In March 2008 the Department of Education in New Brunswick announced sweeping changes to the public education system affecting French second-language edu- cation for all anglophone students in the province. The changes seek to bring at least 70 percent of the children in the province to the intermediate level of French proficiency, while improving their scores in literacy, mathematics and science.

Currently, 19 percent of the children enrolled in anglo- phone schools follow a program of early French immersion (EFI) that begins in grade 1; 7 percent follow a late French immersion (LFI) program that begins in grade 6; while 74 per- cent follow a core English program that entails about 30 min- utes of daily instruction in French in grades 1 to 5, and 40 to 50 minutes in grades 6 to 12. The proposed universal French program would offer an intensive French instruction program for all anglophone students in grade 5.

In this new program students would spend about 80 p e rcent of the school day over a five-month period in lan- guage-rich activities devoted to reading, writing and speak- ing French. In the other one-half of the school year they would receive the regular subjects, including French, with the normal time allotments. When students enter middle school in grade 6, they would elect to follow a late immer- sion program or a core English (CE) program with enhanced French instruction.

There have been critics to the reform arguing that the status quo should be maintained, or that the system should have just two streams beginning in kindergarten: EFI and a core program with increased French instruction.

Many parents, with the support of Canadian Parents for French, have supported the critics in their calls to abandon the reform. In June 2008 a court ruling required the Department of Education to conduct a further six weeks of public consultation. If the proposed reform is suc- cessfully implemented, New Brunswick will become the first province to introduce universal French instruction.

The need for change appears to be based on three assumptions. The first is that the current system is not achieving the desired educational outcomes for all anglo- phone students in the province. The second is that the cur- rent immersion system segregates children by social class background and ability, and this has a detrimental effect on their learning. The third is that the previous approach to French instruc- tion did not succeed in bringing a sig- nificant number of children in the province to the intermediate level of French proficiency. In this article, I critically examine the evidence on these three issues.

Findings from the 2000 Programme for International Student Assess- ment (PISA) revealed that 40 percent of New Brunswick youth fail to achieve at level 3 or higher in their reading skills at age 15. The level among Canadian youth is 28 percent. The international assessment of youth literacy skills con- sists of five levels. Those who score below level 3 have diminished chances of pursuing post-secondary education, while those below level 2 are at risk of not completing secondary school. The results in the 2003 and 2006 PISA assess- ments, which emphasized mathematics and science skills, were similar. Among the 10 provinces’ youth, those in New Brunswick ranked in second-last place in mathematics and in last place in science. Results from the School Achievement Indicators Program (SAIP), an assess- ment program conducted by the Coun- cil of Ministers of Education Canada (CMEC), also suggest that students in New Brunswick do not do as well as their counterparts elsewhere in Canada.

The three other Atlantic provinces also scored poorly on these national and international assess- ments, although literacy scores in Newfoundland and Labrador are slightly higher than those in the three other provinces. Part of this literacy gap is attributable to differences in the socioeconomic backgrounds of families, but even when these factors are taken into account, children in New Brunswick do not fare well.

Arguments about whether EFI or LFI result in a segregated school sys- tem have focused mainly on the virtual exclusion from French immersion programs of children with special needs.

In the anglophone sector on aver- age around 17 percent of children are in special education plans for the whole school year. Data from the Department of Education suggest that only around 7 percent of these chil- dren are in French immersion classes. The class sizes in EFI are slightly smaller than those in CE (19.5 vs. 21.3 students in schools with EFI). If students with special needs were equally distributed among all classes, each teacher would on average have 3.4 students with special needs. However, given the segregation of these students in CE classes, the aver- age in CE is approximately 4.3 stu- dents. The problem is exacerbated, however, in schools that have EFI pro- grams: in these schools the average in CE classes is about 5.7 students. It is this discrepancy that is felt by many teachers in the CE program.

However, the segregation associat- ed with French immersion is much broader and deeper than merely that related to special needs children. There are at least four other dimensions to it.

First we also find segregation along social class lines. I have divided all fam- ilies of children enrolled in NB schools in 2007-08 into five socioeconomic groups, based on their family income, level of education and type of occupation. Each group includes 20 per- cent of the population of students enrolled in EFI and CE in grades 1 to 5.

Figure 1 shows the enrolment rates in EFI and CE by socioeconomic group. Compared with children from the middle socioeconomic group, those from the highest socioeconom- ic group are nearly twice as likely to enrol in EFI. In contrast, those in the lowest socioeconomic group are about half as likely to enrol in EFI. Well over half of all children enrolled in EFI are from the two wealthiest socioeconomic groups.

When one compares socioeconom- ic status of those in EFI to those in CE, the divide is comparable to or larger than the divide between non-Hispanic whites and African-Americans in the US.

Indeed, about 28 percent of CE stu- dents are in the bottom quarter of the socioeconomic distribution, while only 20 percent are from the top quarter. In contrast, only 13 percent of EFI stu- dents are from the bottom quarter of the socioeconomic distribution, while 39 percent are from the top quarter. These results are remarkably similar to the ratios of African-Americans to non- Hispanic whites in the US.

