2.1 Variationist studies of the French subjunctive mood

Over the past few decades, the subjunctive mood has been studied in a number of varieties of French from a variationist sociolinguistic perspective. One of the first variationist studies focuses on Ontarian French (Laurier, Reference Laurier, Mougeon and Beniak1989) where subjunctive usage is described according to contexts prescribed by proponents of the standard language. This entailed that subjunctive use in Ontarian French is defined in terms of the (often arbitrary) rules set forth by grammarians and commentators rather than by the local community norms. For example, Laurier (Reference Laurier, Mougeon and Beniak1989: 112) considers whether the subjunctive occurs with certain verbs prescribed by the standard, such as convaincre que ‘to convince’. In fact, the subjunctive mood is not attested in any variety of French spoken in Canada with convaincre que, as pointed out in a later study by Grimm (Reference Grimm2015: 94). Therefore, by comparing the local Ontario French vernacular against an external standard, Laurier (Reference Laurier, Mougeon and Beniak1989) ultimately concludes that the subjunctive mood is undergoing a change in progress, being gradually replaced by other moods such as the indicative and conditional. This envelope of variation was again taken by Neumann-Holzschuh (Reference Neumann-Holzschuh, Brasseur and Falkert2005) for Louisiana and Acadian French, who likewise concludes that the subjunctive mood is gradually being lost. Evidently, studies such as these, which compare use of the subjunctive in the local variety with the standard, can lead analysts to assume that the subjunctive is lacking in its productivity or that it is being lost since it is not found in numerous prescribed contexts, such as after convaincre que or after verbs of volition like vouloir ‘to want’, as in (2) taken from Neumann-Holzschuh (Reference Neumann-Holzschuh, Brasseur and Falkert2005: 135, example 80).

1. (2)

   

In addition, studies which adopt a variable context based on the standard will often identify contexts which show usage of the subjunctive where the standard does not require the subjunctive, such as following après que ‘after’, as in (3) taken from Laurier (Reference Laurier, Mougeon and Beniak1989: 116, example 5).

1. (3)

   

In a study of mood alternation in the French spoken in Ottawa-Hull, Canada’s national capital region, Poplack (Reference Poplack, Laeufer and Morgan1992) proposes that the variable context for the French subjunctive be based on local community norms rather than on the standard language. This represented a major methodological shift in analysing the subjunctive mood in that the local variety would then be measured not against some external and arbitrary norm, but rather against the community’s own local norms. Poplack’s approach consisted of first extracting all tokens which show subjunctive verbal morphology at least once in her large Ottawa-Hull sociolinguistic corpus and then establishing a list of the main clause verbs (falloir ‘to be necessary’, vouloir ‘to want’, etc.). Following this step, all tokens, whether they involve subjunctive morphology or not, which occurred in an embedded clause following a subjunctive-selecting main clause verb were then extracted, giving us a locally-relevant picture as to how the subjunctive occurs in this variety of French.

In a more recent article, Poplack, Lealess and Dion (Reference Poplack, Lealess and Dion2013) present a rigorous analysis of mood variation in three corpora of contemporary Laurentian French as well as an analysis of grammarians’ commentaries on the subjunctive mood.Footnote ⁵ Their comparison of the overall rate of subjunctive use across the three corpora shows a surprising increase, not a decline (contra Laurier, Reference Laurier, Mougeon and Beniak1989; Neumann-Holzschuh, Reference Neumann-Holzschuh, Brasseur and Falkert2005). They ultimately argue that the subjunctive does not function to express particular meanings, but that it occurs with a small set of main clause elements (verbal and non-verbal) and with a small set of embedded clause verbs. Thus, they conclude that ‘the overriding constraint on subjunctive selection, in Quebec French at least, is lexical’ (Poplack, Lealess and Dion, Reference Poplack, Lealess and Dion2013: 170). This conclusion stands in stark opposition to centuries of attributing various semantic functions to the subjunctive mood in French. To account for this discrepancy, they argue that this is due to grammarians’ (and theoretical linguists’) goal of establishing form-function symmetry between the subjunctive morphology and its purported functions. Essentially, by forcing symmetry between a linguistic form and a function, grammarians and theoretical linguists are excluding the possibility of inherent variability (Labov, Reference Labov1972). Thus, the findings for contemporary Laurentian French by Poplack and her colleagues lead them to conclude that the subjunctive ultimately does not have a semantic function. In fact, Poplack (Reference Poplack, Laeufer and Morgan1992: 257) states that ‘[w]e do not wish to imply that the subjunctive mood has lost its meaning in embedded noun clauses in Canadian French. The available evidence […] suggests that it never had one, at least since the time of Classical Latin’.

