Opening up to Native Speaker Norms: The Use of / I / in the Speech of Canadian French Immersion Students

Our study examines the extent to which French immersion students use lax /ɪ/ in the same linguistic context as native speakers of Canadian French. Our results show that the lax variant is vanishingly rare in the speech of immersion students and is used by only a small minority of individuals. This is interpreted as a limitation of French immersion students’ sociolinguistic competence. Within the group of students who do use both variants, we document a positive correlation between female and middle-class students and use of the lax variant and suggest these speakers are generally more sensitive to sociolinguistic variation. A reverse correlation between English cognates and laxing was found. This is taken as evidence that the learning of laxing is lexically mediated.


Introduction
The goal of this study is to examine the extent to which Canadian French immersion students use lax /ɪ/ in the same linguistic context as native speakers. This regional variant, considered socio-stylistically neutral, is used categorically by Québécois Francophones and should therefore be expected to enjoy wide currency in the speech of advanced second language (L2) learners of French in Canada. In addition to considering the overall distribution of variants, we examine the social and linguistics factors that may condition variant choice. Our general results suggest that French immersion students' use of the variable differs greatly from that of their French native speaker (L1) peers since the majority of immersion students make no use of the lax variant in obligatory context. This is therefore yet another case where the sociolinguistic competence of immersion students is markedly different than that of the Canadian Francophones with whom they are compared. Our article begins with a presentation of the French immersion corpus used for the present study and a discussion of the main findings from previous research on this corpus. We then present our hypotheses and results concerning the variable in question in the current study, followed by a general discussion of results.

Theoretical Context
The theoretical framework used in our research is that of Labovian variationist linguistics. This approach makes use of the notion of the linguistic variable, which can generally be defined as two or more ways of saying the same thing. In other words, it examines cases of different linguistic forms (or variants) that are identical in terms of function or meaning, but whose use is influenced by various linguistic and social factors. The ability to use variants according to the different contextual (i.e., social and linguistic) factors is part of native speakers' competence. This same patterning could therefore be expected to be found in the speech of advanced L2 learners.

The Corpus
Our corpus of French immersion students' speech was collected by Mougeon and Nadasdi in 1996 from 41 students who were enrolled in an extended French programme in the Greater Toronto Area. These students were from Grades 9 and 12 and were instructed in French 50% of the time from Grades 5 to 8 and 20% of the time thereafter. Note that these students all came from homes where French was not spoken and that they resided in a community where there were limited opportunities to use French outside the school setting. Furthermore, they had had only limited contacts with L1 speakers of French in a Francophone environment (cf. Mougeon, Nadasdi, & Rehner, 2010). The topics discussed during the hour-long interviews with the students were designed to encourage them to speak at length and cover a range of both formal and informal topics. In short, the interview protocol was constructed following the same standard sociolinguistic protocols used in the research on sociolinguistic variation in the speech of Canadian Francophones.

Previous Results for the Socio-Stylistic Continuum
The primary objective of this project is to examine the sociolinguistic competence of advanced L2 learners in order to determine the extent to which such competence corresponds to L1 speaker norms. In keeping with previous research by Canale and Swain (1980), we are of the view that successful mastery of an L2 involves not only knowledge of grammatical and phonological forms, but also knowledge of formal variants, informal variants and local norms. Mastery of such L1 speaker norms is important for students who lament not being able to interact with L1 speakers in a natural way (Auger, 2002;Genesee, 1978Genesee, , 1981MacFarlane, 2001;Tarone & Swain, 1995;Thibault & Sankoff, 1993). This goal further reflects the fact that the Ontario Ministry of Education (2000) clearly stated that one of the aims of French immersion programmes is to produce students who are able to incorporate colloquialisms and idiomatic expressions into their speech and debate issues both formally and informally. Finally, such competency is important from the perspective L1 speakers. This was evidenced by Segalowitz (1976) in an experimental study of the social-psychological costs associated with the use of too formal a register by L2 speakers when interacting with target language speakers. These latter speakers perceived the L2 learners as too distant and uncooperative.
In order to explore the range of sociolinguistic variants used by French immersion students, we examined variables along a sociolinguistic continuum ranging from marked (vernacular) to hyperformal (bookish) variants (see Mougeon et al., 2010). The full continuum is presented below:

