Unpicking NCELP’s Marketing

Since the launch of the finalised new GCSE content in January 2022, colleagues at NCELP have been very much on the charm offensive, extolling the virtues (or so they see them) of said new GCSE. Lots of this has been random quotes in colourful speech bubbles from various teachers. Who knows whether this effusive praise reflects the real balance of opinion of those who have been on their training courses – after all, we know NCELP at al like to be very selective when it comes to evidence (see this fairly crushing editorial from the highly respected Language Learning Journal for more on this).

Anyway, more insightful than the various fluorescent eulogies are the samples of work that NCELP have been posting: these are pieces of work from NCELP schools. I thought I’d offer a few comments on one of them. Here is the tweet:

I found this particular post interesting because it reminded me of one of the first blogs I ever wrote – which was also a discussion a piece of writing by a Year 8 student. It prompted me to reflect, at the time, on where we might be going wrong in the teaching and learning of MLs in England.

What does this marketing tell us about the NCELP approach?

The piece of work shared by NCELP as part of their twitter marketing

But first, let’s be clear about one thing: none of this critique is aimed at the school nor the pupil nor their teacher. The purpose of this blog is to offer critical reflection on the outputs and marketing of a State-funded pedagogical centre of excellence. This is the sort of discussion that any subject should be able to have, when taxpayer funds are used to finance a particular pedagocial doctrine. All of what follows is directed at NCELP, who are the architects of the approach that this school, teacher, and pupil are all adopting. If you don’t like a building, you criticise the architect, not what the inhabitants have done to try and make a home out of it.

Grammar, grammar, grammar: at the expense of meaning

Critics of the new GCSE and of the NCELP method take issue with the dismembered, grammar-heavy approach it takes. This is very much in evidence in this particular text. Apparently, the lesson during which this writing took place was at the start of year 8, after the week spent on “talking about jobs”. This does indeed feature at the start of Year 8 in the NCELP scheme of work (snip below).

The NCELP Year 8 Term 1 Week 3 scheme of work

Interestingly, the writing task which the student appears to have done does not actually feature in the NCELP lesson plan. There is a writing task at the end of the sequence in the NCELP lesson, but it is this and goodness knows what students are actually supposed to write. No wonder the school in question did a more sensible task!

The writing task as per the NCELP official lesson – replaced by the school for a more natural piece

It’s worth noting two things about the student’s writing. The first thing to pause upon is the quality criteria being used to judge the work. The student’s task was to say what they might want to do when they’re older. Yet this is the quality criteria:

I am not why “more difficult connectives”, “varied intensifiers” or “past tense” are remotely necessary to answer the question what do you want to do when you’re older? This is a sign of the grammar-tail wagging the language-dog: language being judged by its grammatical gymnastics, not on whether it achieves its purpose and communicates what it is supposed to say.

What’s also interesting is what is not featured on the quality criteria. Here is a translation of what the student has written:

Q. What do you want to do when you’re older?

A: Hello! My name is XYZ. I love French with Mrs XYZ because she is fun and funny. my friends are also vrey funny. we like to eat at the restaurant and café. I want to be a lawyer because it’s interesting. school is also interesting sometimes it’s bad, sometimes it’s good! the headteacher is really hardworking and nice. Goodbye!

The teacher – and let me be clear, I am not criticising this teacher, I am criticising the architects of the teaching approach, i.e. NCELP – says that this piece of work is “super”. In my opinion, it is not super, it is weird, and it is a bad piece of writing, regardless of what language it is in.

Normally, a hallmark of good use of language is that it is cogent and makes sense. But there is no requirement for an actual decent piece of writing in the criteria – hence we get this fairly odd, quite repetitious and only partially relevant response. It’s also very interesting that the teacher is correcting relatively unimportant things – like the adjectival agreement on ‘amusant’ – but not correcting more fundamental aspects of the writing such as capital letters and punctuation. The teacher appears genuinely delighted with the adjectival agreements (this is what the lesson sought to teach): the student gets two of these right, but one of them wrong. Interesting that despite an apparent (?) focus on intensifiers, the student still omits the accent from très, writing it as tres. These are perfectly normal mistakes for a year 8 to make, but the NCELP grammartastic approach has not eliminated them. Interestingly, despite the heavy focus on phonics, there is also no accent on the e in ecole or interessant. It’s like, sometimes you say something explicitly lots and lots, and the student still struggles to apply it in real life. Err yeah, we knew that 40 years ago, and the NCELP hyperexplicit method won’t change that.

Students should not be praised for writing texts like this, in my view. Despite its two correct adjective agreements, the text is rambling, irrelevant and content-free. But this student did their best with the limited words at their disposal – as we shall see in what follows.

Vocab own goals

Here I have rewritten the student’s text, but omitted the words which are not in the top 2000 words in French.

Bonjour je m’appelle [name]. J’XXXX le français avec Mme [name] car elle est très XXXX et XXXX. Mes amis sont vraiment XXXX aussi. Nous aimons manger au XXXX et café. Je veux être avocat car c’est intéressant! L’école est aussi intéressante parfois c’est mauvais, parfois c’est bon. Le directeur est vraiment travailleur et XXXX.

For theis passage, the student uses 35 headwords. These are:

bonjour, je, me, appeler, adorer, le, français, avec, Madame, car, elle, très, amusant, drôle, mon, ami, vraiment, aussi, nous, aimer, manger, à, restaurant, café, vouloir, avocat, ce, intéressant, école, parfois, mauvais, bon, directeur, travailleur, sympa.

Of these 35 headwords:

  • four are pronouns (je, me, nous, elle)
  • two are articles (le, ce)
  • two are prepositions (avec, à)
  • two are conjunctions (et, car)

Interestingly, despite the high frequency focus of the y7 NCELP Scheme of Work, I can’t see a single word here that a normal Y7 textbook or course doesn’t already teach. We already do teach high frequency words – pronouns, core adjectives, conjunctions for example- because you can’t build sentences without them. But we also teach other stuff with which we can say specifics.

Overall, four of the headwords – or 11% of the text – are low frequency words. This text would not have met, therefore, the criteria which was originally proposed – namely for vocab lists to be limited to 10% infrequent vocabulary.

There are, though, a few funny words in these text which aren’t quite as high frequency as they might look: travailleur comes in at 1341 in the frequency list, because it is also often used as a noun, meaning worker (quite distinct from its meaning as an adjective), and because of the political/economic emphasis of the corpus used to create the frequency list, travailleur comes up rather a lot as a noun. This is why avocat also makes it as a high frequency word: the only others Bonjour only just makes the cut: it comes in at 1,972, after ordonner and before morale (lol).

It is interesting that the student (we presume he’s male because avocat was not corrected to avocate) says he wants to be a lawyer. This might, of course, be because the student wants to be a lawyer. Or it might be because it is one of the very few words for professions that the student knows – despite having just done a unit of work entitled “talking about jobs”. Indeed, the only words that the unit introduces for jobs are the following:

In fairness, there are a few other words from NCELP’s Year 7 scheme of work that pupils could also use for jobs:
Why does NCELP only teach 8 professions within and leading up to the unit on “talking about jobs”? The answer, of course, is that there aren’t many high-frequency words to denote professions – certainly not useful or plausible ones for 11-12 year olds. Here is the full list from the frequency dictionary – and you’ll be curious to spot that facteur makes it. This is of course because facteur has at least two meanings. Most often, it means factor, not mailperson.
Lonsdale & Le Bras (2009) A Frequency Dictionary of French, New York: Routledge, p86

In the past, of course, “talking about jobs” was taught rather differently, in that it included a variety of appropriate jobs to talk about. My 3rd edition OUP French Vocab Book contains a list featuring 35 professions – and that is just for Foundation Tier. It has everything from médecin to maçon, femme d’affaires to fermier, comptable to caissier. Why? Was it trying to overload students with oodles of unmanageable words in random order? No, it was trying to teach students words they might need or, indeed, want. Even my widely used and respected “Mastering Arabic” textbook contains 7 words for professions – in chapter 2, before students have even learned the whole alphabet or conjugated a verb. Why? Because the emphasis is on being able to use the language, not just fetishise its structural characteristics

Comparing Pre- and Post-NCELP

So the two pieces of writing – the one analysed above from an NCELP school and the one I critiqued in my blog a few years ago – might not be directly comparable. One student might “have a higher target grade” (to use the NCELP parlance from their tweet) than the other. One might be at the end of year 8, rather than the start. One pupil have had longer to write the text than the other. But broadly, these two pieces are comparable: they are both Year 8 French pupils in state schools taught by jobbing teachers like you and me. And they tell us about what has been taught.

