Acquisition of Variation in the Use of alors, donc, fait que by Advanced French-as-a-Second-Language Learners in Ontario, Canada

Abstract

This study examines the acquisition of sociolinguistic variation in the use of French connectors alors/donc/fait que ‘so’ by two groups of advanced French-as-a-second language (FL2) learners in Ontario: (i) high school French Immersion (FI) students and (ii) university students. It considers two types of functions fulfilled by these connectors: (i) the grammatical function of expressing consequence between two clauses and (ii) a range of discursive functions, a dual focus not present in previous research, which considered only one or the other of these two types of functions. Our study shows that: (i) although these two FL2 groups’ use of the connectors is distant from the norms of FL1 speech, the university students achieve a more advanced level of acquisition of this case of variation than do the FI students, reflecting the positive effect of continued learning of French at the postsecondary level; (ii) ‘level of opportunities to interact in French with native speakers’ has a greater positive impact on the acquisition of alors/donc/fait que than ‘time spent learning French’; and (iii) both groups of students evidence incomplete acquisition of the linguistic constraints of connector choice.

1. Introduction

The Government of Canada’s (2018) Action Plan for Official Languages as a way to identifies promoting a bilingual Canada as a way to build bridges between speakers of English/other languages and Francophones. In Ontario (Canada’s most populous Anglophone province), an essential aspect of advanced bilingualism is the development of French-as-a-second language (FL2) learners’ sociolinguistic competence. In line with this, in its high school FL2 curriculum, the Ontario Ministry of Education identifies sociolinguistic variation as an important aspect of advanced FL2 competence. In this respect, the FL2 curriculum states that students will be able to “identify sociolinguistic conventions associated with a variety of social situations in diverse French-speaking communities and use them appropriately in spoken interactions” (Ontario Ministry of Education, 2014, p. 65).

Such considerations are at the heart of a strand of research examining the acquisition of sociolinguistic variation by advanced second-language (L2) learners. This research views successful acquisition of variation as learners’ knowledge of the full repertoire of native variants, their use of variants at nativelike frequencies, and their observance of the (extra-)linguistic constraints on variation adhered to by native speakers. Early studies focused on sociolinguistic variation in English by L2 learners, such as Ellis’s (1987) study of variable past tense use in narrative discourse and Adamson and Regan’s (1991) investigation of the variable realization of –ing. In the following decades, this strand of L2 acquisition research has expanded considerably in terms of the range of languages and sociolinguistic variables investigated.

As part of this strand of research, the present study examines the acquisition of sociolinguistic variation in the use of French connectors alors/donc/fait que by two groups of advanced FL2 learners in Ontario: (i) high school French Immersion (FI) students and (ii) university students. In analyzing the speech of these students, it considers two types of functions fulfilled by alors/donc/fait que. Along with other researchers (e.g., Dessureault-Dober, 1974; Blondeau et al., 2022; Bigot & Papen, 2021), we distinguish: (i) the grammatical function of expressing consequence between two clauses (see examples 1–4 in Section 3.2) and (ii) a range of discursive functions (e.g., turn taking, pause filling, topic introduction—see examples 5–10 in Section 3.2), a dual focus not considered in previous research, which considered either the grammatical function (e.g., Rehner et al., 2003) or the discursive ones (Lemée, 2025; G. Sankoff et al., 1997).

The present study contributes to current understandings in three main ways. The first is by showing that continued learning of French at the postsecondary level leads to a more advanced level of acquisition of this case of variation in that the university students display more nativelike frequencies of use of alors/donc/fait que than do the FI students. In the second contribution we compare for both groups of FL2 students the relative impact on the acquisition of alors/donc/fait que of ‘time spent learning French’ versus ‘level of opportunities to interact in French with French-as-a-first-language (FL1) speakers’. Our study shows for the FI students that time spent learning French is more influential than opportunities to interact with FL1 speakers but for the university students that level of opportunities is the better predictor of acquisition. The third contribution is to show that the advantage afforded by postsecondary learning of French does not extend to the acquisition of nativelike linguistic constraints on connector choice. In arriving at these findings, we use spoken Montreal FL1 French as an acquisition benchmark.

2. Previous Research

2.1. Sociolinguistic Competence of FL2 Learners

In reviewing past research on FL2 sociolinguistic competence, we focus on studies of advanced FL2 learners in Canada, the bulk of which has examined two groups: (i) Ontario high school FI and university FL2 students and (ii) young adult FL2 learners residing in Montreal, Quebec. This focus reflects the direct relevance of such research to the present study and the fact that the results of the Canadian research are in line with those of studies of FL2 learners in other countries (e.g., Dewaele, 2007; Howard, 2006; Howard et al., 2006; Regan, 2022a, 2022b; Regan et al., 2009; Sax, 2003; Kennedy Terry, 2022). Further, the focus on Canadian FL2 learners provides a principled means to narrow down the vast body of FL2 sociolinguistic studies, since as Howard et al. (2013) and Long (2022) point out, French is among the most-studied languages by L2 variationist researchers. Our review considers the FL2 students’ repertoire of sociolinguistic variants, the frequency with which they use them in comparison with FL1 speakers, and the impact of (extra-)linguistic factors on such frequency.

Previous research on the acquisition of sociolinguistic variation by the Canadian FL2 learners mentioned above has examined cases of morphosyntactic, lexical, and phonological variation, as well as the use of discourse markers. It has found that even though these learners have acquired many of the variants used by FL1 speakers, there are some gaps in their repertoire of variants, notably an absence of several hyper-formal and socially marked vernacular variants in the speech of FI and FL2 university students in Toronto, Ontario (R. Mougeon et al., 2010; F. Mougeon & Rehner, 2015) and of several discourse markers in the speech of university FL2 students in Thunder Bay, Ontario (Lemée, 2025). Lastly, these same authors have found that some of the lesser advanced learners make occasional use of non-native forms instead of the expected FL1 variants (e.g., use of the infinitive instead of the future tense).