marked > informal > neutral > formal > hyperformal
Marked variants are described as those which are typically stigmatized, such as the use of m'as instead of the formal je vais or informal je vas (all meaning I go) form of the first person periphrastic future (cf. . They are more commonly used in lower social classes and are used somewhat less frequently by L1 speakers than informal variants. It is not expected that immersion students would make frequent use of marked variants because of their low frequency and status in Canadian French. Indeed, previous studies (Mougeon & Rehner, 2001;Nadasdi & McKinnie, 2003;Nadasdi, Mougeon, & Rehner, 2008) have discovered that the French immersion students interviewed for this corpus made very little, if any, use of vernacular variants. Nadasdi, Mougeon, and Rehner (2005) attributed this to the fact that the students, in a classroom setting, had little or no exposure to vernacular variants, whether it was due to their absence in the educational input or to limited contact with L1 speakers (or both).
Many of the variants that are most commonly used by L1 speakers fall into the mildly-marked informal category. Mildly-marked informal variants are commonly found by speakers of all social classes in both formal and informal situations. Such forms are not stigmatized, but they do not conform to the rules of the standard language. The high frequency and lack of stigmatization associated with these types of variants suggest that immersion students should be exposed to them and be able to produce them more readily than marked variants. However, the findings of previous studies indicated that immersion The Canadian Journal of Applied Linguistics, Special Issue: 20, 2 (2017): 18-31 students use informal variants much less frequently than do native speakers (Mougeon et al., 2010;Nadasdi et.al, 2003;Rehner & Mougeon, 1999;Uritescu, Mougeon, Rehner, & Nadasdi, 2004). Informal variants that have near-categorical usage in the L1 speech, such as ne-deletion (99%) or the use of on as the first-person plural pronoun (95%), were used at rates of 27% and 55% respectively by the immersion students (Rehner & Mougeon, 1999;. According to Nadasdi et al. (2005) these usage rates were the highest among the immersion students' use of informal variants; other variants were used even less (e.g., 15% for schwa deletion) or not at all.
Neutral variants are very similar to mildly-marked informal variants in many respects. They are also not stigmatized, and "stand as a default alternative" (Mougeon et al., 2010, p. 9) to other forms. However, unlike mildly-marked informal variants, neutral variants do conform to the standard rules of the language. Examples include the use of auto to mean car rather than the more formal automobile and voiture or socially marked char and machine, and the use of travail to mean paid work instead of the formal emploi and poste or informal job and ouvrage Nadasdi & McKinnie, 2003). In addition to their lack of stigmatization and high frequency, the fact that neutral variants are acceptable in the standard language suggests that immersion students should have few reservations using the variants in this category. The limited number of studies that have examined neutral variants have found that their usage depends on factors such as input frequency, complexity, and similarity of the variant to English forms (see Mougeon et al., 2010). Neutral variants that are (a) found in the input more often, (b) are similar to an English form, and/or (c) are less structurally complex, may be used more frequently by the students than those forms which are less frequent, different from English, and more complex.
Formal variants such as the use of seulement as an expression of restriction meaning only (Mougeon & Rehner, 2001) are characteristic of a more careful style of speech, are typically used infrequently in the semiformal interview context, and follow the rules of the standard language. It has been shown that the immersion students use formal variants at rates much higher than L1 speakers (Mougeon & Rehner, 2001;Nadasdi et al., 2008;Nadasdi & McKinnie, 2003). Such over-use is likely due to the high levels of exposure the students have to the formal variants in the classroom setting.
Finally, hyper-formal variants conform to the rules of standard French, but are rare in spoken discourse and found almost exclusively in the speech of middle-class, educated L1 speakers. Such variants are relatively frequent in the speech of French immersion students (e.g., use of 1 st person plural nous and use of ne before verbs).
It should be noted that while our research programme (cf. Mougeon et al., 2010) considers variation in terms of sociolinguistic variation along a stylistic continuum, some of the variables studied can also be viewed as regional variants. In other words, some variants are simultaneously regional and stylistic variants. This is in fact the case for many variants that occupy the left side of the continuum presented above (i.e., marked, informal, and neutral variants). For example, Nadasdi and McKinnie (2003) examined the use of rester (meaning to live) in the speech of French immersion students. They considered it an informal variant because of its social distribution. However, it is also a default regional variant of Canadian French since it is used 64% of the time by L1 speakers. As we will see, one of the variants in the present study should also be considered a frequent regional variant and should therefore be used frequently by learners of French in Canada (cf. Ringer-Hilfinger, 2012). As outlined in Bachman (1990), both regional and stylistic variants should be taken into account when assessing mastery of sociolinguistic competence.