NCELP exists because Government decided, in 2016, that MFL teaching was dreadful, that bad teaching was the root cause of all ills in MFL. So they created NCELP to improve teaching. You’d expect, therefore, the text from the NCELP school to be a lot better.

The text from the 2018 blog

But that just is not the case. Much though I critiqued the thinking behind the piece in my original blog, it is a lot better. And I’m not talking about the grammar. It’s better because it has content. It says things. It expresses specifics and has meaning. There are so many things the student is able to talk about, because he/she is not limited to high frequency structure words: cuisine, escargots, relaxant, maillot de bain, lunettes, maquillage, BD, hôtel de cinq étoiles, luxueux, frisbée. You get a sense of the text from those words alone. It tells you what the student has written about. And the student was able to use lots, and lots, and lots of words.

Conclusions

The sample of student work from one of NCELP’s flagship schools confirms what we fear about their approach:

  • it prioritises certain aspects of grammar over expression of meaning
  • it fetishes accuracy and structure
  • it artificially and significantly limits vocabulary, and limits what students can say, and will inevitably frustrate learners, therefore
  • NCELP’s own vocab dogma means its lessons can’t achieve what they set out to achieve
  • the current GCSE probably gives us a higher standard of attainment than the new one will

Responding to DfE’s latest statement about MFL reform

On 24th November, in an unprecedented move, nine major organisations (three exam boards, three subject associations and three headteachers’s organisations) joined forces to call on DfE to recognise the overwhelming consensus in the professional community that they need to rethink their plans to reform MFL GCSE. The problems with the proposals are legion, but principally relate to their narrow focus on a reductive and lop-sided word list, at the expense of communication and culture. You can see the full statement here.

The story was covered in TES and the Guardian. DfE eventually provided a response. TES covered it in their story here:

I just wanted to go through this statement with a fine-toothed comb, so here we go.

1. You cannot make a subject more “well-rounded” by reducing its content, in this case to a mere 1200 words for the average learner. In fact, the real figure, once you account for some of the technical aspects of that word list, is more like 1000 lemmas. That is an astonishingly small vocabulary for a language qualification. Middle-attaining students in German high schools (Realschule) learning French aim for more than double this.

2. You don’t make languages more “accessible” via an ill-conceived word list and a grammar-heavy curriculum. You make languages accessible by ensuring every pupil has sufficient classroom time for MFL and through a focus on communication & culture. Note that the more natural, communication-based elements of the current exam- such as the conversation task – will be removed, if the proposals are implemented, and replaced by dictations and read-aloud tasks.

If DfE really cared about the accessibility of MFL, then it would be taking a much more muscular stance on two issues: teacher supply and curriculum time. On the first issue, DfE still doesn’t provide training bursaries for MFL which are as generous as other shortage subjects. On the second issue of curriculum time, DfE is conspicuously silent. Curriculum time is the single most important variable when it comes to proficiency outcomes in language learning. No whizz-bang pedagogy or politically favourable grammar-led curriculum will ever make up for a time-poor curriculum. To learn a language, you need to learn words and you need time to practice them. DfE likes to pretend otherwise. Ofsted’s highly flawed and widely criticised MFL research review was also silent on this critical issue, reflecting the Inspectorate’s remarkably cosy relationship with the DfE curriculum team.

3. This brings me to my next point. DfE have given up pretending that its MFL proposals were based on proper research. They are now saying that the proposals are based on “Ofsted research”. Of course, Ofsted hasn’t done any research on this stuff. They did a so-called research report (see this link), but this of course was never peer reviewed, and chooses to gloss over all the research which doesn’t reflect current DfE thinking. I’ve also started exploring the research which is included in the review, and I’ve found some irregularities: for example, the research review asserted that Studies show that pupils’ self-efficacy consistently results in academic achievement more than other motivational factors. It also improves their language proficiencyStudies show that pupils’ self-efficacy consistently results in academic achievement more than other motivational factors. It also improves their language proficiency. Crucially, though, the references that Ofsted provides to back up this claim do not actually support this claim. No wonder the report was not peer reviewed. I wonder if the research community will review Ofsted’s work anyway? This is exactly what has been going on in the Maths community – and it makes v painful reading for Ofsted.

Ofsted itself seems to find it hard to follow the advice in its own research. The research – like the proposals – talks about how important it is to focus on so-called “high frequency words”. But in the model lesson given as an example by the Lead Inspector for MFL, nearly all the words used are … not high-frequency.

4. DfE asserts, in its response, that students benefit from learning “the building blocks of language […] vocabulary, phonics and grammar”. This is a bit like saying “students benefit from breathing air”. DfE seem to be trying to imply that teachers who reject their proposals somehow reject the idea that language learners should learn words and grammar. Nothing could be further from the truth. If you’re not learning words and grammar, you’re not learning a language. But the key issue is the balance of words, phonics and grammar, and how you then assess them after 5 years.

5. DfE finishes its response by saying “Our proposals aim to increase pupils’ motivation through this approach”. This is fine as an aim, but there is no evidence at all that they will achieve this aim. The approach espoused by these proposals – the NCELP approach – has indeed cost over £4.5m but of course has not been evaluated, and there is no data whatsoever on the motivational impact of this approach. So DfE’s statement here is not even conjecture or hypothesis, it is just hopelessly naive wishful thinking, as MFL teachers have now been saying for months.

What’s also very revealing about this final comment is that they describe their proposed GCSE exam as an “approach”. This makes it clearer than ever that DfE’s real aim here is not to change what the exam looks like – but to change how we teach languages.

What next?

I hope that DfE means what it says when it talks about “working with professional bodies”. Professional bodies have, of course, been excluded from the DfE’s thinking so far. The panel who designed these proposals was narrow and largely hand-picked. Input from professional bodies was absent. So it would be something of a euphemism to say that DfE might “continue” to work with those bodies, but if they would like to start to do so, that would indeed be very welcome, and serve to benefit language learners right across the country.

Are the current MFL GCSE exams unfair?

Fairness is an emotive issue in education, and rightly so – because we all want the best for our learners, and I suspect very many (if not all of us) see education as a means for helping the world become a slightly better place. So whether exams are fair or not is a big issue.

I’ve blogged/tweeted/talked about the issue of fairness a lot before – particularly in relation to the types of topics which currently occur in GCSE, which arguably favour middle class learners in affluent communities. NALA did some good work on this last year – see this link.

In this blog, I’m going to consider fairness from a different angle though, namely vocabulary and the extent to which it is specified by exam boards.

What is the claim?

The claim is as follows. Current GCSE exams are unfair because

  • although they have vocabulary lists (like this one), they don’t include all the words which will be in the exam
  • therefore, students will come across words in the exam which aren’t in the vocab list
  • students might, therefore, come across words which they have not seen in lessons (assuming that lessons are currently delivered according to vocab lists, which I believe is not the case, because the most common textbooks don’t tie in with these vocab lists anyway)

According to be NCELP, who appear to be Government’s de facto policymakers for MFL, this is unfair because (source) the students who “are most likely to succeed” are “very literate students with a broad, sophisticated vocabulary in English, together with a wide general knowledge of the world”. It goes on to say that this is “facilitated by more affluent homes with books, newspapers, discussion at mealtimes etc…”.

I certainly sympathise with the feeling we all have when we see exam questions – and a word we haven’t taught comes up. “They can’t possibly have got that right because they’ve never even seen that word”. Frustrating, sure.