In relation to variant frequency, previous research has found that compared to FL1 speakers, Canadian FL2 learners tend to overuse formal variants and underuse informal ones and that the extent of such overuse/underuse is conditioned by learners’ level of opportunities to interact with FL1 speakers. For instance, use of formal subject pronoun nous ‘we’ (compared to its informal counterpart on) is 45% for Ontario FI students (Rehner et al., 2003), sits at 25% for Ontario FL2 university students (F. Mougeon & Rehner, 2015), and is at a low of only 3% for Montreal FL2 learners (Blondeau et al., 2002), almost matching the frequency documented in FL1 speech.

Previous research has also found that the magnitude of the differences between FL2 learners and FL1 speakers in frequency of variant use is conditioned by the socio-stylistic salience of the variants. For instance, Blondeau et al. (2002) found that FL2 Montreal learners use complex non-clitic plural pronouns (socially-marked vernacular variants) significantly less often than do FL1 speakers but delete negative particle ne ‘not’ (a near categorical feature of ordinary FL1 speech) at a level of frequency close to that of FL1 speakers. Lastly, previous research has found that the systemic properties of sociolinguistic variants may have either a negative or positive impact on their acquisition by FL2 learners. For instance, in research on the acquisition of variable schwa deletion and/l/deletion in subject pronouns il/ils ‘he-it/they’, R. Mougeon et al. (2010) found that FI students almost never delete/l/but delete schwa some of the time. According to the authors, this difference reflects that in English it is possible to delete schwa in some contexts, while/l/is undeletable in final position.

With respect to the acquisition of linguistic constraints on variant choice, previous research has found that FL2 speakers do not uniformly adhere to FL1 linguistic constraints. Uritescu et al. (2004) and Thomas (2002) found, respectively, that FI students and Ontario FL2 university students have acquired the hierarchy of FL1 phonetic constraints on schwa deletion, and Blondeau et al. (2002) found that anglophone FL2 speakers in Montreal have acquired FL1 morphosyntactic constraints on subject doubling. However, Rehner and Mougeon (1999) and Thomas (2002) found respectively that FI students and Ontario FL2 university students have not learned the FL1 constraint on ne deletion associated with the type of postverbal negator. Lastly, R. Mougeon et al. (2010) found that FI students sometimes demonstrate their own linguistic constraints. For example in relation to variable use of restrictive adverbs juste/seulement/rien que ‘only’, R. Mougeon and Rehner (2001) found that FI students sometimes use juste left of the verb, a context where FL1 speakers never use it.

Finally, R. Mougeon et al. (2010) in their investigation of FI students’ acquisition of variants expressing futurity (periphrastic vs. inflected future tenses) and those expressing restriction (adverbs juste, seulement, etc. ‘only’) found that level of extra-curricular exposure to FL1 speech had a greater positive impact than time spent learning French. F. Mougeon and Rehner (2009), for their part, found that time spent learning French had no significant impact on FL2 university students’ acquisition of informal pronoun on in contrast with the strong effect of time spent in a Francophone environment. These results are generally in line with those of previous studies on the acquisition of sociolinguistic variants by non-Canadian FL2 learners (see Dewaele & Regan, 2002; Dewaele, 2004 for a review of these studies).

2.2. Use of alors, donc, and fait que by Canadian FL1 Speakers

As will become clear in the methodology section, Quebec French is the variety of Canadian French to which most of the FL2 speakers in our study have been exposed. Therefore, we limit our review of studies on variable use of alors/donc/fait que by FL1 speakers to those based on Quebec French corpora.

The first such study is by Dessureault-Dober (1974) and is based on a sample of 25 speakers from the 1971 Sankoff and Cedergren corpus of Montreal French. It found that these speakers used fait que—a form that occurs in Quebec French and in other varieties of Canadian French— and alors—a feature of standard French—at roughly comparable levels of frequency but used donc—another standard variant—only marginally. Beyond these general frequencies, Dessureault-Dober found that: (i) alors was used primarily by older high-socio-economic (SES) speakers, (ii) the younger high-SES speakers were evidencing a shift toward fait que—the vernacular variant which was solidly entrenched in the low-SES speakers’ speech, and (iii) topic (in)formality influenced variant choice—formal topics being associated with more frequent use of alors and informal topics with greater use of fait que.

A second study, by Blondeau et al. (2022), draws on the full 120 speakers in the 1971 Sankoff and Cedergren Montreal corpus and on a corpus collected in Montreal 40 years later (50 speakers). The authors found that over 40 years: (i) the frequency of alors decreased sharply at the expense of donc and (ii) the frequency of fait que rose markedly. They also confirmed Dessureault-Dober’s finding of alors’ association with high-SES speech and revealed that donc was also associated with high-SES speakers and that the steady rise of fait que was led by lower-SES women. Lastly the authors examined the impact of the linguistic functions of the connectors on speaker choice. They distinguished two types of functions: (i) the grammatical function of expressing a consequence between two clauses and (ii) discursive functions such as ‘turn yielding’ or ‘pause filling’. They found that discursive functions favor the use of fait que and the grammatical function favors the use of alors and donc.

A third study, by Beaulieu et al. (2019), examined the use of alors/donc/fait que by high-SES speakers interviewed on the French-Canadian Broadcasting Corporation network during a news series aired from 2008 to 2013. The results showed an overall frequency of 57% for donc, 43% for alors, and no use of fait que, suggesting that standard alors and donc are deemed as more appropriate than fait que in formal communication situations, echoing Dessureault-Dober’s (1974) finding of the impact of topic (in)formality on the use of alors/donc vs. fait que.

2.3. Grammatical Use of alors, donc, fait que, and so by Canadian FL2 Speakers

Two studies examined the acquisition of alors/donc/fait que by Canadian FL2 speakers. Rehner and Mougeon (2003) focused on the FI corpus used for the present paper, and Rehner and Beaulieu (2008) focused on a subsample of the FL2 university corpus considered in the present paper. Both studies considered the grammatical use of these connectors.

Table 1. Grammatical use of alors, donc, fait que and so in FL1 and FL2 speech.

