Agreeing to Disagree: Variable subject-verb agreement in Immersion French[1]

Terry Nadasdi, University of Alberta

This study presents a variationist analysis (cf. Sankoff, 1988) of subject/verb agreement in the third person plural, using data from students enrolled in a programme of French immersion in Ontario, Canada. The analysis considers both social and linguistic factors that may influence the use of either the syncretized or standard variants and makes comparisons between the behaviour of this variable among immersion students and native speakers of French. In so doing, the study will contribute to an understanding of the similarities and differences between the grammars of these two groups of speakers. The principle research questions addressed in this study are: a) do the immersion students use syncretized forms to a greater degree than native speakers?; b) do the immersion speakers share the same linguistic and social constraints as native speakers whose use of French is restricted (cf. Mougeon and Beniak, 1991)?

Corpus

The twenty speakers in the present study have been taken from Mougeon and Nadasdi's 1996 corpus of immersion French. All speakers are enrolled in extended French programmes where 50% of courses are taken in French from grades 5-8, followed by 20% in high school. While speakers are from various L1 backgrounds, none are native speakers of French and the school setting has been and remains their primary locus of French usage and learning. Students are from grades 9 and 12 and are from homes where neither parent is a native speaker of French and where French is not spoken. Although these students are not from French-speaking homes, they are by no means all from unilingual Anglophone homes. In fact, 51% of our subjects come from homes where a language other than English is used to varying degrees. Of these students, 39% come from homes where a Romance language is spoken and the rest are from non-Romance language homes. There are approximately the same number of grade 9 and 12 students, more females than males, and over half are from middle class families with all but one of the remaining being from lower middle class families. Most of the students have received between 26-37% of their schooling through the medium of French. The majority of students never use the spoken French media; however, there are more grade 12 students than grade 9s who do so occasionally. The grade 12 students have also spent more time in Francophone environments and with Francophone families than is the case for the grade 9 students. These stays in a Francophone environment or with a Francophone family are, for the most part, in Quebec.

Previous studies of variation in immersion French

Previous research on variation in the spoken French of immersion students has concentrated on the use (or none use) of both vernacular and informal variants in students' spoken discourse (cf. Swain and Lapkin, 1990; Tarone and Swain, 1995; Rehner and Mougeon, 1999; Mougeon, Nadasdi and Rehner, 1999). These studies have show that vernacular variants are almost never used by immersion speakers and that while immersion students do make some use of informal features, for example deletion of the preverbal negative particle ne, the frequency of such features is much less than that found in the discourse of native Francophones. However, there exists no study to date which concentrates on morphological variation in the spontaneous speech of immersion students that does not involve a standard/non-standard or formal/informal split between variants in the speech of native Francophones. The present study concentrates on a variable of this type.

The Variable

The variable understudy concerns the alternation between explicit third person plural verbal forms and syncretized verb forms devoid of plural marking. While a great many French verbs are homophonous in the third person singular and plural, a number of frequent and irregular verbs explicitly mark person in the third person plural by means of a morphological alternation. This can take the form of complete suppletion, as in the case of être, for example il est/ils sont, denasalization, e.g. il vient/ils viennent, change in final vowel quality, e.g. il va/ils vont, the adding of a final consonant, e.g. il dit/il disent or a combination of these last two processes, e.g. il sait/ils savent. Examples from our corpus which illustrate both the non-sycretized (i.e., standard) and syncretized variants are given below:

non-sycretised variant

1) Tous les parents disent quelque chose que les enfants n'aiment pas. 'All parents say something that children don't like'.

syncretized variant

2) Les personnes … dit que … 'People … say that …'.