What’s interesting here is that the new MFL GCSE proposals – with their lists of 1700 words and the requirement that exams only use words on these lists – can therefore be presented as a sort of Boris-Johnson-esque attempt to “level up” the languages curriculum, and stop all those metropolitan élite students with the broad vocabularies nabbing the top grades.

In short, therefore, the accusation is that the vocab lists aren’t clear enough, and this means that students who know more words – in the TL or in English – are going to get higher grades than students who know fewer words.

So what’s the problem?

Seems fair enough, right? Having a vocabulary list would make it fairer? Except it won’t. For lots of reasons – which relate to 1) grading, 2) how vocabulary is learned, 3) linguistic proficiency and 4) overall outcomes.

  1. Grading. We have to remember that the mark a student gets and the grade they get aren’t the same thing. In theory, having a specified vocab list might make it easier to get higher marks, because a student might not lose marks for not knowing words. But the exam board is only allowed to give a certain number of grade 9s, 8s, 7s, 6s etc. So that means the board will have to find other things for students to get wrong. Having a set word list won’t make the exam easier, it will mean they are just differently hard. And given that vocabulary is the main feature of a language, that’s a tricky task for exam boards. They’ll probably have to revert to testing grammar (you know those weird grammatical traps they set? yes, those…. more of them!) instead. The methodology used for assigning and distributing grades is not changing in this reform.
  2. The learning of vocabulary. The reality is – and research bears this out – that we don’t learn all the words we encounter – even with the best teachers, the best materials and the best recycling and chunking and retrieval practice and whatever else. We must not treat vocabulary like content in any other subject like Geography or History. This is for two reasons. Firstly, words in a foreign language are not new entirely content, they are new sound-form correspondences which we map onto our existing L1 vocabulary networks in our brain (see the esteemed Heather Hilton explaining this here, skip to minute 22); i.e. we are applying new words (spellings & sounds) to existing concepts (in L1), with some adjustment where necessary. Secondly, we are not just learning what a word is, like a fact such as the birth date of a historical figure, we are learning to use that word receptively and productively, which of course includes all the knowledge of its part of speech and morphology and collocation and lots more besides. Together, these mean that we can’t and should not treat word learning in the same way we teach factual/content learning in other subjects. In fact, research (Milton 2009, Vassiliu 2001) suggests that we can probably expect average learners to learn about half of the words we teach, if we are lucky.

    So, going back to the levelling-up agenda: if we teach 1700 words, as is proposed, our learners might learn about 900 of them on average. 1100 if we think our pedagogy and curriculum is really amazing. Either way, students will encounter words in the exam that they don’t know. So we’re back to where we started – students with a wider vocabulary will still do better. If anything, the ones with the lower vocabulary will be worse off, because we will have taught them fewer words than we teach at present.
  3. Linguistic proficiency. The statement by NCELP about who gets the highest grades in languages really is pretty remarkable. They are bemoaning the fact that students with a rich vocabulary do well in language exams. Take a moment to read that again: They are bemoaning the fact that students with a rich vocabulary do well in language exams. Let’s be clear: vocabulary is the single most important factor in driving proficiency (even NCELP themselves say that), so of course students with a rich vocabulary should do well. Our job is to make sure that everyone has as rich a vocabulary as possible.

    What Government has done, of course, is to reduce the vocabulary requirements of GCSE to 1200/1700 words. That’s the equivalent of wanting students to do better in History, so just requiring them to learn less History and then pretend that students are better off. It’s a lie. Because if they learn less History, they learn less History, and are less knowledgeable about History, so aren’t as well qualified as others who went before them. Reducing the standard doesn’t make it fairer, it just dumbs it down, and means that everyone gets to learn less. To be clear, it’s not levelling-up, it’s levelling-down, or, more accurately, a race to the bottom.
  4. Overall outcomes. Let’s just remind ourselves of the many reasons we teach languages. It isn’t just about helping kids with their chitchat on holiday. We all know and agree that a languages education is about cultural decentring and about turbocharging literacy and communicative effectiveness. If we reduce the standard by reducing the number of words, then all we do is deny our learners the very benefits that language-learning should bring. NCELP are right to recognise that some students in our schools have a limited vocabulary in English. They are wrong, though, to abandon all attempts to address that. If it is true that some students have a limited L1 vocabulary, then our job is to nourish and grow it. Because we all know – and there is TONNES of evidence for this – that strong literacy skills (i.e. knowing words and being able to use them well) are an arbiter of good outcomes in all sorts of different areas of life.

    Unfair isn’t the absence of vocabulary lists; unfair is the decision to allow students to progress through school with a low level of lexical knowledge. We would be doing our learners a massive disservice.

    NCELP and Ofsted’s claim is, of course, that learners should achieve “mastery” of a smaller number of words rather than learning new words. This is total misleading bullshit (sorry). No amount of grammar can replace words that a learner doesn’t know. Not only that, but all the evidence shows us that we only achieve mastery of words (depth) as and when we also develop our overall vocabulary: in other words, depth, breadth and fluency develop in relation to one another, not independently of one another. It’s the same old “less is more” myth. I say myth because it is just that: myth.

    Think of it like learning to play the violin. We don’t insist on ‘mastery’ of bowing before you’re allowed to play a piece. Your bowing improves, alongside your tuning and vibrato and everything else, the more your expand your repertoire. They all feed off each other. Same is true of football: you don’t wait until you’ve “mastered” kicking before you have a go at a match or try out a corner kick. Yet that is what is being proposed for MFL: achieve “mastery” in a small vocabulary before you learn to actually communicate.

So if the list approach isn’t fair either, what should we do?

The truth is that the current approach is not quite as unfair as it is being presented to be. It is being presented as unfair because this is a convenient way of Government blaming others (i.e. predecessor curriculum designers) for its own successive policy failings. Readers of my blog will know that I’m not here to defend exam boards – so hear me out.

Let’s first be clear what the current approach to vocab specification is. It has two parts:

a) a vocab list, which as discussed, is not “exhaustive” of what will be in the exam

b) a topic list, which tells us broadly what students will be expected to communicate about

The reasons that this is not really unfair are as follows:

  1. The example basically functions by “sampling” the vocab that comes up in the list and in the topics. Students who know more vocab are more likely to know the words that come up in the exam. This preserves the importance of learning large quantities of vocab, which is good, because it is vocabulary that makes us proficient.

    Yes that means that there might be a word in the exam that we didn’t teach in our individual schools. But all schools will experience that. It all balances out. And I think it works: rarely, if ever, do we hear complaints that the wrong students are getting the wrong grade (i.e. it’s not the case that our top students are getting 4s while our middling students are getting 8s and 9s); if the sampling approach was misfiring, this is what we would hear. But we don’t: what we hear is that students’ grades overall are low, which is a different issue entirely. They might be low, but they’re usually in the right order.
  2. It means that exam boards can use vocab to distinguish between the best and less good linguists – and not have to revert to grammar or tricksiness. Vocab is the best predictor of linguistic proficiency (see this research and lots more) so vocab is a fair thing to test.
  3. We know, through the topic lists and sample papers and textbooks, pretty much what sort of vocab is expected and appropriate. We know, for GCSE, that vocab relating to biomedical science, graphic design, the French Revolution and postcolonialism are not going to come up. We know that words relating to weather, sports or holidays probably will. We have some flexibility, too, to tailor the exact words to suit our classes, if we wish.
  4. It is one of the (few) ways in which the GCSE is actually a genuine test of language proficiency: in real life, we come across words we don’t know all the time – in our L1, and hugely more so in L2. Being a linguist means being able to manage that. As learners, we will never know as many words in the TL as we do in our L1, so we need to be adept at dealing with this. It would be unfair to pretend that this is not the case: students would have an almighty shock when they arrived in a TL speaking country, and feel rather unprepared.

This is, by the way, how language exams work all over the western world. For all sorts of reasons, governments (except ours) don’t like defined lists: they lead to hoop-jumping in learning, artificiality in communication and restrict teachers’ professional judgements.

So what improvements could be made?