The Variable
The variable examined in this study is the alternation between /i/ and /ɪ/ in closed, stressed syllables (see Vickerman, 2010). This variable is unique inasmuch as most research on the sociolinguistic competence of French immersion students has examined sociolinguistic variables that involve two variants used by L1 speakers. The present variable is actually an instance of regional allophonic variation in the speech of Canadian Francophones since the variants are in complimentary distribution. More specifically, the "tense" /i/ variant is found categorically in open syllables and in stressed syllables closed by a lengthening consonant (/v/, /z/, /ʁ/ and /ʒ/), for example dit, vivre, disent, rire, and tige, while /ɪ/ is found in stressed syllables closed by other (non-lengthening) consonants, for example site, ride, and prime. Our reason for focusing exclusively on obligatory contexts is that we are interested in determining the extent to which French immersion students' approximate L1 speaker norms in their oral discourse.
It should be pointed out that lax vowels have existed in French since at least the 16th century in certain regions of France (see Dumas & Boulanger, 1982), and should therefore not be attributed to the existence of lax vowels in English. In addition, the lowering of high vowels is in keeping with an existing pattern for the mid-vowels of French (la Loi de position) whereby lower mid-vowels are preferred in closed syllables (not closed by a lengthening consonant). Furthermore, the lax variant has been documented in contemporary varieties spoken in northern France (see Coveney, 2001). That said, it is not inconceivable that the presence of /ɪ/ in English may influence the presence of this variant in the immersion students' French since we have previously documented cases of home language influence (e.g., the prevalence of seulement [only] in the speech of students from a Romance language background). Our analysis of home language use allows us to consider this eventuality.
Note finally that although laxing does occur with all three high vowels, only the front unrounded vowels /i/ and /ɪ/ have been considered for this study. This is because these two vowels are the most different from each other in terms of their acoustic properties (compared to the rounded pairs /y-ʏ/ and /u-ʊ/). That difference makes the /i-ɪ/ distinction more salient and hence it should be easier for learners to perceive than the subtler difference between /u/ and /ʊ/ or the non-English sounds /y/ and /ʏ/. In other words, if students are learning this case of allophonic variation, we expect it to be made first with high, front unrounded vowels.
There is some disagreement as to the correct terminology to use when referring to the vowel sets /i, y, u/ and /ɪ, ʏ, ʊ/. Much of the existing literature dealing with these vowels in Canadian French has referred to the former vowels as "tense" and the latter as "lax." While these terms may be less than exact since the alternation involves tongue position rather than muscular tension, we will continue to use them to conform to conventions of previous research on the distinction.

Socio-Stylistic Status of /ɪ/ in Canadian French
The lax variant /ɪ/ is best characterized as occupying the neutral position of the socio-stylistic continuum previously described. Recall that neutral variants are not sociostylistically marked and conform to the rules of the standard variety. While the notion of "standard" is more all-encompassing when referring to grammatical variables, phonetic variables allow for more variation (cf. Milroy & Milroy, 1999). In the case at hand, the widespread use of the lax variant in Canadian French in a variety of stylistic contexts makes it clear that "neutral" is a more fitting label than is "mildly-marked" (cf. Dumas, 1987;Walker, 1984). In other words, the use of the lax variant in obligatory context conforms to the spoken norm of the variety spoken by the majority of Francophones in Quebec in the majority of situations. It is, consequently, the regional standard in Canada (with the exception of Acadian French).
While neutral variants have been considered in previous research on the sociolinguistic competence of French immersion, the present study is the first to consider a case of phonetic variation involving a neutral variant. The only other studies of phonetic variation that exist involve the (mildly) marked informal variants of schwa deletion and /l/ deletion. Unlike lax /ɪ/, these former variants diverge from written French and can therefore be expected to be used less frequently by French immersion students (and their teachers).

Linguistic Factors
Each occurrence of the variable was coded for the following (potentially) relevant linguistic factors: 1. presence of a tense or lax vowel in English cognates, in order to determine if a case can be made for L1 transfer (e.g., catholique, intensif [Catholic, intensive]); 2. presence of /i/ or /ɪ/ elsewhere in the word (e.g., limites [limits]), to determine any potential effects of vowel harmony; 3. phoneme preceding the target vowel, to determine what effect the phonetic environment of the target phoneme may have on variant choice; and 4. phoneme following the target vowel, to determine what effect the phonetic environment of the target phoneme may have on variant choice.