Previous studies of third person plural syncretism in L1 French

As discussed in Mougeon and Beniak (1995), syncretism in the third person plural has been analyzed in a number of studies concentrating on native speakers of European French (cf. Bauche, 1920 and Frei, 1929), and Canadian French (cf. Flikeid, 1985 and King, 1993). In all of these studies, the syncretized variant is relatively rare, except after the relative pronoun qui or the personal pronoun ils. For example, Mougeon and Beniak (1995) report that in the speech of unrestricted speakers[2], the syncretized variant is rare (2% of tokens) and produced exclusively in the above mentioned linguistic contexts, i.e. after qui or ils. Furthermore, it is particularly the relative pronoun that gives rise to the syncretized variant. On the other hand, speakers whose French language use is restricted make greater use of the syncretized variant (19%) and do not follow these same linguistic constraints. In other words, these latter speakers use the syncretized variant in all linguistic contexts. And, unlike the unrestricted speakers, syncretism is much more likely to occur with low frequency verbs in the speech of restricted speakers. According to Mougeon and Beniak, (1995:54), third person plural syncretism in the speech of restricted speakers results from imperfect mastery of French verb forms due to infrequent use of French. As concerns our immersion speakers, several predictions can be made. For example, given that the immersion speakers use French even less frequently than Mougeon and Beniak's restricted speakers[3], we can expect to find an even higher incidence of syncretism in their oral discourse. This seems likely given the tendency of L2 speakers to regularized complex morphological structures. Alternatively, it may be that for the structure in question the amount of exposure received is sufficient to produce frequencies of subject-verb agreement that are in line with those of restricted native speakers. This would not be a surprising result given Mougeon and Nadasdi's 1999 findings regarding the alternation of je vais/je vas in immersion French. This study shows that immersion speakers rarely regularize the first person singular form and that the speakers "have not experienced a great deal of difficulty in mastering the irregular 1sg vais form". Our results will help shed light on the relative complexity of subject-verb agreement in the third person plural in comparison with the je vais/je vas alternation.

Previous studies of third person plural syncretism in L2 French

One study that has examined third person plural agreement in L2 French is Harley's 1982 study of early and late immersion students. Harley reports that immersion students scored significantly lower than the Francophone comparison group. The highest levels of agreement were found in the late immersion speakers. However, even in this group syncretised forms were found in 70% of occurrences. In Harley's study, no distinction was made between frequent verbs and infrequent ones, which will make comparisons difficult with the present study.

Linguistic Factors

The primary linguistic factors to be considered in the present study are: a) type of subject and b) verb frequency. As mentioned, those few cases of syncretism found in the speech of unrestricted Francophones occurred after ils and qui. Our study will provide additional information regarding the role of this factor by considering L2 data. We hope to determine whether or not the same qualitative difference which distinguished the occurrence of syncretism in restricted and unrestricted Francophone speech obtains in the immersion speakers' French. For the second linguistic factor group, i.e. verb frequency, we use Mougeon and Beniak's division which places avoir, être and aller in the category of frequent verbs (37%, 34%, and 9% respectively). The category of infrequent verbs includes all others verbs having two morphologically distinction forms for the third person singular and plural, for example, dire, venir, devoir, etc. (none of these verbs constitute more than 4% of tokens). Given that this factor exercised a significant effect for restricted Franco-Ontarians (cf. Mougeon and Beniak, 1995), we anticipate that it will be selected as influencing the variable in the speech of immersion students.

In our analysis of linguistic factors, we have also examined environmental elements not controlled for in Mougeon and Beniak's study. The first factor we considered was the presence of an element intervening between the subject and the verb, as in example 3):

3) Ils ne veut faire rien. 'They don't want to do anything'.

The hypothesis underlying the inclusion of this factor group is that agreement will occur less frequently when the subject is not immediately adjacent to the verb since the link between the two elements has been broken.

The fourth linguistic factor group considered allows us to ask the following: does the presence of an overt plural marker on the subject lessen the likelihood of marking this same information of the verb, or on the other hand is there a priming of effect such that marking of overt plurality on the subject increases the likelihood of marking verbal plurality? In the category of subjects containing an overt plural marker we include all lexical subjects preceded by an article (des or les) or a quantifying adverb, such as beaucoup, plusieurs, etc. We have also included cases where a subject pronoun is pronounced /ilz/ as is sometimes the case before vowels AND consonants, for example:

4) [ilzvaparle] "ils vont parler". 'They will speak'.

This factor group was considered in Mougeon and Beniak's original (1991) study of third person plural syncretism. Their inclusion of this factor group stems from the functionalist hypothesis that syncretism "might be blocked or at least significantly reduced when plurality is not overtly marked in the subject, failing which, singularity rather than plurality would be conveyed". Although this factor group did not produce a significant effect on the variable in Mougeon and Beniak's study of Francophones, we have included it in our analysis since it may be relevant for second language speakers.