Some things Government could consider would be:

  • a core vocab list of high frequency words, including those which it currently feels are under-taught (this appears to be their concern but they have never said which of these high frequency words students don’t currently know)
  • a guaranteed proportion of words which relate to words on the vocab list

    and most importantly
  • a guaranteed minimum free entitlement to MFL lessons in KS3 and 4, which would make sure that students have the time they need to learn enough words to pass the GCSE. This is what would actually make MFL fairer

This final point is crucial. All this nonsense about using a defined word list to level-up (in fact, level-down) MFL provision is just that – nonsense. And it’s a distraction from some core facts. The way to make sure everyone has a good chance of doing well in languages is to make sure everyone gets enough teaching. There is good research which shows how long it takes students to reach different levels of proficiency: we can measure this stuff, and lots of countries do. In France and most German states (e.g. here) there are rules which make sure that students get enough language provision. So let’s have a national minimum entitlement – and stop this business of pretending we can do amazing things with 90 mins per week because we’re following whatever latest fad methodology – because we can’t. That national minimum entitlement should be of the order of 2.5-3 hours per week. No excuses.

Schools might also consider measuring the vocabulary of their learners at year 7 (in English). Schools where learners have below-average vocab might then apply more time to English and MFL, or indeed other humanities subjects where lexis can be enriched.

Is it beneficial to learn high frequency words first?

I have written before about what high frequency words are and why they are not sufficient for a language-learner who seeks to be able to communicate in real life.

But in this blog I want to analyse the issue differently – and consider the arguments presented by the advocates of a frequency-led vocabulary curriculum.

What is the argument for a frequency approach?

The Government has recently proposed a new MFL GCSE for England in which 90% of the words on the curriculum would be from the top 2000 words in each language. Their justification for this was as follows:

This reflects the messages delivered in various training programmes provided by the Government-funded MFL Centre for Excellence, NCELP. In these materials, it is stated that you get more “bang for your buck” if you teach high frequency words first.

So what’s the alternative?

The alternative to teaching high frequency words first is to do what we do at the moment – to teach words according to what our communication goals are. The table below gives you a flavour of the current loading of our course materials, compared to the proposed GCSE

Proportion of High Frequency words
Studio textbook series49%
A typical AQA French GCSE exam73-74%
The current AQA French vocab list58%
Proposed GCSE vocab90%

The proposed GCSE is the outlier here. The Studio textbook series actually aligns most closely with the advice of one influential researcher, who suggests the following:

So there are broadly two competing visions of how we should structure vocabulary in the languages curriculum:

Model 1: Frequency-first – i.e. the NCELP approach advocated by the Government, whereby the overwhelming majority of words in a curriculum are high-frequency words

Model 2: The 50/50 approach, which is broadly similar to current textbooks and vocabulary lists.

Which model is better?

In order to answer this, I am going to apply two tests.

  1. Which approach gives students the most coverage of a text? This is important because the justification of the frequency-first approach (Model 1) is that it gives students decent coverage of a normal text.
  2. Which approach allows students to communicate?

Which approach gives the best coverage?

You would expect that a frequency-first approach gives learners significantly better coverage of texts than a 50/50 approach. This is not true though – and this might surprise readers. Assuming that high frequency words are taught in more or less the same order, but at different speeds, I modeled the coverage gains of the two approaches and found them to perform almost equally as well. Check out this graph (and see the end of this blog for notes on how I made it):

This chart shows us that students learning via a 50/50 approach, after 15 school terms, have more or less the same coverage of a typical text as those who learn via a frequency-first approach. The difference is less than 2%. How on earth can this be? Well, the answer lies in the detail of what those 2000 high frequency words are. They aren’t all as frequent as each other; the top 100 words account for 50% coverage all by themselves, in fact the most frequent word accounts for 10% on its own. So to get that overall good coverage of 70% or so, most of that is achieved by the top 500 words, not the top 2000. We can conclude that learners who learn via a 50/50 approach – as they do today – are not being short-changed in terms of coverage.
Chart showing cumulative lexical coverage in French

Which approach is the best for communication?

So if the 50/50 model isn’t such a sacrifice in terms of coverage after all, does it have other advantages, which justify this method instead of the frequency-first approach? Yes, absolutely. A I have explained elsewhere on these pages, high frequency words and basic words are not the same thing. Lots of basic words – for the commonest foodstuffs, for the weather, for basic things like saying hi and asking where the toilets are – are actually “less frequent” words. This means that for basic transactions that a learner will expect to be able to understake, low-frequency words are vital.

Moreover, often the most important word in a text might be the least frequent word. Take the sentence j’étais très déprimé hier: the word déprimé is the most important, but it is not a high frequency word. Another example would be this photo I took of a sign in the Paris metro just yesterday, which warns passengers not to buy tickets from touts or peddlers. The most important words here are jamais and sauvette – and the latter is definitely not a high frequency word.

A poster at Gare du Nord, 21/10/2021

This reminds us that coverage and comprehension are not the same thing: just because I understand the high frequency words in a text does not mean that I understand the text. Sometimes, some of the high frequency words might not be adding much meaning at all, because they are particles or structure words. Low frequency words are vital for communication because they are the specific content words that gives a text its meaning. If learners don’t know them, they will be handicapped. We know this from our L1: I’ve been doing some building work on my house recently and have had to learn a tonne of words relating to masonry and construction.

Concluding thoughts

This analysis has shown us that there is no business-case for a frequency-first approach, as proposed in the new GCSE. The improvement in lexical coverage that it yields is, in fact, marginal. Yet the communicative and motivational gains achieved by allowing students to learn infrequent and frequent vocabulary are vast. There are also other reasons why the frequency-first approach doesn’t actually work in practical terms – because of the challenge this stores up for the moment when learners do eventually move onto low-frequency words – but I will write about that another time.

Notes on how I built the graph

Coverage figures used were those from the Lonsdale / Le Bras Frequency Dictionary of French. Assumptions made were as follows:

  1. That students learn 150 word per term in both models
  2. That the high frequency words are learned in the same order, but at different speeds
  3. That the coverage value for words over 2000 is 0.003188% per word. This is the average figure for all words between 2000 and 7500 in the Lonsdale/Le Bras text.

Derivational morphology: what is it? And so what?

In the proposed new GCSE, students will be expected to be able to use certain morphological patterns to understand words grammatically related to ones they already know. Specifically, using French as an example, it says that students will be expected to know:

  1. Adverbs: ending in –ment or –ement only where the English equivalent is -ly
  2. Adjectives: starting in in- only where the English equivalent is ‘un-’, or meaning ‘opposite of’
  3. Verbs: starting in dé- only where the English equivalent is ‘de-’ or meaning ‘not’

Words which follow these patterns will not need to be listed on the vocabulary list. This means that if “libre” is on the list, students will be expected to know “librement“. Ditto, if “efficace” is on the list, they will need to know “ineffecicace”. If “limiter” is on the list, they will be expected to know délimiter.

What does this mean in practice?

Firstly it has implications for the word count. Higher tier students will be expected to know 1700 words; but in reality, 300 of these will be ‘grammatical words’, i.e. words on the list because of the grammatical requirements, such as irregular forms of verbs. But, if you include the words expected to be known via “derivational morphology” along the lines above, that opens up another 200 words or so. About 170 of these words will be adverbs of the -ment/-ement variety. Of course not all of these will end up in the final exam board list: Exam boards need to select about 950/1400 words from the top 2000, and that means they won’t be including all of the adjectives from which you can form adverbs. They might select 100 or so of those 200 adjectives. This would mean the overall target vocabulary size will be something like: 1700, minus 300, plus 100 = 1500. Very small.

Are there any issues?

It will be slightly odd that 100 or so of the students’ words – more than 10% in most cases – will just be adverb variants of adjectives they know. This doesn’t give them much new to say, so the vocabulary might be argued to be somewhat unbalanced. But there is another problem – and namely that the two of the three ‘derivational morphology’ rules required don’t really make any sense when students are learning so few rules in the first place. Let me explain.

Adjectives: starting in- only where the English equivalent is ‘un-’, or meaning ‘opposite of’

By my rough calculations, there are about 20 words in the top 2000 where you can add ‘in-‘ and get a negated version of the adjective in French. Examples would be actif/inactif; capable/incapable; volontaire/involontaire.