Hypotheses
Given the neutral status of the lax variant and the fact that it is a high-frequency regional variant, we expect this form to be used often by students. This hypothesis is based on two observations. First, teachers have been shown to make frequent use of neutral variants (e.g., they use the periphrastic future 77% of the time) and also use some mildly marked variants at high levels (e.g., they use first person plural on 83% of the time). Therefore, it can be surmised that the lax variant would have been abundant in the speech to which students have been exposed in the classroom and that they will have learned this variant through the input to which they were exposed. (Note that we are unable to empirically confirm this since there are no recordings of the actual teachers; previous teacher data are limited to Allen, Cummins, Harley, and Swain's 1987 transcribed corpus.) This seems all the more likely given that the Ontario Ministry of Education's (2000) The Canadian Journal of Applied Linguistics, Special Issue: 20, 2 (2017): 18-31 guidelines for the teaching of French immersion emphasized the fact that students should develop familiarity with Canadian French.
Our second reason for predicting frequent use of the lax variant is that previous studies of other neutral variants have found relatively high usage by immersion students, as shown in the following We see here, for example, that the periphrastic future is used in a majority of cases by both Francophones and French immersion students. That said, our study is the first to examine a neutral phonetic variable and it may be that grammatical and lexical neutral variants may be salient and more readily mastered by students.
As for social correlations, it is possible that female and middle-class students will show a preference for the tense variant since it could be viewed as more standard in a scholastic setting. Still, the neutral status and great prevalence of the lax variant in Canadian French may trump this potential influence. Another social factor that could prove relevant is students' home language. More specifically, we hypothesize that the presence of lax vowels in English versus their absence in Spanish and Italian will result in greater use of the lax variant by speakers who use English in the home. Finally, we expect that there will be a correlation between lax vowel usage and greater time spent in a Francophone environment since it would provide greater exposure to this vowel.
With regard to linguistic factors, our principle hypothesis is that words having a lax vowel in an English cognate will more often contain a lax vowel in French. A similar trend was found in the case of schwa deletion, which was more common in English cognates that did not contain schwa, for example, gouvernement/government (cf. Uritescu et al., 2004). We also predict potential vowel harmony effects (as discussed above).

Results
Thirty-eight of the 41 French immersion students in our corpus provided 249 occurrences of the variable in obligatory laxing context. General results for the variable are presented in Table 2 (14) 215 (86) 249 (100) As seen in Table 2, the lax variant is surprisingly rare in the speech of Canadian French immersion students. And, this result is in spite of the fact that a) the lax variant is used categorically by Canadian Francophones in this same linguistic context, and b) the lax variant is by no means difficult to pronounce (and is in fact readily used by these same students when they speak English). However, the lax variant does have some currency in immersion students' discourse. Let us now consider the students who do make some use of the lax variant and examine the linguistic contexts in which this variant is found.

Social Factors
An important general result that emerges from the distribution of variants is that the lax variant is only used by 10 of the 38 students who provided tokens of the variable. In other words, 28 students (i.e., 74%) display no productive knowledge of the variant that is typical of Canadian French. Among the 10 students who do show variation, the lax variant is used 35% of the time (34/96). Given the small number of students who display variable usage, it is difficult to run a GoldVarb analysis on the data. We have therefore decided to look at the social characteristics of the individuals who do make at least some use of the lax variant in order to see if meaningful patterns emerge. The general dispersion of speakers using at least one instance of the lax variant is presented in Table 3. Although the fact that there are only 34 tokens of the lax variants does not allow for a robust statistical analysis, it is still instructive to consider and interpret these results in light of our initial hypotheses and previous results related to the immersion corpus.
Looking at the group as a whole, it is apparent that female students make greater use of the lax variant than do males. While males are far outnumbered by the females in this sample, only one male (or 11% of the males vs. 31% of females) used the lax variant. Recall that we had hypothesized the contrary, that is, that females would have a greater tendency to use the "formal" variant /i/ than males, which does not appear to be the case in these results. Assuming students were exposed to the lax variant in teachers' speech, it is possible to interpret our results not in terms of greater sensitivity to an external standard norm, but rather in terms of greater sensitivity to/acceptance of teacher speech or classroom input. However, without audio recordings of the teachers, it is impossible to verify this assertion. Regardless of the explanation, the results do suggest that female students are more aware of variation and that their sociolinguistic competence is more advanced.
The same pattern for gender can be seen in the results by social class. Here we find greater dispersion of the variant /ɪ/ in the middle class compared to the working and lowermiddle classes. This suggests greater sensitivity to classroom input among middle-class speakers.
Regarding exposure to a Francophone environment, we predicted that greater exposure to French outside the classroom would lead to an increase in use of the lax variant /ɪ/. However, no clear trend emerges from the four-way grouping we have used in the past since the five students who had the most exposure to a Francophone environment (more than 3 weeks) did not produce any tokens with a lax vowel. Still, if we simply made a binary division for exposure to French language environment, the results would be 17% for no exposure and 31% for some exposure and would therefore support our initial hypothesis.
As concerns home language, we see that the lax variant is more widely dispersed among English speakers than it is among speakers of Spanish and Italian. This is not surprising, given the fact that unlike English, the latter two languages do not have a lax /ɪ/ in their phonemic inventory.