Social Factors

As mentioned, previous sociolinguistic analyses of the spontaneous spoken French of immersion students have concentrated on variables that correlate with speakers' sex and SES in L1 speech. The results of these studies have shown that students display patterns of sex and/or social class stratification that are comparable to those of Canadian Francophones. The explanation proposed by Mougeon and his associates is that students

infer the sociostylistic value of the variants on the basis of their teachers' usage (i.e. what they prefer and use in class, what variants they reinforce, what variants they self-correct in their speech or in that of their students and what variants are used in teaching materials). While there is no evidence that the syncretized and non syncretized variants are distributed according to a vernacular/standard split, there is reason to believe that social factors may play a role in the immersion corpus since sex/gender has often been cited as an important variable in SLA studies. For example, if we consider the various studies that have found that girls score higher than boys in measurements of L2 achievement (cf. Burstall, 1975; Boyle, 1987) we might expect to find higher rates of the non-syncretized variant among female students, which would constitute evidence of greater mastery of French morphology by these latter students. It needs to be pointed out, however, that results concerning sex differences and SLA are often conflicting (cf. Ellis, 1994; Ehrlich, 1997). There also exist a number of studies that suggest no difference between the two, or that it is boys who have the advantage (cf. Buegel and Bunk, 1996). The vast majority of studies that have found differences in either direction, concentrate on test data. The present study hopes to contribute to the findings in the area of sex/gender difference in SLA by examining results from boys' and girls' spontaneous L2 production.

In addition to examining sex and social class, we will consider the role of extra-scholastic contacts with native speakers. The corpus used for the present studies controls for this variable since students have indicated the number of days, weeks, etc. spent in a native Francophone environment. Since the variable has been correlated with verb frequency in the speech of restricted Franco-Ontarians (the greater the verbs' frequency, the more likely it will give rise to the non-syncretized variant), it can be hypothesized that speakers having more contact with native speakers, and hence who receive a greater amount of input, will use the syncretized variant less often than those who have less contact with native speakers.

Results

Results for the general distribution of the variants are presented in Table1a:

Table1a. Distribution of variants

	Number	Percentage
Syncretized variant	118	19%
Non-syncretized variant	490	81%

As we can see, the syncretized form is relatively rare in the spoken discourse of French immersion students. In other words, students do not display a great deal of difficulty with subject verb agreement. This is true in general terms, and also in comparison with native Francophones since the frequency of the syncretized variant in the immersion corpus is close to that found in the corpus Ontario French where it accounts for 12% of all tokens. In fact, one finds fewer syncretized forms in the immersion corpus than one does in the speech of Franco-Ontarians living in the minority language community of Pembroke where this variant accounts for 27% of all tokens! (cf. Mougeon and Beniak, 1995).

The results presented in Table1a suggest that the presence of third person plural agreement in Immersion French is similar to that found in native speaker French, particularly when considering the speech of Francophones whose language use is restricted. This results is based on the assumption that third person plural forms are evidence of third person plural agreement. Such an assumption is unproblematic in native speaker French, but is less straightforward when working with interlanguage data. As pointed out by Corder (1967) and Gass and Selinker (1994), L2 speakers are capable of producing target like forms which do not necessarily reflect knowledge of a particular target language rule. We will therefore consider the validity of this assumption by revisiting the data and examining not only third person plural, but other verb forms as well.

Closer scrutiny of the data suggest that while many students do have distinct forms for third person singular and plural, this is not always the case. Consider the data in 5) from speaker 33:

5) quand ils ont fait la confirmation … il[z] doivent aller au … au église pour deux années toutes le[z] dimanches. 'When they have done confirmation they must go to church'.

At first blush, this exercpt suggests that speaker 33 has mastered the rule of plural agreement for the verb devoir since the plural form doivent is used (and not the singular doit). However, other data from the same speaker reveal that this may not be the case. Consider the data in 6), also taken from speaker 33:

6)

a) je doive[4] parler parce que j'ai une amie qui parle seulement l'Italien alors quand je l'ai à la maison je doive parler l'italien à lui. 'I have to speak because I have a friend who only speaks Italian so when I have him at the house I must speak Italian to him'.

b) Si tu peux, tu doives donner d'argent pour faire les recherches. 'If you can, you must give money to do research'.

c) elle se doive trouver un autre travail. 'She has to find herself another job'.

d) alors il va mourir alors une autre personne l'a tué après il doive aller à un autre place comme un autre pays. 'so he is going to die so another person killed him, after he must go to another place, like another country.'

e) maintenant nous sommes finis et nou[z] doive fait un examen. "Now we are finished and we must do an exam'.

Each of these examples contains an anomolous verb form homophonous with the third person plural. They suggest that the speaker has a single finite form for the verb devoir, which is inconsistent with all varieties of L1 French. These data are important since they suggest that those cases where the student pronounced ils doivent may not be bona fides examples of third person plural morphology. As such, they should be excluded from quantitative analysis for third person plural agreement.