But there are somewhat more numerous examples where of words where the -in rules does not work when limited to words within the top 2000. Specifically it doesn’t work with…

  • words where we might expect it to work, because we can add un- in English, but where we can’t add in- in French, such as réel, naturel, commun, critique, complexe, étonnant, naturel, original (in English we can say unreal, unnatural, uncommon, uncritical, uncomplicated, unexpert, unsurprising, unnatural, unoriginal)
  • words where there is a negated form in French, but where it isn’t formed with in-, such as précis, réel, moral, parfait (these exist as imprécis, irréel, immoral, imparfait)

So while it is fine to teach this in+adjective = negated adjective rule as a rule for receptive understanding, it is highly risky to promote it as a rule for productive use. Because students will end up creating words which do not exist. Remember, in these proposals, the aim is that productive knowledge should be the same as receptive knowledge.

Verbs: starting dé- only where the English equivalent is ‘de-’ or meaning ‘not’

French has this pattern just like English does: construct/deconstruct. Though in English it sometimes turns into ‘dis-‘, such as apply/disapply. So how relevant is it to the top 2000 words?

As far as I can see, not very relevant at all. There are only a handful of top 2000 verbs where you can apply this rule, and they’re not exactly priority verbs for your average grade 3-5 student.

  • faire/faire
  • composter/composer
  • conseiller/conseiller
  • monter/monter
  • limiter/limiter
  • former/former
  • lier/lier
  • charger/charger
  • régler/régler

Of course the exam board has to select 50-70% of the top 2000 for its vocab list, so it’s possible that only three or four would end up on the final list. Yet the exam expects students to have learned this as a rule?

More worryingly, learning this as a rule in such early stages of language acquisition could actually be very risky. Because there is a long list of verbs in the top 2000 where you can add de-, but where the meaning is not “de” or “not”. I think these verbs where the rule doesn’t work outnumbers the verbs where it does work in the top 2000 by roughly 2 to 1. For example:

  • tenir / détenir (détenir does not mean “not hold”, it means “detain”)
  • porter / déporter (déporter does not mean “not carry” or “decarry”, it means “deport”
  • montrer / démontrer (etc)
  • noter / dénoter
  • compter / décompter
  • laisser / délaisser
  • poser / déposer
  • passer / dépasser
  • terminer / déterminer
  • finir / définir
  • penser / dépenser
  • priver / dépriver
  • figurer / défigurer
  • marcher / démarche (=noun)
  • céder / décéder
  • nier / dénier

Again this rule doesn’t really make sense as one to teach when students have only got a few hundred words under their belt, if that. There’s nothing wrong with pointing this out as patterns which can and might be helpful when understanding a text, but it’s something else entirely to include it as a “high frequency pattern of derivational morphology” when students know so few words: the patterns aren’t necessarily that high frequency at all, and at this stage, might well mislead them.

But it’s another example of the structuralist nature of these proposals: they represent a world view where languages are really about rules first, then words slot in. In reality, languages are about words first.

MFL exams overseas

There’s lots of debates about what our new GCSE should look like. I’ve been looking at how MFLs (focussing on Languages Other Than English – “LOTE”) are assessed in different jurisdictions – you can see a summary table below, to which I plan to add more.

What strikes me is how straightforward it all is. In England we’re used to a highly choreographed, highly prescribed, highly engineered system with specific task types and very specific mark schemes. These leads to lots of “teaching to the exam task” such as “here is how to do a 16 marker” or lessons on “describing a photocard“; we also over-emphasise opinions and reasons because it’s precisely what exams require. Meanwhile we find ourselves teaching our students not to read and comprehend, but to spot distractors and traps. Does it work? No. Do we like it? No. Does it motivate students? No. Elsewhere, students appear free to get on with learning the language and using it to meaningful effect. There’s also much less obsession with individual points of grammar, and a greater focus on the overall picture, coherence and cogency.

Another key difference is revealed in the mark schemes – which remind us that elsewhere, grades are criterion-referenced (e.g. a certain number of marks or a certain level of skill gets you a certain grade or outcome), whereas our exams are significantly cohort-referenced (the grade you get depends on how well you did compared to other people).

CountryLanguage assessedExamPaperLink
FranceSpanishBaccalauréat – second modern language (Science stream) = B1 level, between GCSE amd ASReading & Writinghere
FranceGermanBaccalauréat – second modern language (Science stream) = B1 level, between GCSE amd ASReading & Writinghere. Mark scheme (inc writing) here
GermanyFrenchRealschule B1 (=Foundation GCSE)Reading Writing Speaking ListeningOverall info and example questions here   Audio samples and sample papers here   Speaking & writing markscheme
IrelandFrenchJunior Cert (has now been replaced) (= GCSE ish) Question paper here   Audio file here   Mark scheme here

If you have got links to other papers from other jurisdictions, please contact me and we’ll add them to this grid.

How many words is 1700 words?

It is proposed that in GCSE, students will be expected to learn 1700 words at Higher Tier. How does that compare? Well, it is less than the 2100 words that Realschule (equivalent of Foundation Tier) students learning French in Germany (for more comparisons, see this post)

But I also wanted to compare this vocab list with O-level, because as many have pointed out, this new GCSE does have a lot in common with O-level: lots of grammar testing, a dictation and less (if any) spontaneous output. So bought I.C. Thimann’s “A French Vocabulary for Ordinary Level “, published 1965. It wasn’t a comprehensive list, as this one in the new GCSE will be, nor was it endorsed by the Government as a full reflection of the curriculum. So we don’t know whether it correlates to what teachers taught. But it should serve as a decent indication.

O-level versus GCSE

Before we go much further it’s worth looking at some evidence from Milton (2008) which compared O-level learners with GCSE learners: are we as proficient in MFL today at the upper levels of attainment as we were in the days of O-level?

Milton tested the vocabulary of 21 individuals who had studied O-level French between 25 and 45 years earlier, but who had not continued learning French in any way since then (even in evening classes). They had scored a mix of Grades 1 (the top grade) to Grade 9, which was the greatest degree of failure possible. He compared them with the best 24 students from a cohort of 449 GCSE students in 2006, so the cohorts should be comparable. This is important because we know that O-level was a qualification only tackled by the brightest students (although by the mid 1980s there were more students taking O-levels in MFLs than there are taking GCSE today – and vastly more than than take GCSE at Higher Tier).

This is what he found:

Milton 2008: French vocabulary breadth among learners in the British school and university system: comparing
knowledge over time, French Language Studies 18 (2008), 333-348

A few things are interesting. Firstly, the ‘minimum’ scores are relatively similar – in other words the lowest-scoring participants in each test had roughly the same vocab size – but we must remember that 20-40 years had elapsed for the O-level learners. Secondly, the maximum scores are wildly different. This means that students who studied O-level 20-40 years prior were considerably more proficient than the best higher Tier linguists at GCSE in the year of the study. Today’s highest attaining Higher Tier linguists are presumably no less able and talented than the best students at O-level were, and arguably the curriculum should allow them to make the same degree of progress. But it appears not to. GCSE Higher is a considerable downgrade from O-level. Today’s learners end up being considerably less proficient than in pre-GCSE times.

I should also be clear that this is not a blog about single word teaching versus chunks. I merely using target vocabulary size as an indicator of proficiency. Vocabulary size is, after all, the best indicator of proficiency (Staehr 2008) . How you embed that vocab and encourage its uptake is a different issue. But one thing is certain: the more words you know overall, the better your proficiency in a language.

The purpose of this vocab book

The next interesting thing to look at is the preface to the 1965 vocab handbook. I’ve scanned it in for you at the bottom of this blog. It shows us a few things:

  • the preoccupation for picking the “right” words for learning is not at all new. This concern has been there for a good 50 years.
  • the vocab handbook is not intended to be exhaustive, unlike the new GCSE vocab list. It deliberately excludes “very common” words and words that it expects students to have come across in grammar lessons – such as pronouns and common irregular verbs. So we can assume that the total number of words contained in this book is an understatement of the learning envisaged. It excludes “easily ‘guessable’ items” such as exact cognates – position, importance etc. Again, this is the opposite to what is being proposed in the new GCSE list. I think that the list is an understatement of the learning expected by anything between 300 and 1000 words, having done some quick counting, but I haven’t included these in my totals below.
  • there was already an oral exam in O-level French in1965, as well as translation and composition
  • the author recommends that the words in this book be learned over three years or so – roughly one list per week (there are about 150 lists). That equates to about 5 school years if we assume no learning in the holidays. In the 1960s this was, apparently, deemed reasonable alongside other French “preparation” (i.e. homework).