Speakers	Alors	Donc	Fait que	So	Total
FL1 Montreal (2012)	(68) 4%	(388) 22%	(1323) 74%	0	1779
FL2 Immersion	(484) 78%	(96) 15%	0	(44) 7%	624
FL2 University	(578) 73%	(195) 25%	0	(14) 2%	787

presents the results of these studies and includes data on the grammatical use of the connectors by FL1 speakers, taken from Blondeau et al.’s (2022) analysis of the second Montreal spoken French corpus (see Section 3.1.3 for a description of the FL1 corpora). As can be seen, in sharp contrast to the Montreal FL1 speakers, both FL2 groups strongly favor standard alors and make no use of vernacular fait que. With respect to donc, the FI students use it less frequently than the FL1 speakers, but the difference is only relative, and the FL2 university students use it somewhat more often than the FL1 speakers. Lastly, as Table 1 shows, these two studies found that the FL2 groups make some use of English connector so, a form that is absent in the 2012 Montreal corpus.

According to Rehner and Mougeon (2003), their findings reflect that the FI students have had limited or no opportunities to hear French outside the immersion classroom, a setting in which teachers use alors and donc almost exclusively. As for Rehner and Beaulieu (2008), they found that those university students with some exposure to FL1 French via short stays in a Francophone environment had a higher rate of donc and a lower use of so than the remaining students.

2.4. Use of Discourse Markers by Canadian FL2 Speakers

Three studies have investigated the acquisition of French discourse markers by Canadian FL2 speakers. G. Sankoff et al. (1997) examined a corpus of 17 young adult Anglophone FL2 speakers residing in Montreal; Lemée (2025) used a corpus of 15 FL2 university students collected in Thunder Bay, Ontario; and Rehner (2004) used the corpus of 41 FI students examined in the present study. In contrast with the FI students, the Montreal FL2 speakers have had significantly higher levels of exposure to FL1 French via extra-curricular or community activities, or in their neighborhood and, for some of them, on their jobs. While Lemée does not provide information on the FL2 university students’ extra-curricular exposure to French, she considers them as advanced learners of French since they have continued to learn French beyond high school, and they all intend to teach French as a second language.

These three studies arrived at the following general findings. The range of discourse markers used by the FL2 speakers is narrower than that found in FL1 speech. Thus, several of the markers documented in FL1 speech are either absent or used by only a limited number of the FL2 speakers. For instance, Lemée found that t’sais ‘you know’, pis, ‘and/then’, and ben ‘well’ are not used by the Thunder Bay FL2 university students. Sankoff et al. found that discourse marker bon ‘well/good’ is used by only four of the 17 Montreal FL2 speakers and that pause filler là (a ubiquitous feature of FL1 Quebec French) is used to a non-negligible extent by only five of the 17 FL2 speakers. In Rehner’s study, là is almost completely absent in the speech of the 41 FI students (FL2 speakers who have had limited opportunities to interact with FL1 speakers). As noted by Sankoff et al. and Rehner, the marked under-use of là and bon reflects in part that both markers lack an English discursive counterpart. Conversely, discourse markers with an English discursive equivalent (e.g., comme ‘like’) are used by the great majority of the FL2 speakers in their studies. Thus, for some of the discourse markers examined by these authors, inter-systemic factors can either hinder or facilitate acquisition.

Another illustration of the impact of inter-systemic transfer shown by all three studies is the occasional use of English discourse markers during French interviews. These include ‘you know’, ‘like’, and ‘well’. As noted by Lemée, such cases of direct transfer from English are mostly in the speech of the least advanced FL2 leaners.

Discursive alors/donc/fait que are among the discourse markers included in the G. Sankoff et al. (1997) and Lemée (2025) studies. In

Table 2. Discursive use of alors, donc, fait que and so in FL1 and FL2 speech in Thunder Bay and Montreal.

Speakers	Alors	Donc	Fait que	So	Total
FL1 Montreal (1971)	(1000) 48%	(22) 1%	(1078) 51%	0	2100
FL1 Montreal (2012)	(41) 2%	(372) 16%	(1928) 82%	0	2341
FL1 Thunder Bay	(134) 26%	(184) 36%	(134) 26%	(64) 12%	516
FL2 Montreal	(55) 83%	0	(6) 9%	(5) 8%	66
FL2 Thunder Bay	(130) 78%	(23) 14%	0	(11) 7%	164

, we provide the frequencies reported by these authors for alors/donc/fait que and for the conjunction so (for the Thunder Bay corpus). To lend perspective, we have included the frequencies documented by Blondeau et al. (2022) for discursive alors/donc/fait que in their analysis of the 1971 and 2012 FL1 Montreal corpora1. Table 2 also includes the frequencies found for alors/donc/fait que and so reported by Lemée in her analysis of a small corpus of FL1 speech collected in Thunder Bay.

As Table 2 shows, both FL2 groups overuse discursive alors compared to their respective FL1 benchmark norms. Both FL2 groups also make some use of discursive so, a form that like grammatical so is absent in Montreal FL1 French (see Section 3.2 for examples of grammatical and discursive uses of this connector). Thus, the occasional use of this form by the FL2 speakers is likely yet another case of transfer of a discourse marker from English. However, in the case of the Thunder Bay FL2 students, it may also in part result from exposure to the local variety of French, since, as is shown by Table 2, the Thunder Bay FL1 corpus does include this form2. Where the two groups of FL2 speakers differ is in their patterns related to donc and fait que. The FL2 speakers in Montreal do not use donc at all, in keeping with its virtual absence in the 1971 FL1 benchmark corpus. In contrast, the FL2 students in Thunder Bay use this variant, in line with its non-negligible frequency in contemporary varieties of Canadian French, indicated by its presence in the FL1 Thunder Bay and 2012 FL1 Montreal corpora. As for fait que, the table shows that the Montreal FL2 speakers make some use of it, while the FL2 students in Thunder Bay do not. Further, the few FL2 Montreal speakers who had learned the discursive use of fait que were those whose social networks brought them into close contact with FL1 speakers who made regular use of the vernacular (see G. Sankoff et al., 1997). Lastly, the overall limited to non-use of fait que across the FL2 corpora contrasts sharply with the frequencies documented in the FL1 benchmark corpora (an issue we address in the Section 5).