A similar problem is found in the interview of speaker 35:

7) C'est comme de[z] enfants … ils vendent des drogues et toute ça et ils va à l'école … ils battent avec les autres personnes et … le directeur directrice de l'école elle … quand l'école est fini elle va à une restaurant et elle comme vende le[z] drogues et tou[t] ça. 'It's like some kids … they sell drugs and all that and they go to school... the fight with other people and … the principal of the school she … when school is over, she goes to a restaurant and she like sells drugs and all that'.

As was the case with speaker 33, speaker 35 provides little evidence of an agreement rule for the verb vendre (or aller for that matter)[5]. In Standard French, vendre presents distinct forms in the third person plural singular and plural since the former ends with a nasal vowel while the latter ends with a voiced dental consonant (il vend/ils vendent). As such, the target-like ils vendent should be excluded from our analysis.

The following table contains revised figures, after having excluded for which there was no evidence of a singular/plural distinction.

Table 1b. Distribution of variants (revised)

	Number	Percentage
Syncretized variant	118	20%
Non-syncretized variant	474	80%

This was arrived at by removing third person plural forms which were homophonous with the singular, for example il doive/ils doivent, elle vende/elles vendent, il peuve/ils peuvent, etc. The criteria used for this was that if there was not evidence in other verbal persons of the standard singular form (in the first, second or third person, e.g. je pars, tu dois, etc.), the third person-like plural form was excluded. While the general percentages have not greatly changed, the revised table does exclude 16 occurrences that were previously considered instances of 3^rd person plural agreement. All the cases of excluded instances involved infrequent verbs, in other words, immersion students never use suppletive 3^rd person plural form with singular subjects, that is, the corpus contains no occurrences such as *je sont".

Linguistic factor results

Let us now consider the results according to verb frequency presented in Table2:

Table 2. Use of syncretized forms according to verb frequency

Verb frequency	N	Total	Percentage	Factor effect
High	33	458	7%	.333
Low	85	134	63%	.915
Totals	118	474	20%

As indicated in Table2, verb frequency is an important conditioning factor for the variable: frequent verbs are unlikely to give rise to the syncretized variant, infrequent verbs show a high incidence of syncretism. As such, the immersion speakers follow the same rule as the restricted Franco-Ontarians, and differ from unrestricted Franco-Ontarians since this factor is only selected as significant for the former group. Also of note, immersion speakers evidence the same level of syncretism with frequent verbs in comparison with restricted Franco-Ontarians since both groups of speakers syncretize 7% of frequent verbs. On the other hand, the results in Table2 suggest that in the case of infrequent verbs, the immersion students slightly "outperform" the restricted Franco-ontarians from Pembroke. Mougeon and Beniak (1995) report that these speakers use the syncretized variant in 65% of occurrences whereas as the immersion speakers only use it in 63% of tokens with infrequent verbs.

Let us next consider the role of type of subject on the variable in immersion French. These results are presented in Table3:

Table 3. Use of syncretized forms according to subject type

Subject type	N	Total	Percentage	Factor effect
Lexical NP	47	191	25%	NS
Ils	57	323	18%	NS
Qui	14	74	19%	NS

Results for subject type show that, once again, immersion students resemble restricted Franco-Ontarians since this factor group is not a significant predictor of variation for either group. Unlike the unrestricted speakers, immersion speakers are not more likely to syncretize with subject relative pronouns. In other words, they are not subject to the same vernacular linguistic constraint found in that speech of unrestricted Francophones.

Consider next results for explicit plural marking on the subject which are presented in Table4:

Table 4. Use of syncretised forms when plurality is explicitly marked on the subject

	N	Total	Percentage	Factor effect
Overt plurality	62	219	28%	.608
Non overt plurality	56	373	15%	.436

As mentioned, this factor did not exercise a significant effect for any of the Franco-ontarian students considered in Mougeon and Beniak's 1991 study. However, it does influence variant choice in the immersion corpus. These results support the functionalist hypothesis originally entertained by Mougeon and Beniak: subjects which do not explicitly mark plurality are less likely to give rise to the syncretized variant. It would appear then that while the immersion students do mark plurality in the clause, they tend not to do so redundantly.