In more recent work, Grimm (Reference Grimm2015) considers the subjunctive mood in two corpora (collected in 1978 and in 2005) of Ontarian French spoken by adolescents in four communities: Cornwall, Hawkesbury, North Bay and Pembroke.Footnote ⁶ While these communities differ in terms of relative proportion of francophones at the community level, Grimm (Reference Grimm2015) also considers variation in speakers’ daily use of French, referred to as language restriction.Footnote ⁷ His results show that falloir is by far the most frequent subjunctive-selecting context, as was found in other spoken varieties of Laurentian French. However, he uncovers a correlation whereby the total number of tokens of falloir (regardless of the mood choice in the embedded clause) decreases as the population of francophones in the community decreases. Thus, in the communities with fewer francophones, they are simply using falloir (the most frequent subjunctive-selector) less than in other communities. To better understand this pattern, Grimm (Reference Grimm2015) then considers one of the main functions of falloir, namely to express necessity, in order to see how speakers might be expressing this function differently. He found that speakers in the communities with fewer francophones (and speakers who are more restricted in their use of French) use devoir ‘to have to’, a deontic modal verb which selects not a finite clause, but rather an infinitive, as a complement (e.g., je dois manger ‘I need to eat’), instead of falloir.Footnote ⁸ According to Grimm (Reference Grimm2015), the avoidance of falloir by communities with fewer francophones is motivated in part by avoiding to conjugate a verb in the subjunctive, a more morphologically complex option. The use of devoir accomplishes this since the following verb is in its infinitival form. In his detailed analysis across communities, Grimm (Reference Grimm2015) notes that use of devoir to express necessity increases (and so falloir conversely decreases) as the relative proportion of francophones decreases at the community level and use of devoir likewise increases as level of language restriction increases at the individual level. The consideration of the expression of necessity (rather than variation in mood choice) is an important methodological point to which I will return in Section 5.2.

An analysis of the expression of necessity, along with an analysis of mood choice, is also conducted by Kastronic (Reference Kastronic2016) in a comparative study between Laurentian and Hexagonal French. Regarding the mood choice variable, Kastronic (Reference Kastronic2016) reports a slightly lower overall rate of subjunctive in Hexagonal French than in Laurentian French, but finds similar patterns in terms of the lexical effects in both varieties (as found by Poplack, Lealess and Dion, Reference Poplack, Lealess and Dion2013). However, one notable difference is that she finds that Hexagonal French speakers use falloir much less frequently than in Laurentian French, where it accounts for most of the subjunctive-selecting contexts. In a similar vein to Grimm’s (Reference Grimm2015) approach, she then analyses how Hexagonal French speakers express necessity, but she considers a larger set of variants and finds that falloir followed by an infinitive (e.g., il faut manger ‘it is necessary to eat’) accounts for most of the tokens to express necessity. Regarding the mood choice variable, she ultimately concludes that subjunctive use is variable and is largely constrained by lexical effects.

A recent article by King, LeBlanc and Grimm (Reference King, LeBlanc and Grimm2018) compares subjunctive use in contexts with the main clause verb falloir ‘to be necessary’ in five varieties of Acadian French spoken in Canada’s Atlantic Provinces. Their analysis reveals that mood choice is variable across varieties, with rates of the subjunctive ranging from 73 per cent in Abram-Village, Prince Edward Island to 90 per cent in Ⓘles-de-la-Madeleine, Quebec. The one exception to this range (73–90%) is the low rate of 32 per cent found in L’Anse-à-Canards, Newfoundland. To account for the divergent finding for L’Anse-à-Canards, they highlight that an important proportion of the community is composed of Metropolitan French speakers who had immigrated from Brittany in the nineteenth century. Based on historical sources, such as the Atlas linguistique de la France (Gilliéron and Edmont, Reference Gilliéron and Edmont1902–Reference Gilliéron and Edmont1910), King, LeBlanc and Grimm (Reference King, LeBlanc and Grimm2018) show that the areas of Brittany from where the L’Anse-à-Canards settlers originated had highly variable subjunctive usage at the turn of the twentieth century. This contrasts with the areas which were the main points of origin of Acadians (namely, the centre-west regions of France) where subjunctive use is categorical in the Atlas linguistique de la France. Therefore, they argue that dialect contact with speakers who had variable subjunctive use in the case of L’Anse-à-Canards accounts for the different patterns. Furthermore, this study shows that the subjunctive varies with the indicative mood in many varieties of Acadian French, which will provide a striking contrast once we consider the results for Baie Sainte-Marie.