3. Methods

3.1. Corpora

The present study draws on two spoken French corpora collected in Ontario among FL2 learners, one from FI high school students and the other from university students. To measure the learners’ progress towards nativelike use of the connectors, we use as comparative benchmarks the findings of studies using three corpora of FL1 speech, namely Sankoff and Cedergren’s corpus of FL1 Montreal (1971) spoken French; Blondeau, Frenette, Martineau, and Tremblay’s corpus of FL1 Montreal (2012) spoken French; and Lemée’s corpus of spoken French collected in Thunder Bay, Ontario from 2013 to 2020.

3.1.1. FI High School Student Corpus

The FL2 Immersion high school student corpus was collected by R. Mougeon and Nadasdi in 1996 via semi-directed one-on-one interviews with 21 Grade 9 students (age 14–15) and 20 Grade 12 students (age 17–18) that focused on a variety of topics encouraging the students to speak about issues of interest to them. The students also filled out a questionnaire gathering information on the frequency with which they used French and other languages in and out of the FI classroom. The FI program in which the students were enrolled is housed in English-medium schools located in the greater Toronto area and is characterized by 50% French-medium instruction in Grades 5 to 8, followed by 20% from Grades 9 to 12. Further, in the schools where the program is located, the great majority of the administrative, teaching, and maintenance staff and students are not French speakers. In other words, the classrooms where these students took their courses in French and the resource rooms attached to the FI program were about the only school settings in which the students had the opportunity to use and be exposed to French. Outside of the school in daily life, the students reported never or only marginally using French, reflecting the local scarcity of Francophones. However, some students reported having had opportunities to interact with FL1 speakers via stays of varying lengths in a Francophone environment, for the most part in Quebec. Sixteen students reported having stayed in a Francophone environment for less than seven days, eleven students from seven to thirteen days, and thirteen students for more than two weeks.3 These short stays mostly took place during trips organized by the schools where students were unlikely to have had the occasion to experience intensive interactions with FL1 speakers.

3.1.2. FL2 University Student Corpus

The FL2 university student corpus was collected by F. Mougeon in 2005 and 2008 on the bilingual (French-English) campus of a university in Toronto. Data collection consisted of surveys and semi-directed one-on-one interviews conducted with 40 1st year and 40 4th year students. Both cohorts included relatively equal proportions of students from traditional FSL programs (i.e., Core French) and FI programs (1st year: Core = 21, FI = 19; 4th year: Core = 17, FI = 23). The interviews followed a similar protocol to that used with the FI high school students. All the university students were taking undergraduate FL2 courses and some were taking courses in other subjects taught through the medium of French. The students reported interacting to varying extents in French on the campus with FL1 staff, administrators, service providers, and fellow students. They also reported varying levels of commitments to using French off campus, either in Toronto (e.g., social contacts with FL1 speakers, part-time work in a local Francophone business) or outside of Toronto (e.g., trips to visit friends in Quebec). This information on the students’ use of French on and off campus was used to calculate a global index of commitments to use French (F. Mougeon & Rehner, 2015). In the present study, we use these authors’ categorization of the 80 students according to three levels of engagement: 29 minimally engaged, 29 moderately engaged, and 22 highly engaged.

3.1.3. FL1 Corpora

Like the corpora of FL2 speech used by the present study, all three FL1 corpora mentioned above were collected with one-on-one semi-directed interviews. The FL1 Montreal (1971) corpus includes a total of 120 speakers, and it is stratified according to speaker age, SES and gender (D. Sankoff et al., 1976). The FL1 Montreal (2012) corpus includes a total of 50 speakers, and it is also stratified according to age, SES, and gender (see Blondeau et al., 2022). As for the Thunder Bay corpus, it includes 15 speakers, mostly females and ranging in age from 17 to 78. These speakers’ SES is unknown (see Lemée, 2025).

3.2. Data Analysis

As mentioned above, the present study examines the acquisition of both the grammatical and discursive uses of the connectors. This differs from previous studies that focused on one or the other of these uses. Further, our study expands the number of tokens considered in its quantitative analyses since it uses the entire data set of 80 FL2 university students instead of the subset of 61 students examined by Rehner and Beaulieu (2008).

The grammatical function examined is that of consequence between two clauses or sentences (examples 1–4). The discursive functions are primarily that of topic closure (examples 5–8) and, secondarily, topic introduction (example 9), and pause filling (see example 10). These examples are taken from the FL2 corpora.

• Grammatical Function

• ma soeur a eu un emploi alors elle ne peut pas aller (FI)
  ‘my sister has a job so she cannot go’
• les examens c’est surtout sur le vocabulaire donc je peux apprendre les mots (University)
  ‘the exams are mostly on the vocabulary so I can learn the words’
• puis quatre mois de coop ça fait que on finit en même temps (University)
  ‘then four months in the coop program so we complete [the term] at the same time’
• nous avons le même âge so alors ça c’est bon (FI)
  ‘we are the same age so so that is good’

• Discursive Functions

Topic Closure

5.

Student: j’ai pas étudié le l’espagnol depuis deuxième année ou troisième année alors (pause) Interviewer: donc il faudrait reprendre (University)

Student: ‘I stopped studying Spanish after second or third year so (pause)’ Interviewer: ‘so you should start taking it again’

6.

Student: ahm well je suis catholique donc (pause) Interviewer: oui mais tu pourrais être catholique et aller dans une école publique (FI)

Student: ‘um well I am catholic so (pause)’ Interviewer: ‘yes but you could be catholic and go to a public school’

7.

Student: je prends les études internationales maintenant fait que (pause) Interviewer: donc nécessairement vous vous y intéressez (University)

Student: ‘I am taking international studies so (pause)’ Interviewer: ‘so necessarily you are interested in that’

8.

Student: je suis bon dans les sciences et les mathématiques so (pause) Interviewer: d’accord donc l’année prochaine tu suis combien de cours (FI)

Student: ‘I am good at science and math so (pause)’ Interviewer: ‘alright so next year how many courses are you taking?’

Topic Introduction

9.