Linguistics Factors
While the tense/lax variant is not variably conditioned by linguistic factors in L1 Canadian French, we did consider a range of potentially relevant factors (see previous list of four linguistic factors described above). However, only one factor exercises a significant effect on variant choice. The factor in question concerns the existence of an English cognate that has a lax vowel in final position. Table 4 lists the different types of words that were produced by the 10 students displaying variation, along with the total number of times each word was produced with a tense or lax vowel, and whether or not the word has an English cognate (with either a tense or lax vowel).  The general quantitative effect of English cognate words is presented in Table 5.  Table 5 is real, it seems likely that the cases of laxing that do exist should not be attributed to English cognates, nor should they be attributed to the learning of a general phonetic rule since it is applied too selectively to justify such a claim. Rather, they must have been learned as part of a lexical unit. In other words, when proper nouns like Christine and Phillipine were learned, the final lax vowel was simultaneously acquired in these words.

Discussion
While the data available are limited, there is one very clear result that emanates from our study. Namely, that the French immersion students make surprisingly little use of the lax variant and, consequently, their sociolinguistic competency is far from native-like (as has been the case for the majority of variables studied, cf. Mougeon et al., 2010) and is not in keeping with the local norm. Indeed, it is only a minority of students (26%) that make any use of the lax variant. This result is somewhat surprising, given previous studies of neutral variables. And, this is in spite of the fact that the lax variant is likely present in these students' input. We say this because there are nonetheless some occurrences of the lax variant in non-cognates and indeed in cognates containing a tense vowel. Still, it may well be the case that the tense variant is the most abundant in the students' input and that it is the first one they learn when mastering French phonology in a classroom setting. It is only the more sociolinguistically advanced students that display a more native-like pronunciation that respects L1 allophonic variation.
Another factor worth considering is that the present results could be related to reasons of identity or attitude. In other words, some students may be aware of the lax variant's regional status, but unwilling to use regional features for personal reasons (see Van Compernolle & Williams, 2012) or because they have a more global and less Canadian view of themselves as L2 French speakers. Regardless, it is likely the local, lax variant that would be most useful in the workplace (cf. Gutiérrez & Fairclough, 2006). Furthermore, the lax variant would not be a difficult variant to master and would contribute to a more Canadian pronunciation without concerns of using a stigmatized variant. We therefore suggest that it be the object of explicit instruction in order that students become more aware of linguistic variation (see Ringer-Hilfinger, 2012).
With regard to gender and social class results, we saw that females and middle-class speakers are more advanced in terms of sociolinguistic competence. It therefore seems reasonable to suggest that these speakers are more sensitive to allophonic variation and more aware of relative subtleties in the input to which they are exposed. Furthermore, they may be more accepting of regional norms. Previous research on the immersion corpus has often found differences whereby females make greater use of the standard variant. However, since the lax variant is a neutral regional form, no such claim can be made for the variable under study. We have argued instead that it is variation in general to which female and middle-class speakers are more sensitive in the classroom. This interpretation could also be applied to previous variables where a female/male difference has been found. Consider the results presented in Table 6: In the first three cases, it appears that females are indeed making greater use of the (more) standard variant. In the fourth case (habiter/vivre), it could be argued that the reverse is true since habiter is a stylized variant in Canadian French (though vivre too is obviously a standard form). What these four variables share, however, is the same trend that we have identified for the laxing variable; namely, that males are more categoric in their usage of variants (82% periphrastic, 71% on, 66% juste, and 79% habiter), while females display greater variation. If variation is learned through a series of progressions whereby one begins at a stage of categoric usage and eventually progresses to a more native-like variable system, the data presented above all suggest that female students are more advanced along this trajectory.
As for the influence of English on variant choice, we did note a tendency among English home-language students to make greater use of the lax variant. Or, conversely, we saw that students from a Romance language background showed a stronger preference for the tense variant, in keeping with their L1s. However, the cognate results cannot be