ARCHIVED – An Examination of the Canadian Language Benchmark Data from the Citizenship Language Survey

Additional correlational and regression analyses

Simple correlational analyses

Additional correlational analyses were carried out to investigate relationships not discussed above. Table 8 provides intercorrelations (Pearson r) among a number of continuous and nominal variables of interest. Among the most noteworthy findings, a significant negative correlation was observed between CLBA scores and age of immigration. This outcome is consistent with findings well-documented in the second language acquisition literature, which show that ultimate proficiency is closely related to age of second language learning (Long, 1990). In fact, participants’ chronological ages (not shown in Table 8) were also significantly correlated with CLBA scores (r = -.258, p < .01), such that older individuals overall tended to have lower proficiency.

It is striking that there is no significant correlation between the number of years in Canada and CLBA scores. In fact, second language research literature indicates that length of residence effects on ultimate language attainment are typically much smaller than age of learning effects (Oyama, 1976). Furthermore, no significant relationship was observed between language training (in months) and CLBA scores. However, it is important to recognize that without a measure of language proficiency on arrival in Canada, it is not possible to assess the overall effect of language training. In addition, dramatic variations in the quality of instruction in language training programs adds considerable complication to the interpretation of these results.

*significant at the p<.05 level
** significant at the p<.01 level

Multiple regression analysis

As an exploratory investigation of the contributions of participant variables to CLBA scores we carried out a multiple regression analysis (pairwise), in which CLBA scores served as the dependent variable and a wide range of predictors were included. The purpose of this analysis was to identify the best combination of variables that would predict CLBA scores. This evaluation must be regarded as preliminary because of some difficulties with the data identified above. However, it may prove useful in providing directions for further work.

Total variance accounted for by the final model (Adjusted R Square) was 41.6%. A list of variables that contributed significantly to the regression is provided in Table 9, along with slope of the regression line, beta coefficients, t- scores and significance levels. LINC training (as compared to other language training), official language training in general, and fee-based training were all tied to lower CLBA scores. It is important not to interpret this outcome as an indication that LINC or other training has a negative impact on language proficiency. This finding is probably a result of the fact that participants who sought such training were likely to have lower or no language proficiency in English on arrival. Education in Canada was positively associated with CLBA scores. Level of education was also a contributor, because, as noted earlier, participants with university or college training scored considerably higher on the CLBA than did most other participants. Age of immigration continues to be negatively correlated with CLBA scores, even when other factors are taken into account. Immigration class was associated with CLBA scores because, as noted above, the independent class achieved higher average CLBA scores than either the family or refugee classes. Toronto residence tended to be associated with lower CLBA scores. This finding is worth investigating in future studies. There are several possible explanations, none of which can be confirmed with this data set. For example, the under-representation of eligible cases for the city of Toronto may have affected the results. Furthermore, because Toronto receives more immigrants than any other location in Canada, the existence of large compatriot communities may have had an impact on language learning. In some cases, it may not have been necessary for individuals to learn an official language in order to obtain employment or services. Finally, the results may be due to a complex interplay of demographic variables that affects Toronto differently than the other cities.

Two significant negative predictors of CLBA scores were ‘East Asian Mother Tongue’ and ‘Southeast Asian Mother Tongue’. Overall, the former group scored significantly lower (M = 5.89) than other major language groups in the analysis. This group comprised speakers of Mandarin, Cantonese, Filipino/Indonesian, Other Chinese, Korean/Japanese, and Vietnamese/Cambodian. However, it should be noted that 69% of this group consisted of speakers of Mandarin, Cantonese, and Other Chinese. In fact, as noted earlier, the Tagalog speakers (from the Filipino/Indonesian grouping) scored significantly higher overall than the Chinese speakers. Participants from Southeast Asian languages (M = 6.41) also were at a disadvantage compared to the other language groups. Of the several Southeast Asian languages represented in the sample, the Tamil-Dravidian language speakers exhibited the lowest CLBA scores (M = 5.99).

Other variables used in the analysis which did not contribute significantly: sex, number of years in Canada, Edmonton residence, Ottawa residence, Vancouver residence, African mother tongue, and Indo-European mother tongue.

• Previous
• 1
• 2
• 3
• 4
• 5 (current)
• 6
• 7
• 8
• Next