Student: pourquoi je vote pour cette personne? Interviewer: ouais Student: alors maintenant je m’implique plus dans ça (University)

Student: ‘why do I vote for this person? ’ Interviewer: ‘yeah’ Student: ‘so now I am more involved in that’

Pause Filling

10.

je ne sais pas comment dire mais uhm so ils ont décidé de déménager (FI)

‘I don’t know how to say it but um so they decided to move out’

All of the tokens of the connectors that were produced by the FI and university students were extracted from the transcriptions of the recordings and were coded for a statistical analysis of their frequencies according to several factors, which allowed us to examine the research questions mentioned below. The evaluation of the statistical significance of intergroup differences in the frequency of use of alors/donc/fait que and so was assessed via GoldVarbX (Tagliamonte, 2006). GoldVarb calculates variant use frequencies as percentages and assesses the impact of (extra-)linguistic factors on such use, as indicated through factor effect values ranging from 0.01 (a highly disfavoring effect) to 0.99 (a strong favoring effect). Categorical use or non-use of a variant is indicated with the short form “k.o.”(knock out) in the GoldVarb analysis. Statistical significance rates, calculated by GoldVarb, are provided for the intergroup comparisons, with 0.05 as the significance threshold.

3.3. Research Questions

The present study expands on previous research on the acquisition of alors/donc/fait que by FL2 learners in several ways. First, it is the first study that compares high-school FI students with university students, with previous research having examined only one or the other of these two groups of FL2 learners. Thus, our study asks whether the FL2 university students achieve a more advanced level of acquisition of the connectors than do the FI students, suggesting a positive effect of continued learning of French at the postsecondary level. Second, this study seeks to answer the question, as in previous research on the acquisition of sociolinguistic variation by FL2 learners, does the factor of ‘time spent learning French’ have a greater positive impact on the acquisition of alors/donc/fait que than the factor ‘level of opportunities to interact in French with FL1 speakers’. Lastly, the present study investigates the question of what impact the linguistic functions of the connectors (i.e., grammatical vs. discursive) have on variant choice and to what extent such impact matches that documented in FL1 speech. This question was also not investigated in previous research.

4. Results

The results are presented below in response to the research questions guiding the study.

4.1. Does Postsecondary Study Lead to a More Advanced Acquisition of the Connectors?

Table 3. GoldVarb analyses of the frequencies of grammatical use of alors, donc, fait que and so by FL2 Immersion and FL2 university students and FL1 Montreal (2012) benchmark.

Groups	Alors		Donc		Fait que		So		Total (N)
	(N) %	Effect	(N) %	Effect	(N) %	Effect	(N) %	Effect
FL2 Immersion	(506) 83	[.59]	(81) 13	[.37]	0	k.o.	(26) 4	[.68]	613
FL2 University	(1058) 74	[.46]	(346) 24	[.56]	(3) 0.2	--	(21) 2	[.42]	1428
Significance	<0.001		<0.001		NA *		<0.001
FL2 Immersion	(506) 83	[.97]	(81) 13	[.39]	0	k.o.	(26) 4	--	613
FL1 Montreal (2012)	(68) 4	[.23]	(388) 22	[.54]	(1323) 74	--	0	k.o.	1779
Significance	<0.001		<0.001		NA		NA
FL2 University	(1058) 74	[.92]	(346) 24	n.s.	(3) 0.2	[.02]	(21) 2	--	1428
FL1 Montreal (2012)	(68) 4	[.13]	(388) 22		(1323) 74	[.96]	0	k.o.	1779
Significance	<0.001		0.108		<0.001		NA		T = 3820

* NA: not applicable.

and

Table 4. GoldVarb analyses of the frequencies of discursive uses of alors, donc, fait que and so by FL2 Immersion and FL2 university students and FL1 Montreal (2012) benchmark.

Groups	Alors		Donc		Fait que		So		Total (N)
	(N) %	Effect	(N) %	Effect	(N) %	Effect	(N) %	Effect
FL2 Immersion	(291) 72	[.45]	(43) 11	[.40]	0	k.o.	(71) 18	[.80]	405
FL2 University	(535) 78	[.53]	(131) 19	[.56]	(4) 1	--	(15) 2	[.31]	685
Significance	0.02		<0.001		NA		<0.001
FL2 Immersion	(291) 72	[.99]	(43) 11	[.41]	0	k.o.	(71) 18	--	405
FL1 Montreal (2012)	(41) 2	[.33]	(372) 16	[.52]	(1928) 82	--	0	k.o.	2341
Significance	<0.001		0.008		NA		NA
FL2 University	(535) 78	[.98]	(131) 19	[.54]	(4) 1	[.006]	(15) 2	--	685
FL1 Montreal (2012)	(41) 2	[.23]	(372) 16	[.48]	(1928) 82	[.82]	0	k.o.	2341
Significance	<0.001		0.047		<0.001		NA		T = 3431

provide an overview of the intergroup frequencies of grammatical and discursive uses of alors/donc/fait que and so by the FI and FL2 university students and FL1 speakers. These intergroup frequencies were examined via GoldVarb univariate analyses. As can be seen, the university students are, by and large, more advanced than the FI students in their acquisition of the connectors, both for the grammatical and discursive functions. For donc, this is indicated by the higher percentages and factor effects for the FL2 university students compared to the FI students and by the (near) lack of statistical significance of the differences between the FL2 university students’ use and that of the FL1 benchmarks. For so, the FL2 university students’ more advanced acquisition is shown by their lower grammatical and discursive uses compared to the FI students, which bring them closer to the non-use by the FL1 speakers. In fact, the 18% rate of discursive so for the FI students makes this their second most frequent variant, indicating the importance of L1 transfer in their discursive repertoire of connectors. With respect to fait que, the FL2 university students’ advantage is indicated by the presence (albeit marginal) of this highly frequent FL1 variant in its grammatical and discursive functions, a use that was absent in the subsample examined by Rehner and Beaulieu (2008) (see Section 2.3). This use, while marginal, contrasts with the complete absence of fait que in the FI students’ speech. As for alors, the university students are once again ahead of the FI students in their grammatical use of this variant with a statistically significantly lower rate (even if it is still far from the FL1 norm). While the same lower rate is not evident for the university students’ discursive use of alors, it must be kept in mind that the FI students’ overreliance on discursive so is at the expense of alors, making it look as if they are closer to the FL1 norms.