Second, girls are more likely than boys to be enrolled in French immersion. In a typical class of 20 children in EFI, there are on average 11 girls and 9 boys, while the ratio is reversed in CE classes. This is arguably small, but one must also consider that boys are more prone than girls to have difficulty in reading.

Third, there is some segregation according to ability. Parents’ deci- sions to enrol their children in EFI are based in part on their desire for their children to become bilingual, but also on their perception of their children’s likelihood of success in the program. To discern the extent of segregation associated with children’s ability, data on children’s skills prior to grade 1 are required, and the Department of Education has only recently begun to collect these data. However, data for nine New Brunswick schools with EFI programs were collected by the Canadian Research Institute for Social Policy at University of New Brunswick.

Table 1 shows the average scores and standard deviations on the Early Years Evaluation " Teacher Assessment (EYE-TA) for over 900 chil- dren who entered EFI and CE pro- grams in 2004 and 2005. The EYE-TA is a comprehensive assessment designed to assist educators in assess- ing the skills and behaviour of chil- dren aged 4 to 6 as they prepare for and make the transition to school. It assesses five domains of early child development that are closely related to school preparedness and emergent literacy skills. The scale for each domain ranges from 0 to 3.

Table 1 shows the percentage of children in each program who scored below 1 in each domain. These chil- dren are vulnerable in the sense that without a concerted effort on the part of parents and school staff, they are at risk of falling behind their peers in literacy skills.
In all five domains of early development, chil- dren enrolled in EFI have significantly higher scores. Moreover, the proportion of vulnerable children in CE classes is more than twice that in EFI classes. The differences are most pronounced in measures of cognitive and lan- guage skills, which are important predictors of academic success.

Table 1 shows that there is segregation based on behaviour. It shows the percentage of children who had very low scores on the EYE-TA meas- ures of Social Skills, Behaviour and Approaches to Learning. Children with very low scores on this scale typically have problems of inatten- tion, physical aggression, or anxi- ety/ depression. The difference in the proportions of children with low scores in the two programs: 10 per- cent in CE compared with 5.6 per- cent in EFI " is important as it indi- cates that at this age (Kindergarten) the majority of children with behav- iour problems do not enrol in EFI.

A report by Patricia Lee and Jim Croll for the Department of Education found that only about 20 to 25 percent of the students who enrolled in EFI in the mid-1990s persisted in French immersion through to grade 12 and presented for the oral proficiency assessment. Nearly all of the students who completed the p rogram achieved intermediate proficiency, but less than half of them achieved an advanced level or higher. The authors did not have data on the success rate at the end of grade 10 as the oral pro- ficiency assessment was not offered for that grade.

The report has been criticized by the proponents of EFI, especially concerning the way that attrition rates were determined. Croll and Lee examined current enrolment rates for each year, which is common practice among school district administrative staff concerned about class allocations from year to year. They reported an average rate of 16.9 percent for the period 2001 to 2006. Their calculations were correct, but they mistakenly called the differ- ences ”œattrition rates.”

One prominent critic correctly used a longitudinal approach, for example, comparing the number of children that enrolled in grade 1 in 2001 and finished grade 5 in 2006. His average estimate was 23.5 perc e n t , and he argued Croll and Lee under- estimated attrition. Two other critics used a similar approach but compared their estimates for that period to Croll and Lee’s estimates for 1997 to 2002, which is not standard practice. They argued that Croll and Lee over-esti- mated attrition.

However, the correct approach for using longitudinal data to estimate attrition is to take account of move- ment in and out of the program, which none of the analysts did. About one-third of families with young chil- dren in New Brunswick move every five years, but one cannot discern from available data how much of the move- ment is to other schools that have early immersion programs. I would estimate that the real attrition rate over that period is about 20 percent, or about half-way between the Croll and Lee estimate and one of their critics.

The current enrolment rates in EFI, LFI and CE programs are shown in figure 2. Although one cannot dis- cern what the success or persistence rates of students who recently enrolled in French immersion will be in 10 years, the current enrolment pattern supports the findings of Lee and Croll. Around 27 percent of grade 10 students and 18 percent of grade 12 students are currently enrolled in French immersion programs.

Among the EFI students who do persist to grade 12, about one-third achieve a standing of advanced profi- ciency, while the remainder achieve a level of intermediate or intermediate plus. Overall, given the low persistence rate through to grade 12, only 10 per- cent of the cohort that enrols in EFI in grade 1 achieve advanced standing. It would be fair to presume that the major- ity of EFI students who persist in immer- sion up to grade 10 have at least an intermediate level of proficiency. Lee and Croll note that the two main rea- sons cited by students for not complet- ing the immersion program are that they wanted to complete their educa- tion in English to better prepare for studies at an English university, and that they view the last two years of immer- sion as ”œmaintenance” years. The per- sistence rate to grade 12 of LFI students is about twice that of EFI students, but only about 10 percent achieve advanced proficiency.

