Results over three years

In order to measure the effects of the pedagogical approach on the children, in 2011 the association Agir Pour l'École agreed to fund a rigorous and ongoing evaluation of the experiment. The purpose of the chosen tests was to measure: attention span, working memory (by measuring memory span), conceptual reasoning (in terms of classification and induction), knowledge of the letters of the alphabet (by testing their ability to vocalise the letters), the accuracy of their hand-eye coordination, their abilities in arithmetic, and reading; oral expression and lexical level (in describing images), their ability to infer a concept and their general culture (through guessing games), and their awareness of phonology (by breaking down syllables into phonemes); and finally, their graphomotor and visuoperceptive abilities (through the reproduction of geometrical figures).

It was envisaged that these tests would be conducted each year over a three-year period. Unfortunately, the evaluation was abandoned after the first year for two reasons:

The first was that it proved very difficult to establish a formal monitoring framework at the institutional level. This lack of a formal framework prompted the Ministry of Education to discontinue the tests at the end of the first year of research.

Methodologically speaking, it was moreover difficult to measure the progress of children in reading and knowledge of the alphabet system, because the aforementioned rigorous tests were designed to measure the children's ability to read in block letters and to recognise the letters by their name. However, the children in the Gennevilliers class began reading in cursive (recognition of block letters coming later) and knew the letters by their sound. The names of the letters was given to them later on, once reading had become automatic. In order to measure the reading and letter-recognition abilities of the children, it was therefore necessary to adapt the aforementioned tests by changing the text of certain tests to cursive and asking the children what the sound of the letters was rather than what they were called. Due to the very nature of the pedagogical approach used in the class - initiation to reading through the sound of the letters and not their name, and focusing on cursive before block letters - we were thus confronted with methodological difficulties: there are no standard formal tests for measuring the ability of such young children to give the sound of letters (rather than their name) and to read in cursive rather than block letters. Over the following two years, the children were able to take these tests because their reading had become automatic and they were reading block letters; however, the tests which they took were the standard tests for older children. In addition, we would have needed to include a control class in this study, which we were not in a position to do.

For these two reasons – institutional and methodological – formal monitoring was conducted during the first year, but was not extended into the second and third years. These institutional and methodological constraints also led the association Agir Pour l’École to withdraw its support for us. Over the following two years, we therefore arranged for tests to be conducted with the help of an independent psychologist and subsequently the Laboratoire Unicog, headed by Stanislas Dehaene, but this was done outside school hours with the support of the parents. Under such conditions (outside school hours), only about fifteen children were able to take part in the monitoring over the following two years. As a result, although the tests are indicative of a very positive trend, the small number of children who took them places limits on how representative they can be considered to be. In order to take things further, one would need to conduct a longitudinal study involving a larger number of children and a control group.

Pending the development of tests which are adapted to this type of approach, and the testing of a larger sample of children in the future, it is nevertheless important to bear in mind that the results for these three years, which are set out below and supplemented by parent feedback, were highly positive: the children, despite being from an area which the French government has designated as a zone of falling education standards and violence in schools, blossomed to a striking extent: they developed significant social skills and were able to read and count; they were in cheerful spirits and enjoyed going to school and learning.

Year 1 • 2011-12

The monitoring of this first year was therefore organised by Agir Pour l’École: the tests were selected by a psychologist and technical expert contracted to the Laboratoire de Psychologie et Neurocognition, which is affiliated to the CNRS*; the testing was conducted by independent psychologists, and the data was analysed by the association. In November 2011, a pre-test was carried out to measure the children's level in all the areas listed above, and a post-test was conducted in June 2012 on these same areas in order to measure progress.

Extracts of test reports, produced by the association Agir Pour l’École

'All of the pupils, with one exception, are making faster than normal progress, and many have made very significant progress. The pupil who is not making faster than normal progress has had the highest number of absences over the year.

'On average, in terms of phonological awareness, the children are very far above the norm.'

'57% of second-year children (aged 4) exceeded the January assessment score of pupils in the preparatory class (aged 6).'

'The accuracy of hand-eye coordination appears to substantially ahead of the norm.'

'In terms of short-term memory, progress is very substantial.'

During this first year, the children embarked on the learning process in a striking way: nine of the fifteen second-year children and one first-year child began to read. The tests indicated that these children were already one year (or indeed two years) ahead in terms of reading. The extremely encouraging results of this first year nevertheless demonstrate that one area – vocabulary acquisition – needs to be attended to, since it appears to be lagging behind in comparison to other areas.

During this first year, one 3-year-old child decided to learn to read on their own initiative in order to imitate the 4-year-old children in the class. We would nevertheless like to stress that our approach did not involve asking the children to read, but rather enabling them to do so when they felt the urge: some felt the urge at the age of 3, others at the age of 5.

For their part, the families quickly noticed that their children were better able to concentrate, were more autonomous, interacted more harmoniously with others and showed greater self-discipline than before. Their children were happy to go to school and felt a greater sense of self-confidence and fulfilment.