4.2. Does the Level of Opportunities to Interact in French Have a Greater Positive Impact on Acquisition than Time Spent Learning French?

Table 5. GoldVarb analyses of the impact of extra-linguistic factors on grammatical use of alors, donc and so.

Groups	Factors	Alors		Donc		So
		(N) %	Effect	(N) %	Effect	(N) %	Effect
FL2 Immersion
Francophone Environment	<7 days	(184) 92	.67	0	k.o.	(17) 9	.80
	7–13 days	(119) 72	.34	(38) 23	n.s. **	(8) 5	.69
	2 weeks +	(199) 82	.47	(43) 18		(1) 0.4	.15
Grade	9	(222) 88	.59	(14) 6	.32	(16) 6	n.s.
	12	(284) 78	.44	(67) 19	.61	(10) 3
Significance		0.009		<0.001		<0.001
FL2 University
Engagement	Low	(280) 88	.64	(27) 8	.30	(13) 4	.74
	Mid	(493) 92	.76	(36) 7	.25	(8) 0.7	.43 *
	High	(285) 50	.20	(283) 49	.82
Year of Study	1st	(388) 83		(61) 13	n.s.	(16) 3	.74
	4th	(670) 70		(285) 30		(5) 0.5	.38
Significance		<0.001		<0.001		0.007

* Due to the low number of tokens of so, there was an interaction for engagement. Therefore, we have regrouped the mid and high categories. ** n.s.: (not significant) no significant factor effect has been found.

and

Table 6. GoldVarb analyses of impact of extra-linguistic factors on discursive uses of alors, donc, and so.

Groups	Factors	Alors		Donc		So
		(N) %	Effect	(N) %	Effect	(N) %	Effect
FL2 Immersion
Francophone Environment	<7 days	(106) 66		(6) 4	.33	(48) 30	.79
	7–13 days	(71) 72		(11) 11	.56	(17) 17	.69
	2 weeks +	(106) 80		(26) 20	.66	(1) 1	.10
Grade	9	(91) 63	.40	(4) 3	.29	(49) 34	.68
	12	(200) 77	.56	(39) 15	.62	(22) 8	.40
Significance		0.007		0.009		<0.001
FL2 University
Engagement	Low	(184) 91	.67	(11) 5	.27	(8) 4	n.s.
	Mid	(241) 90	.66	(21) 8	.35	(4) 2
	High	(110) 51	.18	(99) 46	.85	(4) 1
Year of Study	1st	(178) 80	n.s.	(32) 14	n.s.	(10) 5	.73
	4th	(357) 77		(99) 21		(5) 1	.38
Significance		<0.001		<0.001		0.008

present the results of multivariate GoldVarb analyses comparing the predictive value of ‘opportunities to interact in French with FL1 speakers’ with ‘time spent learning French’ for the grammatical and discursive uses of the connectors. Data were collected and analysed from the FI and the FL2 university student corpora presented in Section 3.1.1 and Section 3.1.2 above. Note that fait que was excluded from these analyses due to its limited use by the FL2 students. That said, it should be pointed out that three of the four students who make some grammatical and discursive use of this variant are highly-engaged university students, while only one is in Year 4. As Table 5 and Table 6 show, opportunities to interact with FL1 speakers is clearly more influential than time spent learning French for the FL2 university students’ grammatical and discursive uses, while the picture is less straightforward for the FI students. Beginning with the university students, for both uses there is greater spread in the percentages and effects for level of engagement indicating its greater influence compared to year of study. In fact, year of study is influential only for their grammatical and discursive use of so, whereas engagement is influential for their grammatical and discursive uses of alors and donc and their grammatical use of so. In contrast, for the FI students, both factors are influential without a clear pattern emerging. For instance, grade is more influential than time spent in a Francophone environment for their use of discursive alors, whereas this pattern flips for their use of grammatical so. Both factors appear to have a roughly equal influence on the FI students’ use of grammatical and discursive donc.

Table 5 and Table 6 also allow us to zero in on the influence of opportunities to interact in French with FL1 speakers (including stays in a Francophone country, bilingual jobs, interacting with Francophone staff on campus, and building a network of bilingual and Francophone friends). A clear effect of this factor is observable at the highest level of engagement for the university students. Specifically, these students display dramatically lower levels of grammatical and discursive alors (50% and 51%) alors and dramatically higher levels of grammatical and discursive donc (49% and 46%) than do their counterparts with mid or low engagement levels. As for the FI students, the influence of time spent in a Francophone environment (e.g., short stays in a Francophone family) is more complex. For grammatical and discursive donc and for grammatical alors, it is the group of FI students with the least time spent who stand out. In contrast, for so, it is the FI students with the most time spent in a Francophone environment whose grammatical and discursive use of this connector comes closer to approximating FL1 benchmark norms.

4.3. Do the FL2 Students Respect the FL1 Linguistic Constraints on Connector Choice?

To determine if the FL2 students adhere to the linguistic constraints on connector choice documented in FL1 Montreal spoken French (see Section 2.2), we have performed GoldVarb analyses measuring the impact of the linguistic functions of the connectors (discursive vs. grammatical) in the spoken French of the FI students, the FL2 university students, and the FL1 Montreal speakers (2012 corpus). The results are presented in

Table 7. GoldVarb analysis of the influence of linguistic functions on connector choice by FL1 Montreal (2012) speakers, FL2 immersion and FL2 university students.