In 2004, Statistics Canada com- pared the reading literacy results of immersion to non-immersion stu- dents based on the results from PISA 2000. The average score among New Brunswick EFI students (550) was slightly higher than the national average (534), and the same as the average score of all students in Alberta, the top-performing province.

The author attempted to control for ”œselection effects,” using a weak measure of socioeconomic back- ground; however, she was unable to control statistically for the ability of students on entering elementary school. Given the high degree of segre- gation associated with French immersion programs, a study of their effects on students’ academic progress would require a design in which students were randomly assigned into immersion and non- immersion programs.

Results in other provinces revealed a difference in performance favouring immersion students, but as a much smaller proportions of students in these provinces enrol in EFI, the selection effects are likely to be even greater.

The question of the success of EFI has also been framed as being ”œIs learning a second language early better than later?” The answer to this ques- tion is clearly ”œyes.” But can children who begin to learn French at a later age also achieve success? The answer to this question is also clearly yes. Among students who entered LFI, the proportion attaining at least the intermediate level was about 90 per- cent in four of the years between 2002 and 2006. However, a smaller propor- tion " on average about 10 percent " achieved an advanced level. Results for Canada from the International Adult Literacy Survey indicate that adults who immigrate to Canada learn either French or English quite rapidly, especially during the first five years.

The most fundamental choice of parents in a public school system is the right to enrol their children in a school where they can learn with their peers. But school choice is not a right when it has a negative effect on the educational choice for other children, especially those who are most vulnera- ble. And this is the perverse effect that early French immersion is having in New Brunswick.

In fact, in every country that participated in PISA, the socioeconomic com- position of schools had a significant effect on student achievement. When children with lower ability or children from lower socioeconomic groups are concentrated in particular schools or classes, they tend to perform worse than when they are in mixed ability classes. Children from higher socioeconomic groups tend to do well in any setting. Consequently, countries that practise early streaming, such as Germany and Austria, tend to have a disproportionate number of children with very low PISA scores, while countries without stream- ing, such as Finland, tend to have supe- rior scores. Many countries that practise early streaming are attempting to over- haul their school system to delay streaming until the later stages of sec- ondaryschool.

Our results clearly show that the streaming problem associated with early French immersion cannot be simply fixed by adding more resources. If I were to design the cur- rent reform I would maintain some French instruction during the ele- mentary school period and offer a second intensive French program in grade 10. I would also want written and oral proficiency testing con- ducted at the end of grades 5, 8 and 10, with formal certificates awarded as students reached successive levels. I would also propose having only one program during the middle school years, thereby delaying any direct form of streaming until sec- ondary school.

For the children who would oth- erwise have enrolled in EFI this fall, there is little risk of them not succeed- ing if the proposed reform is imple- mented: they are a select group to begin with; they will have a more solid base in their first language at the end of grade 4; they will receive the intensive French program in grade 5 prior to enrolling in LFI; the school culture will be stronger during the elementary and middle school years; and they will have a more intensive program of French instruction during secondary school. The proportion achieving advanced oral proficiency may in fact be higher.

As for those students who would have enrolled in the core English pro- gram next fall, there is no contest. Under the new system, many of these students will achieve intermediate proficiency in French, and some may pursue advanced standing. Even if the new program is only partially successful in achieving the Department of Education’s goals, there will be a dramatic increase in the overall proportion of students achieving inter- mediate standing in French.

But the broader question is whether the changes in the program will lead to an improvement in the literacy skills of all children and a decrease in the proportion of children who fail to make the transition from learning-to- read to reading-to-learn. For most chil- dren this occurs by the end of grade 3. I have no doubt that the reform would be successful in these terms; there are many examples of reforms where eliminating or delaying streaming resulted in improved academic performance.

In other provinces there is a small- er proportion of students enrolled in EFI, and while the problems associated with selection are evident, they are not as severe as in New Brunswick. Howev- er, if the New Brunswick reform suc- cessfully brings the majority of its population to the intermediate or advanced level of oral and written pro- ficiency in a second language, educa- tional leaders will undoubtedly follow New Brunswick’s lead.

Will it be enough? The develop- ment of literacy skills requires quality teaching at the appropriate level, a positive classroom environment, ade- quate learning time, and the engage- ment of students and their families.

Data from PISA and other studies suggest that New Brunswick has a well-qualified and dedicated teaching f o rce. However, the amount of learn- ing time children receive day-to-day and over the course of the school year deserves attention. Also, many children in New Brunswick face chal- lenges associated with poverty, and they therefore need extra resources, especially during the primary school years. Some children will need a sec- ond-chance program if they are still off track at the end of grade 2. The bold move to offer universal French instruction needs to be accompanied by other reforms aimed at improving the literacy skills of children during elementary school. The success of New Brunswick children, especially those who are most vulnerable, will also need the continued support of parents from all walks of life.

 

This research was supported by funding from the Social Sciences and Humanities Research Council through its Canada Research Chairs program.