instruction is woven into every phase of the MARS EARS sequence. How’s the EPI approach to grammar instruction different from the current methodology promoted by the DfE in England?. Here are some key differences:

  • Grammar is not the end goal of instruction but a tool to support the achievement of spontaneity and creativity with the language: it serves a communicative purpose. As picture 3 above shows, it is selected based on the Communicative functions we teach and, more precisely, on the construction(s) which convey that function. For example, in teaching the function “Describing what one did last weekend” in French, I may choose to teach the construction: Time marker + Perfect tense of Aller + prepositional phrase. C’était + adjective. As a result, I will decide to teach the perfect tense of aller. This means that the target grammar structure emerges organically from the context; it is learnt in context. The grammar explanation occurs as a way to give the target construction more generative power, i.e.: teach the learners how to manipulate it using declarative knowledge so that they have control over it and doesn’t stay a monolytical unanalysed chunk of language. In other words, declarative knowledge is taught in order to build on procedural knowledge.

  • The full-fledged explanation of the key grammar point is delayed to the Explanation phase, which occurs after much receptive and productive practice, on the fourth of fifth lesson. This means that grammar teaching is often about reverse engineering, picking apart what the students already know, which makes learning declarative knowledge easier and in some cases even redundant. Delaying the teaching towards the end of each sub-unit (or MARSEA sequence) means that the grammar explanation that at best puts off, at worst excludes a substantive chunk of a mixed ability class at the very outset of a lesson, comes after four lessons packed with inclusive and fun activities, mostly ludic in nature.

  • The core structures that your students must ‘nail’ by the end of each year or cycle, the ‘non-negotiables’ selected for teaching, what I call ‘Universals’ are not as many as those found in course books such as Dynamo or Studio. They are limited because there is a limit to what can be truly ‘entrenched’ in the ridiculously small time allocation language learning typically gets in primary and secondary schools (1 to 2 hours a week). The criterion for selection? How foundational and key they are in the building of the L2 system. In picture 11 below I have listed my universals for a year 7 French group.

Figure 11 – Examples of Year 7 French universal

#2. In E.P.I. we don’t teach phonics

This is another myth. The teaching of Phonics or SSC (Spelling-to-Sound Correspondence) is also woven into every step of the MARS EARS sequence. As happens with grammar, phonics too are  taught through a synergy of Explicit and Implicit instruction.

Figure 12: In EPI, phonics are taught through a synergy of Explicit and Implicit instruction

The phonemes and syllables the EPI teacher focuses on, though, are derived opportunistically from the vocabulary and grammar you selected, as shown in picture 1 above. For instance, if, as part of the communicative function “Describing people”, you plan to teach words like mère, frère and père, this may prompt you to focus on  the phoneme /ɛ/. Moreover, if, as part of the same topic, you are going to teach the present indicative of the verb Etre (suis, es and est) this may trigger a focus on silent consonants s and t. This means that the target phonemes and phonemes clusters will be constantly recycled multiple times across the entire unit.

One of the advantages of modelling language through sentence builders is that the learners hear and see the words being presented simultaneously. Hence, each word the students are taught is concurrently encoded in both its phonemic and graphemic form. Thus, phonics teaching starts implicitly from the get-go, in the Modelling phase.

In the awareness-raising phase, through activities such as “Faulty echo”, “Spot the silent letters”, “Rhyming pairs”, “Write it as you hear it”, “Spot the foreign sound” and many others described in Conti and Smith (2019), the teacher draws attention to specific sounds known to be problematic for the students (e.g. silent consonants in French, nasal sounds, etc.). These noticing activities are used routinely by EPI teachers and are the staples of this initial phase. Input enhancement techniques, both visual (e.g. highlighting silent letters) and acoustic (e.g. stressing specific sounds) are also used in this phase as awareness-raising tools.

Figure 13: Input enhancement applied to the teaching of SSC (phonics)

In the ensuing Receptive processing phase, listening and reading work in synergy to further reinforce SSC. This is done through a number of engaging ‘Scripted listening’ (listening whilst reading) activities, such as ‘Word Bingo’, ‘Sentence Bingo’, ‘Break the flow’, ‘Spot the missing detail’, ‘Spot the intruder’, ‘Listening puzzle’, ‘Slalom listening’, ‘Jigsaw listening’, etc. These aural tasks elicit dual processing, i.e.: the students simultaneously process aural and written input. For instance, in ‘Spot the intruder’, the learner is given a written text and must identify any words contained in that text, which are not read aloud by the teacher. Most EPI scripted-listening activities are designed to promote ‘thorough processing’, i.e. force the students to pay close attention to every single word in the transcript. Thorough processing means that, if you have flooded the input with the target graphemes/phonemes, the chances of the students learning SSC are likely to be multiplied.