	Alors			Donc			Fait que			So
	N	%	Effect	N	%	Effect	N	%	Effect	N	%	Effect
FL1 Montreal (2012)
Grammatical	68	3.8	.61	388	21.7	.55	1334	74.5	.44	0	0
Discursive	41	1.8	.42	373	16	.46	1918	82.2	.55	0	0
Significance	<0.001			<0.001			<0.001
FL2 University
Grammatical	1058	74.1	.48	346	24	.52	3	0.2	--	21	1.5	--
Discursive	535	78.1	.54	131	19.1	.45	4	0.6	--	15	2.2	--
Significance	0.046			0.009			NS			NS
FL2 Immersion
Grammatical	506	82.5	.56	81	13.2	--	0	0	--	26	4	.35
Discursive	291	71.9	.41	43	10.6	--	0	0	--	71	18	.72
Significance	<0.001			NS			NA			<0.001

. For the FL1 speakers, we see their use of alors and donc is favored by the grammatical function and fait que by the discursive functions. For the university students, we see that they favor donc to fulfill the grammatical function, but unlike the FL1 speakers, they favor alors to fulfill discursive functions. That said, this non-native association is weak, as indicated by the small spreads within the percentages and factor effects and the almost non-significant X² value computed by GoldVarb. Like the FL1 speakers, the university students use fait que more often discursively than grammatically. However, this difference rests on a very small number of tokens and is unsurprisingly judged as statistically non-significant by GoldVarb. Lastly, with developmental so, the analysis reveals an absence of statistically significant frequency differences. As for the FI students, like the FL1 speakers, they favor alors to fulfill the grammatical function and use donc more often grammatically than discursively, however this difference is judged statistically non-significant by GoldVarb. Finally, with respect to so, the FI students strongly favor its discursive use, demonstrating trouble curbing their use of this English discourse marker. This non-negligible discursive use of so has the serendipitous consequence of bringing down their frequency of discursive alors to make it fit more closely with the FL1 linguistic constraint.

5. Discussion

5.1. Postsecondary Study Leads to a More Targetlike Acquisition of Sociolinguistic Variants

Beginning with whether postsecondary study leads to more advanced acquisition, our results show that, overall, the FL2 university students achieve a higher level of acquisition of the variable use of grammatical and discursive alors/donc/fait que than do the FI students. Specifically, the university students make less frequent use of standard alors, more frequent use of standard donc, more frequent use of vernacular fait que, and less frequent use of developmental so. Thus, they are closer to FL1 sociolinguistic norms than are the FI students. To understand if the university students’ more advanced acquisition extends beyond the connectors,

Table 8. Frequencies of sociolinguistic variants in the speech of FL1 Montreal speakers, immersion teachers, FL2 immersion students, and FL2 university students.

Variants	FL1 Montreal Speech * %	Immersion Teachers’ In-Class Speech %	FL2 Immersion Students’ Speech %	FL2 University Students’ Speech %
Alors **	4	76	83	74
Donc **	22	23	13	24
Fait que **	75	1	0	0.2
So **	0	0	4	2
Nous	2	17	45	25
On	98	83	55	75
Ne use	0.5	71	70	58
Ne non-use	99.5	29	30	42

* Alors/donc/fait que/so (2012 Montreal corpus); nous/on and ne use/ne non-use (1971 Montreal corpus). ** For the four groups of speakers, the frequencies for the four connectors are for their grammatical use.

sets our findings alongside two previous studies of sociolinguistic variation in the speech of these two FL2 groups, namely standard subject pronoun nous versus informal on (Rehner et al., 2003; F. Mougeon & Rehner, 2015) and standard use of ne in negative sentences versus its informal non-use (Rehner & Mougeon, 1999; Rehner, 2010). The table also provides information on these variants in FL1 speech and in FI teachers’ in-class speech. While no classroom corpus is available for university instructor speech, the FI teacher data provide a sense of the type of educational input that these FL2 university students would have received earlier in their studies.

While Table 8 shows that both FL2 groups are quite far from FL1 frequencies and are much closer to pedagogical norms of teachers’ classroom speech, the university students’ more advanced acquisition is also reflected in the nous vs. on and ne use vs. non-use variables. For each variant, the university students’ rates are closer to FL1 norms than are those of the FI students. Specifically, the university students make greater use of the informal variants that are frequent in FL1 speech and less use of the standard variants that are privileged in the educational input. These patterns are reminiscent of those of FL2 learners living in Montreal (e.g., Blondeau et al., 2002) and FL2 learners with study abroad experience (e.g., Regan et al., 2009) and reflect, as we have seen, that the university students have had more opportunities to interact with FL1 speakers than have the FI students.

5.2. Level of Opportunities to Interact in French Has a Greater Positive Impact on Acquisition than Time Spent Learning French

The results show that opportunities to interact with FL1 speakers is a better predictor of acquisition than is time spent learning French. This is clearly shown for the FL2 university students’ grammatical and discursive use of the connectors (see Table 7). While more years spent studying French only lowers grammatical and discursisve so, those students with high levels of engagement are at a clear advantage over their less-engaged counterparts for the other connectors. In fact, in comparison with the FL1 speakers, the highly-engaged university students are overusing grammatical and discursive donc, but not alors or so. Further, with respect to the highly underused connector fait que, as pointed out above, the majority of the students who produced this variant are highly-engaged university students, while only one is in Year 4. The stronger impact of level of engagement on the FL2 university students’ acquisition of the connectors is in line with the findings of F. Mougeon and Rehner’s (2009) study of the acquisition of the variable use of on vs. nous by these same students and of other cases of sociolinguistic variation by non-Canadian university FL2 students (see Section 2.1).

For the FI students, the pattern is less straightforward, with opportunities to interact with FL1 speakers and time spent learning French both influencing the grammatical and discursive use of these forms. Grade 12 FI students are at an advantage over their Grade 9 counterparts for the grammatical and discursive use of each of the connectors, except for grammatical so. For the impact of opportunities to interact with FL1 speakers, it is the group of FI students with the least time spent in a Francophone environment who stand apart in terms of their grammatical and discursive use of donc and their grammatical use of alors. In contrast, for so, it is the FI students with the most time spent in a Francophone environment whose grammatical and discursive uses of this connector come close to approximating FL1 benchmark norms. That opportunities to interact with FL1 speakers have less of a clearly positive impact on the acquisition of the connectors by the FI students than is the case for the FL2 university students may reflect that the FI students’ extra-curricular level of exposure to FL1 French is not as high as that of the university students.