My analysis and results

I wanted to know how the new GCSE – from a Government famed for rigour in education and promoting knowledge-rich approaches – compares with the old O-level. So I analysed the vocab in the 1965 book and compared it to the list being proposed for the Government’s new MFL GCSE, for teaching from 2023 (1700 words, of which approximately 300 will be pre-determined grammar items, and 90% of the remainder must be words from the top 2000 in word frequency).

What this shows is:

  • Higher tier students will be given a vocab list at least 44% smaller than the vocab list for O-level – and that’s a conservative estimate.
  • This disparity is actually very similar to the difference in attainment observed by Milton: in his study, GCSE learners knew 40% fewer words than their O-level counterparts did 20-40 years post-hoc. This new proposal engineers in a continuous decline in standards.
  • The O-level list was already weighted towards high frequency words – by a considerable margin. The idea of prioritising frequency is not new.
  • BUT the O-level list has roughly 50% high frequency words and 50% low frequency words – precisely the proportions recommended by researchers such as Milton (“An effective textbook is probably going to introduce frequent andinfrequent vocabulary in roughly equal amounts2009)
  • The total number of high frequency (top 2000) words on the list from O-level and the proposed GCSE are very similar: 1200 ish each)
  • But what O-level had was 1000 (at least) additional, extra words, in the lower (beyond the top 2000) frequency bands, which related to thematic and topic words (the ones which allow us to talk about things)
  • Remember that this Thimann book is an under-estimate of what was taught (he excludes cognates, grammar words and very common words), whereas the proposed GCSE list is exhaustive for both productive and receptive purposes.

If you want to see an example exam paper from precisely the year when students might have been using this vocab book, click here.

Current GCSE vs new GCSE

It’s also worth reflecting for a second on whether this GCSE is greater or smaller in content than the current one. Well, one of the panelists who designed it has said on twitter that it represents a “substantial reduction” in content. Is this true?

NCELP have certainly compared the two and suggest that there is a decline in grammar at least. It is hard to say whether the vocabulary expectations have reduced, because the current list is illustrative and not exhaustive. The current GCSE AQA list has about 1150 headwords on it, but it excludes lots of words assumed to be learned at KS3 such as numbers, colours, places around town, family members, and so on. These would presumably add up to several hundred. Bear in mind that the the exams also include words not on the list and not necessarily assumed to be studied, to test inference and deduction skills. So in terms of vocabulary, it probably is a sizeable reduction, given that Higher Tier will contain something around 1400 headwords once you factor in the compulsory grammar listings.

This tallies with the research that has been done looking at how many words students actually ‘learn’. In Annabel David’s study (2008), the average number of words learned by Year 11 was 564, but in Milton’s (2006) study it was 852. These correspond with research that shows in UK classrooms, learners typically uptake 30%-50% of words input during a course. A figure of 800 words learned suggests input was maybe 1600-2000 words – exactly the size of the current AQA lists. If we are now going to be teaching fewer words in the new GCSE, then all the evidence suggests that students will learn fewer words, unless we think there is going to be a significant shift in the quality of teaching to make up for it. There is no reason to suggest this will be the case, unless we believe that UK teachers are ineffective, and I don’t think that’s the case. Why would it be? Why would MFL teachers be any more ineffective than any other teachers?

Conclusions

This new GCSE appears to be the worst of both worlds. The assessment style and teaching approach of O-level (grammar-led with translation trumping spontaneity) with the ever-lower standards of Higher Tier GCSE (which was a 40% reduction compared to O-level, for equivalent students, and which is continuing to be downgraded). This is a Government which is engineering a decline in standards.

The Preface to Thimann’s vocabulary

(Thiman, I, 1959, A French Vocabulary for Ordinary Level, London: Harrap)

Why the GCSE proposals are disproportionately grammar-heavy

There are many, many problems with the GCSE proposals – which I’ve been blogging about for a while (see this summary). The problems relate both to what they were say and how they were drawn up. One of the many problems is the balance between vocabulary and grammar.

The 2016 Pedagogy Review, on which these proposals are based, had already made it quite clear that the these proposals were going to be heavily grammar-led (or “structuralist”). It declared “We use the grammar of a language to say what we wish or need to say” – which is a gross overstatement – because we first and foremost use words, and can communicate a great deal with just words alone. Anyway that’s not today’s debate, and I’m not really anti-grammar anyway (as any of my students would confirm).

The question for this blog is: is there too much grammar in these proposals? The answer is yes, without a shadow of a doubt – far too much grammar relative to the number of words (950 at Foundation and 1400 at Higher, once you allow for the grammatical items which, according to the proposals, will have to be separately listed within the total 1200/1700 words at each tier).

How much grammar is too much grammar might feel like a subjective question – one which just disintegrates into the grammar vs. communication debate, a debate which is as old as language-learning itself. So I’ve opted for a very simple definition. Too much grammar is grammar that you can’t actually use within the words specified. And remember, the main design principle of these proposals is the focus on high-frequency (top 2000) words.

We should bear in mind that some scholars (such as Vivian Cook) suggest that the acquisition of grammar is conditioned by the lexicon – i.e. you have to have lots of words and examples in order to absorb a rule or pattern. Because there are so many and so varied rules in a language you need a large vocabulary to access the examples and to make sense of them in context. This is just like learning vocabulary itself – it’s considered likely that you will need to see a word between 6 and 20 times before you know it. I wonder how many encounters with a grammar rule you need before it makes sense and sticks? Certainly several, with different words each time.

The Evidence

The more I look, the more I see that grammar is overloaded. I’ve based this analysis on the French grammar appendices. I’ll look at German in due course. I’m not a Spanish speaker so if anyone can do this, contact me and we’ll publish a blog!

Please note that I had undercounted some of the examples in the nouns/adjectives section so please read again! Many thanks to Natalie from NCELP for the feedback.

Nouns

The proposals require students to know how to make various types of French noun plural, including those nouns which end -z,al and –eu.

  1. Yet there is one word in the top 2000 ending in -z.
  2. There are only four ending in -eu (feu, jeu, lieu and milieu)
  3. There are six ending in -al (journal, tribunal, animal, hôpital, canal, moral) and various other words listed as nouns but which are far more commonly adjectives (such as occidental and fédéral; to teach these words as nouns within a rule for nouns would be very confusing for secondary pupils). Of those 6, only four-ish might end up on the final list. This is because Boards will produce lists of 1700 words; 300 or so of these are pre-specified grammar items; a further 10% of the reminder will be lower-frequency words, so we’re looking at 1260 of the top 2000 words, judging by what’s in the proposals. Is four words enough to meaningfully learn a rule among so many rules? It’s hard to say, but very possibly not.
  4. There are very few feasible nouns ending in -eau, too (niveau, eau, réseau, bateau, cerveau, bureau, tableauI don’t think we would teach ‘nouveau’ as a noun), so it seems curious that students must learn this rule so early in their career. Bearing in mind that exam boards will have to pick 1260-1400 words from the top 2000, it’s possible that fewer than 5 of them would make the final list. So students would be learning the rule with very few examples indeed. I’d also say that “eaux” is an unusual, but not impossible plural; “cerveaux” is certainly a very unusual plural, because we all only have one brain (and remember the French rule for saying “my pupils’ brains” would be le cerveau de mes élèves, because each pupil has one brain; English doesn’t acknowledge this)

Adjectives

Similarly the proposals require students to learn feminine adjective ending rules for adjectives whose masculine form ends in –x, –en, –er, and –et.