Dictations are also commonly used in EPI, occasionally in synergy with Scripted listening. For instance, in ‘Faulty transcript”, the students need to identify and note down the differences between what they listen to and the corresponding transcript; for example, the text they see might say “Me llamo Paco” whereas the text the teacher reads out would say “Me llamo Juan”. Some of the dictation tasks I use in this phase are detailed in this post. Whilst Scripted listening and Dictation tasks build a strong SSC implicitly, explicit SSC-focused episodes can still be embedded through corrective feedback (e.g. on dictations) or pre-task activation knowledge (e.g. prior to a “Track the sound” task, where the sound to be tracked is /ɛ/ the students may be reminded explicitly of the relevant SSC declarative knowledge in synergy with physical awareness).

Figure 14: Chunking aloud consists of oral-pairwork games in which sentences or texts flooded with the target sentence patterns are repeated many times over to induce the structural priming effect first documented by Bock (1986). Chunking aloud games are very inclusive and foster phonological-encoding and articulatory fluency

In the Structured Production phase, prior to the Chunking aloud games typically staged with beginner to pre-intermediate classes, another short pop-up phonics session may occur to sensitize and prime the students, in which you would stage phonological awareness classics such as ‘Minimal pairs’, ‘Phonemes bingo’, ‘Contrast and response’, ‘Rhyming pairs’ and others detailed in this post  . During each chunking-aloud game you will of course walk around monitoring student output, correcting when necessary and making mental notes of the most common mistakes. After each game, you will use the so-gathered observational data to provide whole-class feedback on their decoding skills, before the students proceed to play the next activity. For some EPI classic chunking aloud games, follow this link.

Figure 15: Sentence Stealer is the most popular chunking-aloud game I have created. Tip: use post-its instead of cards and ask your students to stick them on their mini-whiteboard.

In the Fluency-training phase, tasks and games aimed at speeding up accurate production of the target sounds (e.g. “Chain reading”, tongue twisters, etc.) are staged.

Figure 16: Chain reading

Besides the activities typical of each phase I have just described, in EPI many other techniques and initiatives are carried out which cut through the whole MARS EARS cycle, aimed at promoting alertness to sound, physical awareness, critical listening and other dimensions of sound-related metacognition.

As you can see, just like Grammar teaching, Phonics instruction is pervasive but not overly explicit in EPI; and, because the SSC focus stems opportunistically from the vocabulary and grammatical content of each unit, extensive recycling throughout the MARS EARS sequence is guaranteed.

#3. E.P.I. doesn’t teach the meaning of single words. The target construction are taught as unanalysed chunks.

This is another preposterous misrepresentation. Anyone vaguely familiar with sentence builders knows that every L2 word in the sentence builder is translated in the L1. In fact, in order to make sure that the L1-to-L2 meaning mapping is as unambiguous as possible, I encourage the use of literal translation. For instance: ‘J’ai besoin d’argent’ would be translated in a typical sentence builder as ‘I have need of money’ (instead of ‘I need money’) or ‘J’ai onze ans’ as ‘I have eleven years’ (instead of ‘I am eleven). Hence, from the get-go the students are fully aware of what each constituent of a target construction means. This approach is taken in the brilliant EPI-based website www.sentencebuilders.com as evidenced by the example in Figure 17, below.

Figure 17: In sentence builders incorrect, literal translation (aka ‘dodgy translation) is often used to raise awareness of crucial differences between languages and pre-empt misgivings arising from the correct translation. In the sample sentence builder above (from http://www.sentencebuilders.com), the verb ‘have’ is used to translate in English the verb ‘avoir’ in the context of telling one’s age to pre-empt the assumption that ‘Avoir’ in English means ‘to be.