Results concerning the role of elements intervening between he subject and the verb are presented in Table5:

Table 5. Use of syncretised forms when an element separates the subject and verb

	N	Total	Percentage	Factor effect
Separating element	44	122	36%	.647
No separating element	74	470	16%	.461

Like the presence of explicitly plural subject, the presence of an element separating the subject and the verb, for example an object pronoun, or the negative particle ne, promotes the syncretized variant. In other words, when there is a rupture of the link between the subject and the verb, the likelihood of agreement decreases. This factor group was also considered in Mougeon and Beniak's (1991), however it was not shown to exercise a significant effect on variant. This suggests that the constraint in question is particular to the immersion students` interlanguage. It should be pointed out, however, that Mougeon and Beniak did not consider this factor group separately for each level of language restriction. It may indeed have been found to exercise a significant effect when only the restricted speakers were considered. As shown by Mougeon and Nadasdi 1998, subgroups within the Franco-Ontarian speech community do not share all the same linguistic constraints (cf. Mougeon and Beniak, 1995).

Social Factor Results

Only one social factor exercises a significant independent effect on variation, namely French medium instruction. Results for this factor group are presented in Table6:

Table 6. Use of syncretised according to amount of French language instruction

	N	Total	Percentage	Factor effect
Less than 25% French instruction	37	132	28%	.660
More than 25% French instruction	81	460	18%	.453

Initially, subjects were divided into three categories of French medium instruction: a) 0-25%; b) 26-37% ; c) 38%-100%. While the factor was selected, the division between that latter two groups was not significant and, as such, they have be collapsed into one category. These results suggest that syncretism greatly increases below a certain threshold of French language instruction/exposure. While amounts above the 25% threshold do not seem to influence the variable, those below it do. This is particularly true for infrequent verbs:

Table7. Cross tabulation of the effect of level of instruction and verb frequency on verbal

syncretism

	frequent verbs	infrequent
Less then 25% instruction	111/101 (11%)	26/31 (84%)
Greater than 25 % instruction	22/357 (6%)	59/103 (57%

As revealed in Table7, it is the infrequent verbs that cause the greatest difficulty for those having low levels of French language instruction. It needs to be borne in mind that there is nearly perfect overlap between frequency and morphological complexity. The frequent verbs are, for the most part, the suppletive verbs avoir and être. The third person plural of these forms does not involve a complex morphological rule applied to the singular form. Rather, they are distinct and no doubt learned as separate lexical items. The infrequent verbs, on the other hand, all involve some type of morphological process which relates the singular and plural forms. It may well be this factor and not the verb’s frequency that is the most influential factor. However this is difficult to disentangle given the overlap between frequency and morphological complexity.

Students’ extracurricular exposure to French was also considered however this factor was not selected as significant for verbal syncretism. This suggests that the few weeks of extra exposure to French in Francophone settings are insufficient to have a positive effect on the mastery of subject-verb agreement although this factor has had some positive impact on the acquisition of informal variants, for example the use of on vs. nous (see Mougeon et al., 1999). This may suggest that lack of sociostylistic saliency with the variable understudy neutralizes the effect of extracurricular exposure to French.

As concerns sex and L2 variation, we can conclude that for the present variable at least, sex does not seem to be a relevant factor when considering spontaneous oral discourse. The fact that sex and social class were not selected for the present variable, but have been often found to correlate with variables having a standard and non standard variant suggest that the sex difference is not one of proficiency. Previous studies using the same corpus have found sex to correlate with the use of standard features when the French to which students are exposed contains both a non standard and a standard feature, with the standard feature being most prevalent in the input, usually in the speech of classroom teachers (cf. Mougeon, Rehner and Nadasdi, 1999). Our results suggest that this is not the case when one of the variants is found categorically in the input, as would be the case with the non syncretized form of the third person plural variable[6].

Conclusion

The first finding of our study is that third person plural syncretism in the immersion corpus is similar to what is found in the speech of restricted Franco-Ontarians (20% and 19% respectively). This is true for both frequent and infrequent verbs. One key difference, however, is that a number of speakers in the immersion corpus use third person plural forms for other grammatical persons (e.g.: je doive).

Concerning linguistic factors, we have found that immersion speakers share no linguistic constraints with the unrestricted Francophones, they share the constraint of verb frequency with the restricted Francophones and they alone are influenced by the factors of overt plural marking on the subject and the presence of an element separating the subject and the verb. The findings concerning linguistic factors are summarized in the following table:

Table. 8 Summary of 3pp syncretism along the French language proficiency continuum

Ling factor	Immersion	Restricted franco	Unrestricted franco
Subject type			X
Verb frequency	X	X
Subject plurality	X
separating element	X

As concerns social factors, our result show that neither SES nor exercises a significant effect on the variable. While some studies have found sex related differences involving an alternation between L1 and interlanguage forms, our study is unique inasmuch as the results have not been obtained using test data, but rather from spontaneous L2 speech. The only social factor that was selected is level of French language education. Our results show that students having received less than 25% of their schooling in the target language have considerably higher levels of syncretism, particularly with infrequent verbs.