  1. Yet there are only 4 adjectives ending in –et (complet, net, inquiet, concret). It is inconceivable, almost, that the final list of 1400 words would include all of these adjectives, so once again we would be asking students to learn rules with very few examples using words they’re learning.
  2. There are 6 adjectives in the top 2000 ending in -x (nombreux, sérieux, heureux, religieux, précieux and dangereux). Given that the final list will include 1260 of the top 2000 words, we might reasonably assume that only four of these will feature in the final list.
  3. There are 11 adjectives ending ending in -en (moyen, ancien, aérien, européen, canadien, quotidien, israélien, italien, indien, chrétien, irakien), though I wonder how many of these Boards will want on their list and it’s interesting to see how political texts has influenced the top 2000 here (the list was created in 2009 presumably featuring lots of texts about conflicts in the Middle East). I suspect that the frequency of moyen is elevated artificially due to the homonyms le moyen (the means) and la moyenne (the average). Note that while chrétien is a high-frequency word (thus better “bang for buck”), musulman is not (and therefore considered less of a priority. This is not OK).
  4. There are 12 adjectives ending in -er (léger, fier, cher, premier, dernier, étranger, particulier, entier, financier, policier, ouvrier and régulier). This is starting to look like a sensible number (accounting for the curtailed number which will appear in the final list) with which students might absorb the pattern, although ouvrier is more commonly an adjective not a noun. Note here two common words which ‘break’ the phonics rules relating to the sound of the letters -er at the end of the word: cher and fier

Singular demonstrative pronouns

  1. The proposals say that students must learn singular demonstrative pronouns – mien, tien, sien etc. Yet these are themselves not high-frequency words. I.e. the grammar is trumping the vocabulary – we are being told to learn grammatical structures which are less frequently used than the words we’re being told to learn.

Verbs

  1. The proposals specify that we need to learn that we use jouer and faire with DE (the partitive article) when we are talking about musical instruments and sports respectively. Yet again, there are no musical instruments in the top 2000 words, and the only sport in the top 2000 words is lutte (‘wrestling’, which is high frequency because of its primary meaning, ‘fight/struggle’, as in la lutte contre le racisme)
  2. The proposals specify that students must learn certain irregular or pattern changing verbs, but the vocabulary parameters prevent us from really using these verbs. For example:
    – We need to learn boire but the only vocab items relating to drinks in the top 2000 are café and eau. Note that lait, coca, sirop, bière and vin are not high frequency words.
    – The verb manger is mentioned as one whose spelling changes in the stem. It says that this spelling will “not be credit-bearing”, so presumably the expectation here is that we teach this irregularity even though it won’t be credited in the exam. Note that NCELP materials do teach these patterns from what I’ve seen, such as the ç in some forms of commencer). Interestingly, though, relating to the verb manger, the only food items in the top 2000 are poisson and fruit. Déjeuner, dîner, légume, viande, pain, riz, pâtes, salade, repas etc are all low frequency words, not in the top 2000. Why even mention the verb manger if the nature of the vocabulary being proposed precludes things words for what we eat?
    – similar the verb jeter is listed as one with irregular spelling which presumably we are encouraged to teach, but I have not yet found anything in the top 2000 words that I could throw

Please note that the original version of this post said that ‘aujourd’hui’ is a low frequency word. This was incorrect and due to an error in NCELP’s online profiling tool http://www.multilingprofiler.net.

Curiously, in yet another example of astonishing incongruity (astonishing because I really would have expected the “expert panel” to have checked all this out before publishing the proposals – especially because there was a year’s hiatus between them being written and their publication), there is no expectation in the proposals for students of French to be able to understand or use the relative pronoun dont. This is despite the fact that dont is the 74th most frequent word in French according to the Government’s preferred list. They cannot simultaneously claim to be led by frequency data and then ignore that data in so many ways. If this is all about “bang for buck” and frequency, why is dont not on the grammar list?

So what?

Above I have given some concrete examples of the grammatical overloading of these proposals. The grammar is incongruent with the really important content – the words. This means that time and time again, students will be learning rules with no real words with which to apply them, or too few words as examples. This will be catastrophic for motivation. MFL is being turned into some weird puzzle. The GCSE can’t legitimately call itself “GCSE French”. In reality it will be “GCSE French Grammar with a few illustrative words.” I say this as someone who loves teaching grammar – I nerd out on it and always have done since my A-level Latin and GCSE Ancient Greek – but there’s no point whatsoever in teaching grammar before students have sufficient words in which to encounter and deploy it.

Another way of look at the weighting of grammar and vocabulary is this: for Foundation level (and remember the majority of candidates do score grade 5 or below), students will be expected to know 1200 lexical items. 250 or so of these will be grammatical terms, so the real number of ‘headwords’ they’re learning will be lower. And as ever, students won’t learn everything we teach them; in research studies it is suggested that students usually uptake around 50% of the vocabulary they see in a course. But let’s ignore that and be optimistic that students of this GCSE – a “substantial reduction” in content – will learn as many words as in the past. Evidence from UK classrooms tells us that the average number of words learned might be between 564 (David 2008) and 852 (Milton 2006). This is a very very very small number of words – roughly A1 level in the European Framework. This makes GCSE a beginner qualification. Is the amount of grammar in the proposals commensurate to a beginner qualification?

There’s nothing wrong with the grammar in the proposals. It is all correct grammar. But we need more words than is being proposed for it to make sense and be valuable.

The implications

The implications of this grammar-fetishistic approach are, for me, two-fold:

  1. Because there are relatively few words but lots of grammar, the assessment will inevitably be very grammatical; far more so than now. That means that there will be more of the questions which currently irritate and frustrate teachers and learners. Why learn adjective agreements if not even all colours are in the top 2000 words?
  2. Students will be able to communicate in the TL less effectively than at present. Not just because they will be required to learn fewer words (and the fewer words you learn, the less you can say and understand – it’s as simple as that) but because they’ll have spent time learning grammar that they can’t digest, deploy or practice.

What can I do about it?

Write to your MP to tell them how ridiculous these proposals are and demand that a new panel be convened, with full transparency, broader expertise and a more inclusive process. On this page I have provided a draft letter for you to use.

I’d also really encourage all readers to sign this petition – aiming to secure a full redraft of the proposals, and a new panel.

PS

There’s an embarassing typo in the proposals. Presumably the new exams will tolerate spelling mistakes too?

Write to your MP to express your concern about the proposals for the MFL GCSE

As you may know, there are some grave concerns about the GCSE proposals, not just in terms of what they say but also in terms of the process DfE followed to draw them up.

Please consider writing to your MP to share your concerns. You can do so by using this website (be sure to scroll down to find your MP) or this one. I have drafted this text for you to use in your correspondence: you can copy and paste it, if you like:

You can also share your concerns via one of the exam boards, Pearson, in this survey. Responding to this survey will help the Board build an evidence base to oppose the proposals.

Dear INSERT MP NAME

I am aware that the DfE is consulting, until 19 May on a new GCSE in Modern Foreign Languages. The proposal is that we limit exams to a small pre-specified word list based overwhelmingly on so-called high frequency words; at least some of the most important open-ended tasks (such as spontaneous conversation) have therefore been removed.

Despite ostensibly making sense, this is actually a dangerous, risky, and wholly untested proposal. DfE is punting the science angle in its proposals, but in reality ignoring it. It will further weaken MFL in the UK for a variety of reasons:

  • there is nowhere near enough space or credit for natural, spontaneous conversation in the proposals
  • the proposals specify a word list which is too prescriptive, too limiting and inappropriate for young people
  • the proposed content counteracts efforts to diversify, update and decolonise learning within MFL
  • it will cap the attainment of the most capable linguists and undermine KS5 study
  • the prescriptive list will stymie independence and creativity for learners and teachers
  • the content will not prepare learners for real-world communication or contexts
  • teaching in the way envisaged in the proposals this way does not account for the important linguistic differences which exist both spatially, socially and temporally

But most importantly, this approach flies in the face of so much of the research we have about what successful language learning looks like.