Furthermore, plenty of vocabulary building activities used in EPI in the Receptive and Productive phase elicit focus on single words. These include: (1) traditional vocabulary-building activities such as ‘Gap-fill’ tasks, ‘Odd one out’, ‘Categories’, ‘Find the near synonym’, ‘Match L1 and L2 equivalents’ etc. as well as EPI classics such as (2) Sentence puzzles with L1 translation, ‘Find the L2 equivalent in the text’ ‘Gapped translation’, ‘Faulty translation’, Tangled translation’, Word-substitution, etc.

In EPI, the target vocabulary is selected based on three principles:

  • Relevance to the students. Research shows unequivocably that when the target vocabulary is perceived by L2 learners as relevant, it is more likely to be successfully acquired.

  • High frequency. Vocabulary which is high frequent is more likely to be useful, as the first 2,000 most frequent words in a language give access to at least 80 % of any generic L2 text.

  • High surrender value. Vocabulary which is useful in the learning of other vocabulary or even grammar structures should be obviously prioritised. For instance, ‘aller’ has high surrender value, as it is the necessary pre-requisite for the learning of the Immediate future in French. Learning ‘Mettre’ paves the way for the learning of Promettre, Admettre, Sousmettre,etc. High-frequency vocabulary often has higher surrender value, so if one applies criterion (2), one partially satisfies this criterion too.

Figure 18: The Language Gym ‘Vocab trainer’ teaches vocabulary across three skills implementing the principles and task-types discussed above


Many misconceptions are being circulated in UK MFL circles by entities and people who are either misinformed or have a vested interest in portraying EPI as an anti-grammar and anti-phonics approach whereby language learners are fed unanalysed chunks of language whose meaning they learn ‘holistically’, without truly grasping the meaning of each individual lexical item they contain nor the underlying grammar that glues them together.  Easy to understand why: grammar, vocabulary and phonics are considered these days by OFSTED as the ‘three pillars of progression’, the key areas, that is, which school inspectors are going to investigate when they visit schools in order to assess teaching and learning. Hence, the message is clear: you will fail OFSTED if you embrace EPI. This is, of course, not true: one can teach EPI and still show ‘progression’ in all the above areas.  

In this post, I have attempted to demonstrate that in EPI both grammar and phonics are practised extensively through a powerful synergy of implicit and explicit learning. All the greatest Applied Linguistics theorists and researchers would agree that this synergy is key to successful learning and that implicit (or procedural) knowledge is what teaching should concern itself mostly with. As Ellis and Shintani (2013) posit: “ Instruction needs to be predominantly directed at developing implicit knowledge of the second language while not neglecting explicit knowledge“.

As for the notion that EPI is about the teaching of unanalysed formulaic chunks of language, I hope I have shown that is not the case at all – although, of course, some chunks have the potential to stay unanalysed (e.g. ‘Il y a’ in French or ‘Es gibt’ in German). The meaning of and usage of individual words is modelled and practised in context at all times, faithful to the notion (central to EPI) that ‘you will know a word by the company it keeps’ (Firth, 1957).

In fact, one advantage that EPI has over approaches like the one championed by NCELP, is that, not just the target vocabulary, but also the target phonics and grammar are selected as tools which enable the learner to fulfill a communicative purpose and are seamlessly and organically integrated in the linguistic and even cultural fabric of each unit of work, which allows for abundant meaningful recycling. Take the NCELP approach instead: each lesson consists of two completely disjointed sections, one which deals with explicit phonics instruction following the PPP sequence and one which deals with grammar and vocabulary. In other words, phonics instruction occurs in isolation, as a self-standing episode. Hard to see the logic of such an approach. In addition, the words are not selected based on a unifying theme, but pretty randomly. Moreover, the linguistic content has been selected by the NCELP curriculum designers top-down and without any apparent guiding principle – apart from high frequency for the target lexical items. The result is a random and decontextualised list of phonics, words and grammar rules for the students to regurgitate. Add to this the fact that the guiding principle for vocabulary selection – high frequency – flouts the all-important ‘relevance-to-the-learner principle in that the corpora used for the selection of the target lexical items include mainly texts intended for adults – not adolescents (e.g. European Commission or Parliament documents and newspapers).

In the next post I will concern myself with the remainder of the misrepresentations on that list.

If you want to know more about my approach, do get hold of my books, co-authored with Steve Smith: “Breaking the sound barrier: teaching learners how to listen” and “Memory: what every language teacher should know”.