Year 2 • 2012-13

During this second year we included a third age level : all three nursery school ages were therefore brought together. The children helped each other and collaborated more, and their autonomy continued to increase rapidly. All the oldest children, as well as 90% of second-year children, began reading: some fifty albums had to be borrowed from the library every fortnight to cater to the children's desire to read, and they would take one or several books home every evening. The frequency with which they read resulted in an improvement in their level of language. In addition, the children continued to assimilate major mathematical concepts (the concept of adding one, positional notation) with enthusiasm and ease, as well as cultural components of Geography, Geometry and Music.

As mentioned previously, for this second year we conducted a new series of tests, but this time outside school hours. Only about fifteen children were able to take these tests and so, although the test results indicated a very positive trend, the small number of children tested placed limits on how representative the tests could be considered to be. In order to confirm the positive results of this pedagogical approach, such tests would be need to be extended to a larger number of children. The tests were selected by Manuela Piazza and Karla Monzalvo from the Laboratoire Unicog, whose director is Stanislas Dehaene. These tests focused on mathematics and reading. The holding of the tests and their analysis was carried out by a Doctor of Cognitive Psychology.

Extract of the data analysis:

Reading comprehension

The results achieved by the children in the text comprehension test are astonishing. Although the test is not standard for their schooling level, they nevertheless managed to perform as well as the average of the benchmark group. In addition, these results were achieved despite the fact that half of the available points should have put them at a disadvantage. All of the children tested fell within a standard minus or plus one range of the average. In other words, they demonstrated a comprehension of the text they had just read which was at least as good as the average 7 to 8-year-old pupil.

Reading conclusions

Both third-year and second-year children had already successfully begun to learn to read, as evidenced by their text comprehension level and their general reading level. The prerequisites of learning to read have been mastered, as evidenced by the results achieved in the test which assessed their skill in manipulating the graphophonemic code and in tests on phonological processing.

Complete and unified representation of the numerical code

Given the results of the children tested, it is possible to assert that they all possess a solid representation of the numerical code. Only two children did not obtain the maximum score in the two tests. The oral numerical choice test was passed by all third-year children and by one second-year child, despite the fact that this test is standard for 8 to 9-year-old children. The children who obtained a score of 12/12 in this test not only achieved the best results in their age category, but also in the 8 to 9-year-old category.

Comparison of numbers

Once again, we are led to acknowledge that all the children responded brilliantly to these two tests, indicating an astonishing mastery of numerical quantities for children of their age.

Operations and problem-solving

Simple operations carried out within a narrative framework were performed with remarkable success by all the children. Five children obtained the maximum score in this test, and the other two children were in the upper half of their age category. The pupils who obtained the maximum score have the same level as end-of-year 6 to 7-year-olds, which is to say that they are one year ahead.

Mathematics conclusions

As with the results observed in the reading tests, the performance in arithmetic of the children in the Gennevilliers class is excellent. The analysis of the results reveals a sound command of the numerical code and a very satisfactory semantic understanding of number and of operations.

The children achieved results which generally rank them among the best pupils in their age category. We have chosen to present the children's results by comparing them as far as is possible to the benchmark group for their age category. It nevertheless transpires that certain children obtained scores which exceeded the upper limit of their age category. This phenomenon was virtually systematic in the numerical choice and number comparison tests. When the children obtained maximum scores for their age category, they often attained the level of end-of-year 6 to 7-year-olds. Three of the third-year children sometimes even ranked among the best of the 8 to 9-year-old pupils.

General conclusion

What has emerged is that in the two essential areas of school learning, reading and arithmetic, the children in this class display skills which often exceed their stage of schooling. This observation is especially true when it comes to the children's reading skills. Contrary to what is expected of 6-year-olds, all the children in our group can be considered to be readers. The only child who cannot decipher a text is only in their second year and can nevertheless decipher the letters. The children's skills in arithmetic are equally surprising. Once again, they surpass the expectations which we might have had of them. It must be acknowledged that all the children are at least one year ahead of what is expected of them.'

Year 3 • 2013-14

In September 2013, the children in the class who could read underwent MRI scans in Saclay (NeuroSpin), organised by the Laboratoire Unicog. The purpose of these tests was to establish whether beginning to read before the age of 6 to 7 had an impact on the wiring of the neural circuits associated with reading. The scans are still being analysed, but the data so far would seem to indicate:

a neural rewiring associated with reading which is entirely normal,
a reading level one-and-a-half years in advance of their age.

However, although the tests are indicative of a very positive trend, the small number of children who took them places limits on how representative they can be considered to be. In order to take things further, one would need to conduct more systematic monitoring of a larger number of children and include a control group.

The tests are discussed beginning at 16'20 min.

* An error went unnoticed in “The Natural Laws of Children” (p.18) which suggested that the tests of this first year 'were conducted by the CNRS in Grenoble'. This inaccuracy will be corrected in the next edition: the tests were selected by a psychologist and technical expert contracted to the Laboratoire de Psychologie et Neurocognition, which is affiliated to the CNRS, but the testing was conducted by independent psychologists and the data was analysed by the association Agir Pour l’École.