In addition, I object to the process whereby DfE has developed these proposals. FOI requests, whose responses are in the public domain, show that just 7 so-called experts designed these proposals, in no more than 6 meetings. Four of these panelists were handpicked by the Minister himself. The conclusions of the review were pre-determined, and had to align with the findings of another – itself not unproblematic – report on pedagogy from 2016. None of the representative professional or subject associations were invited to comment on the proposals with the panel present.

This may seem like a technical issue, but we must act now to prevent further damage to MFL in England by calling on the DfE to withdraw these plans and start again – with a new, more inclusive and transparent content review panel, and a more holistic remit. I would be grateful if you could urgently make this request to the Department for Education and Minister Nick Gibb MP on my behalf.

Yours sincerely,

INSERT YOUR NAME, ADDRESS AND POSTCODE (THIS IS VERY IMPORTANT!!!)

How did the Department for Education come up with these proposals?

All of this material comes from a Freedom of Information request, available here. The proposals themselves are available here. They amount to the biggest change in the secondary MFL Curriculum since 1987 and relate to a subject widely accepted to be in crisis, yet one of crucial importance to post-Brexit Britain.

Key documents

Details on how the panel was appointed and when it met:

The Minister’s decision to create the panel and approve its membership

Terms of Reference of the panel

What does this FOI request show?

The panel

  1. It was convened very hastily – in 8 working days, seemingly for political reasons:
    24th October 2019: Minister Nick Gibb MP pre-selects some names that he wants to see on the panel. Its creation is prompted by Ofqual’s work on GCSE grading in MFL.
    25th October 2019: DfE officials email the Chair (whose name was pre-decided) saying that they have the go-ahead to proceed
    28th October 2019: The Chair replies and agrees to Chair the group
    Intervening Days: Panelists’ names are drawn up
    31st October 2019: A paper is submitted to Nick Gibb MP for him to formally sign-off creation of the panel
    4th November 2019: DfE emails panel members to sign them up and agree confidentiality of the work.
    5th November 2019: The Panel is publicly announced alongside Ofqual’s decision on grading.
    6th November 2019: Purdah period commenced, prior to the 2019 General Election.

    DfE was keen to time the announcement of the panel alongside Ofqual’s work on grading. This is expressed as a wish to “align comms” and explains why the DfE said “[they]’d like to move fairly quickly”. They note that the wording of comms and statements would depend on the purdah period and “grid slots” from Number 10.
  2. Ofsted and Ofqual were present in most meetings, despite not actually being on the panel. None of the subject associations were invited to comment, or invited to the meetings, as participants or observers. In this way, the Association for Language Learning was excluded, as was ISMLA, NALA, the UCML, LAGB, the Chartered Institute of Linguists, the British Council, and the British Academy. National cultural and linguistic associations present in the UK (Goethe Institute, Institut Français, Instituto Cervantes) were excluded, as were organisations representing lesser-taught languages (such as Confucius Institute or the Qatar Foundation International). Professional bodies such as the Chartered College of Teaching, NEU and ASCoL were excluded, too.
  3. It did not meet very often; 4 times as a full panel (Nov 2019, Dec 2019, Jan 2020, March 2020) and twice additionally as a seemingly ‘core’ group (Feb 2020). This was not curtailed by Covid-19. It was only ever intended to meet 6 times.
  4. The Chair noted the “relatively short timescale” and preferred a “smallish” panel of “6-8” people. My comment: this is a small group given the fundamental nature of the proposed changes.
  5. The Chair prefers to discuss the “balance of expertise as a whole” on the phone, not in writing. I wonder why…
  6. The Minister himself asked officials to approach 3 people to join the panel, along with the Chair: Emma Marsden (Director of NCELP), Katrin Kohl (Professor of German Literature) and John Bald, (former Ofsted inspector and blogger for Conservative Home, an “independent consultant”). This means that four of seven panelists were nominated by the Minister.
  7. It is written in black and white that DfE officials were drafting criteria for panel membership AFTER panelists had already been nominated / chosen by the Minister.
  8. The DfE has declined to disclose the rationale for the appointment of individual panel members, citing that to do so might
    – “discourage future broad and innovative policy development”
    – “significantly increase the risk that there will not be an in depth and open discussion at official and Ministerial level which would adversely affect good policy making”
    My comment: If broad and innovative policy development are such priorities, why did the panel have so few members and so few meetings?
  9. The Terms of Reference, not dated, detail criteria for panel membership. Only one criterion had to be met to qualify for the panel. Thousands of teachers would have qualified for this panel according to these criteria. One criterion stands out: “Independent advisor on MFL – who can bring a wider perspective of working with a range of schools and teachers”. It is not explained what “independent” means, but in the subsequently published panel list, John Bald is named as an “independent languages and literacy consultant”. He writes for the Conservative Home blog. The Terms of Reference confirm that panel membership be based on “recommendations” [i.e. not applications]. The Chair was invited to advise on panel members.
  10. DfE asserts confidence that the panel had “the necessary expertise and experience”, enabling the review to be “broad and balanced”, despite the fact that the panel only had 7 members, none of which were full-time classroom teachers.
  11. Attendance at meetings raises questions: In the Terms of Reference, it was stated that “[p]anel members must be able to make the commitment to attend these meetings”. But many panelists were not invited to two of the 6 panel meetings, suggesting the existence of a “core” panel comprising Ian Bauckham and Emma Marsden.
  12. As an aside, the Chair uses insider shorthand for DfE’s Headquarters, which is located in a premises called Sanctuary Buildings in central London. Mr Bauckham refers to this as “SB”.

The proposals

  1. DfE have been sitting on the proposals since 18 March 2020. A year later, on 10 March 2021, the proposals were published, precisely during the time in which teachers are grappling with how to assess work and award grades for the Summer 2021 GCSE/A-level cycle.
  2. It is confirmed that the panel did not consider grading standards or boundaries. In other words, this review has nothing to do with grading, severe grading or any other grading issue, and there is nothing to suggest grading will change.
  3. The outcome of the review was pre-determined before the review started, insofar as
    – the recommendation of topics was out of scope (and the proposals suggest that GCSEs no longer be topic-based)
    – all recommendations were required to be “compatible with the MFL pedagogy review” from 2016
    – a specific task for the Chair was to “assure that all recommendations by the panel about GCSE subject content are aligned to the MFL Pedagogy Review”
    This was also apparent in the emails from October 2019 shared between DfE Officials and the soon-to-be panel Chair, in which DfE officials state that the job of the Chair will be to “ensure that all recommendations from the group align with the your/the [sic] Teaching Schools Council MFL pedagogy review”. Note that the review panel and the Pedagogy Review were chaired by the same person.
  4. Prior to the review commencing, the DfE had already decided what it thought the problems with the existing GCSEs were: “some lower priority or extraneous content is taking pupil and teacher time away from the core study of vocabulary, grammar and phonics and other essential core knowledge” (although it does not elaborate on what this “other essential core knowledge” is)
  5. DfE set out very deliberately to “consider the direct and indirect impact of the subject content on teacher practice“. This is interesting as it appears to confirm teachers’ concerns that these proposals seem designed to influence how they go about their classroom practice
  6. Exam boards saw the draft proposals twice in 2020 – in March and November. They were formally submitted to Ofqual in December 2020.
  7. DfE appear to have an understanding of what “teaching hours” are usually allocated to MFL, as this was part of the overall consideration for the review. To my knowledge, the DfE has never published any guidance of or research on teaching hours for MFL.
  8. The DfE noted the Ofqual finding that low attainment in MFL may relate to “the extent to which content motivates students”. This is confirmed in the Terms of Reference. However, the subsequent proposals for actual content (words, phonics and grammar) do not clarify how the changes will ensure that content is indeed more motivating.
  9. The DfE was keen to identify where there are items in the current GCSEs which are “unreasonably [formatting sic] difficult”. It is not clear what the distinction is between ‘reasonably’ difficult and ‘unreasonably’ difficult.

If, like me, you think that MFL deserves better policymaking, better standards of transparency and above all a proper, expert-led, open review, then please write to your MP. Curriculum design is political, so our response has to be political, too. You should also sign this petition too, calling for a new panel and a new set of proposals.