To understand properly the pattern of increased donc use with greater FL1 contacts for both the FI and FL2 university students, one needs to consider the impact of the students’ educational input. As pointed out above (see Table 8), FI teachers’ in-class speech features overwhelming preference for alors (78%) and moderate use of donc (20%) (Rehner et al., 2003). Thus, as the students progress in their learning via more years spent learning French and greater community-based exposure to the language, they move away from the norms of educational input and get closer to the FL1 norms of 18% donc and 3% alors. Another factor that may contribute to the university students’ increased use of donc at alors’ expense is that some are targeting the learning of a more formal register as they advance towards career decisions (F. Mougeon & Rehner, 2015). This may explain, in part, why the highly-engaged university learners are overusing grammatical and discursive donc. What is interesting with the case of the variable use of alors/donc/fait que is that community-based exposure brings about the rise of a formal variant (i.e., donc), unlike past research that has amply documented this beneficial effect on the acquisition of informal non-standard variants by L2 speakers (cf. Bayley et al., 2022; Howard et al., 2013).

In considering the marginal to non-use of fait que by the FL2 students in spite of its very high frequency in FL1 Montreal speech (grammatical = 74%, discursive = 82%), one can invoke again the impact of educational input, which features only 1% fait que in FI teachers’ in-class speech (R. Mougeon et al., 2010), whereas other non-standard variants such as ne deletion or subject pronoun on for nous are more frequent in such speech (Rehner & Mougeon, 1999; Rehner et al., 2003). As mentioned in the literature review, these FL2 learners are not alone in their marginal to non-use of this connector. Recall that the FL2 university students in Thunder Bay (Lemée, 2025) make no use of this form, and the FL2 speakers in Montreal (G. Sankoff et al., 1997) use it less than 10% of the time despite their increased opportunities to interact with FL1 speakers. These strikingly convergent findings across settings for FL2 speakers with differing levels of opportunities to interact in French suggests that some additional factor is at work. What may be at play is that fait que is often realized as a phonetically reduced form ([fak] or [fek]) in Canadian spoken French (cf. Blondeau & Tremblay, 2022 for Montreal French and Falkert, 2006 for Acadian French spoken in the Madeleine Islands). This reduction may lower the perceptual salience of this form for FL2 speakers, who may fail to hear it and consequently fail to associate this reduced form with fait que even when exposed to its use. In fact, some FL2 students in our own university classes were unable to hear this connector in videos played for them with the specific purpose of introducing them to this form. The weak perceptual salience of this connector suggests the need for targeted pedagogical interventions aimed at raising learners’ awareness of this variant and at providing them with information on its socio-stylistic status.

With respect to the decline in the use of so as time spent learning French and opportunities to interact with FL1 speakers increase, this pattern suggests a developmental trajectory for so. This is unlike the use of English discourse marker ‘like’ (versus its French equivalent comme) by the same FI students examined here (Rehner, 2004). In the case of ‘like’, increased interactions with FL1 speakers did not result in its decreased use, leading Rehner to conclude that it was not a developmental variant but rather an automatized feature of their English that they had difficulty suppressing in their spoken French.

5.3. The FL2 Students Respect Some of the FL1 Linguistic Constraints on Connector Choice

Our study has shown that in FL1 speech the grammatical function is favorable to the use of alors and donc, and the discursive functions are favorable to the use of fait que. Analysis of the FL2 students’ speech has revealed that each student group has acquired one the FL1 linguistic constraints and a different constraint in each case (the association of alors with the grammatical function for the FI students and the association of donc with the grammatical function for the university students). However, the FI students’ acquisition of the linguistic constraint for alors may be serendipitous and an indirect consequence of their non-negligible use of discursive so. As for the university students, they exhibit a reversed non-native constraint for alors (i.e., more frequent use of this connector when it fulfills the discursive functions than the grammatical one). Thus, while the university students exhibit frequencies of connector use that are closer to the FL1 norms than are those of the FI students, they do not have the edge in relation to the linguistic constraints on connector choice.

This less-than-complete acquisition of the linguistic constraints is very much in line with past studies of the FI students, or the university students, or other FL2 learners even when they have had extensive opportunities to be exposed to FL1 speech (see Blondeau et al., 2002). That said, when considering the few studies that allow for a direct comparison between the FI and university learners, the picture is more complex. The findings of Uritescu et al. (2004) versus Thomas (2002) allow for a comparison of these two groups’ acquisition of the linguistic constraints for schwa deletion, and those of Rehner and Mougeon (1999) versus Thomas (2002) for ne use/non-use. The comparison reveals that both groups acquired the hierarchy of FL1 phonetic constraints on schwa deletion, namely that the word internal context is much more favourable to deletion than is the context of monosyllabic words. In both of these contexts, the FL2 university students were closer to FL1 frequencies than were the FI students. In contrast, neither FL2 group learned the FL1 constraint on negation, namely that ne non-use is more frequent with postverbal negator pas than with other postverbal negators. While the university students’ rates of non-use were almost identical across the two contexts, the FI students demonstrated a non-native constraint with higher frequencies of ne non-use with postverbal negators other than pas. Thus, like the present study, the FI and university students have acquired some FL1 linguistic constraints but not others and have exhibited non-native linguistic constraints.

6. Conclusions

In line with the scope of this volume, the present study has focused on two groups of FL2 learners representing different stages in the acquisition of sociolinguistic variation. It is the first study where these two groups are compared systematically. While this comparison allows for important insights into the development of sociolinguistic competence, it is based on only one sociolinguistic variable. Thus, there is the opportunity to undertake this type of comparative research with other cases of variation. An opportunity also exists to expand this comparative analysis of the acquisition of sociolinguistic variation by placing these FL2 groups on a broader FL1–FL2 speaker continuum, facilitated by the availability of additional corpora collected with the same methodologies. Two of our previous studies, which examined the development of linguistic competence by focusing on the acquisition of difficult aspects of French morphosyntax, have made use of this FL1–FL2 speaker continuum (i.e., R. Mougeon et al., 2024; Rehner et al., 2022). In these studies, the FI and university students were compared with groups of bilingual FL1 speakers from majority and minority Francophone communities in Ontario who evidence differing degrees of dominance in French and English. This comparison has brought to light differences and similarities in the linguistic competence of student groups on this broader speaker continuum, revealing, for example, that in situations where French and English come into contact the concepts of FL1 and FL2 speakers are more relative than absolute. Thus, in the future we hope to expand our research on the development of sociolinguistic competence by investigating the variable use of the connectors and of other sociolinguistic variables across the FL1–FL2 continuum.