Given that few social factors correlate with subject-verb agreement, the present variable is a special case that differs from other variables studied using the same corpus of immersion students (cf. Mougeon et al., 1999). Subject-verb agreement is not a "classic" case of variation involving a salient vernacular variant that alternates with a standard form. This may help explain the absence of a quantitative difference between the immersion students and the restricted Franco-Ontarian students. In other words, the case of variation understudy is symptomatic of a developmental lag in the mastery of the third person plural distinctive forms and not of learning a vernacular variant

References

Allen, P., Cummins, J., Harley, B. and Swain, M. (1987). Development of Bilingual Proficiency Project. Toronto: OISE/UT.

Boyle, J. 1987. "Sex differences in listening vocabulary". Language Learning 37: 273-284.

Buegel, K. and B. Bunk. 1996. "Sex differences in foreign language text comprehension: the role of interests and prior knowledge". The Modern Language Journal 80: 15-31.

Burstall, C. 1975. "Factors affecting foreign-language learning: a consideration of some relevant research findings". Language Teaching and linguistics Abstracts 8: 105-25.

Ehrlich, Susan. 1997. "Gender as Social Practice: Implications for Second Language Acquisition". Studies in Second Language Acquisition, 19:421-446.

Ellis, R. 1994. The Study of Second Language Acquisition. Oxford: Oxford University Press.

Harley, Birgit. 1986. Age in Second Language Acquisition. Clevedon: Multilingual Matters.

Mougeon, R. and E. Beniak. 1991. The Linguistic Consequences of Language Contact and Restriction: the Case of French in Ontario, Canada. Oxford: Oxford University Press.

Mougoen, R. and E. Beniak. 1995. "Le non-accord en nombre entre sujet et verbe en français ontarien: un cas de simplification?". Présence Francophone 46: 53-65.

Mougeon, R. and T. Nadasdi. 1998. "Sociolinguistic discontinuity in minority language communities". Language 74.1. 40-55.

Mougeon, R. and T. Nadasdi. 1998. "Use of Analytic and Synthetic Verb Forms to Express the Future Time in the Spoken French of High School French Immersion Students. Paper read at Trends in Second Language Teaching and Learning, Ottawa.

Mougeon, R., K. Rehner and T. Nadasdi. 1999. "Variation in the spoken French of Ontario Anglophone French Immersion Students". Paper presented at NWAV 28, Toronto.

Rehner, K. & Mougeon, R. 1999. Variation in the spoken French of immersion students: To ne or not to ne, that is the sociolinguistic question. Special Issue of the Canadian Modern Language Review.

Sankoff, D. 1988. "Sociolinguistics and syntactic variation". In F. Newmeyer (Ed.), Language: The Socio-cultural context . Cambridge: Cambridge University Press.

Swain, M. and Lapkin, S. 1990. "Aspects of the sociolinguistic performance of early and later French Immersion students". In R.C. Scarcella, E.S. Anderson and S.D. Krashen (Eds.), Developing communicative competence in a second language (pp. 41-54). New York: Newbury House.

Tarone, E. and Swain, M. 1995. "A sociolinguistic perspective on second language use in immersion classrooms". Modern Language Journal, 79, (2), 166-178.

---

II would like to thank Raymond Mougeon for his helpful comments on a previous version of this article. I would also like to thank Amy Gerald for helping code the data.

[2] While all speakers studied by Mougeon and Beniak are native Francophones, they can nonetheless be divided according to their language use patterns because they reside in localities where Francophones are a minority and are bilingual in English. Three categories are distinguished: a) unrestricted speakers, i.e. those who make almost exclusive use of French; b) semi-restricted speakers, i.e. those who use English and French to a similar degree; c) restricted speakers, i.e. those who use English more frequently than French.

[3] That is, those speakers whose use of French is limited or restricted to a small number of conversational domains.

[4] One explanation for this finding may be that students` acquisition of the subjunctive, which in many cases is homophonous with third person plural, has led them to generalize to the indicative. The fact that devoir is a deontic verb may further contribute to the confusion.

[5] One must still bear in mind that in many cases, the immersion students do evidence a rule of third person plural agreement by adding a consonant to the open syllable of the third person singular form.

[6] We have confirmed the absences of syncretised forms in the input using Allen et al.'s corpus of immersion teachers' speech.