3.1 The Baie Sainte-Marie community

Acadian French refers to varieties spoken primarily in Canada’s Atlantic Provinces (New Brunswick, Newfoundland and Labrador, Nova Scotia, and Prince Edward Island) and in eastern parts of Quebec. There are two varieties of Acadian French spoken in southwestern Nova Scotia and both are highly conservative, having preserved a number of linguistic features lost in most other spoken varieties of French, including the varieties of Acadian French spoken in New Brunswick which currently has the greatest population of Acadian French speakers within the Atlantic Provinces. These two southwest Nova Scotia varieties, the Baie Sainte-Marie variety, spoken in the Municipality of Clare, and the Par-en-Bas variety, spoken in the Municipality of Argyle, retain a number of linguistic features lost in most spoken varieties of French (Flikeid, Reference Flikeid, Mougeon and Beniak1994). The present article will focus on the variety spoken in the Baie Sainte-Marie region, shown in Map 1.

Map 1. Acadian regions of Nova Scotia shown in grey.

In the case of Baie Sainte-Marie, centuries of relative isolation from external varieties of French as well as a fairly homogeneous settler population resulted in a community with very little dialect contact (Comeau, King and LeBlanc, Reference Comeau, King and LeBlanc2016). This sociohistorical context may account in part for the number of conservative features which have been preserved in this variety. For instance, the first-person plural form of je…–ons as well as the third-person plural ils…–ont, are shown in (4) and (5) respectively.

1. (4)

   

2. (5)

   

Conservative features related to the tense system involve productive use of the simple past tense (Gesner, Reference Gesner1979; Comeau, King and Butler, Reference Comeau, King and Butler2012), a tense lost in most spoken varieties of French. The fact that this variety retains a richer verbal morphology than varieties of Laurentian French (and some varieties of Acadian French) has repercussions on the study of the subjunctive mood, to be discussed in greater detail below.

The data used in this article come from two sociolinguistic corpora both collected in the early 1990s: the Butler Grosses Coques Sociolinguistic Corpus and Le corpus du français acadien de la Nouvelle-Écosse.Footnote ¹¹ Both corpora involve audio recordings collected by native speaker community members with the goal of collecting conversational and narrative data.Footnote ¹² The data analysed here from the Butler Grosses Coques Sociolinguistic Corpus were collected in 1990 and are from 22 speakers from the village of Grosses Coques. While Le corpus du français acadien de la Nouvelle-Écosse involves data for four Acadian communities in Nova Scotia (Baie Sainte-Marie, Chéticamp, Par-en-Bas and Isle Madame), only data for Baie Sainte-Marie (11 speakers in total) are retained for analysis. Furthermore, the data from this second corpus are from speakers of the village of Meteghan, situated at some remove from Grosses Coques (approximately 20 km). Since the data came from two different villages (and since Baie Sainte-Marie Acadians are capable of linguistically distinguishing speakers between villages), the data were initially analysed separately, but were ultimately combined since no discernable differences could be observed in the two data sets for this linguistic feature. Data from interviews for both women and men and from a wide age range (18–84) were extracted and form the basis of the analysis presented in this article. Taken together, the data represent the speech of 33 speakers of Baie Sainte-Marie Acadian French.

In most varieties of Laurentian French (Poplack, Reference Poplack, Laeufer and Morgan1992; Grimm, Reference Grimm2015), approximately half of all tokens of the subjunctive variable are excluded from the analysis due to ambiguous morphological forms (i.e., we cannot determine whether they are indicative or subjunctive mood), as shown in (6) taken from Grimm (Reference Grimm2015: 115).

1. (6)

   

For the class of –er verbs (e.g., manger ‘to eat’, danser ‘to dance’, parler ‘to speak’), only verbs conjugated with a first-person plural nous and second-person plural subject vous are morphologically distinct for mood and can be retained for analysis.Footnote ¹³ However, the Baie Sainte-Marie variety does not have the same degree of ambiguous forms, due to the retention of traditional conjugations first-person plural je…–ons and third-person plural ils…–ont, as shown in Table 1.

Table 1. Verbal tense paradigms for –er verbs in Baie Sainte-Marie Acadian French

The retention of these conjugations entails that for the class of –er verbs, it is only the singular forms (the cells shaded in grey in Table 1) which must be excluded due to ambiguity at the morphological level. If we compare the three plural forms below the cells shaded in grey in Table 1, we can see that the subjunctive forms are in fact distinct from the indicative ones. Therefore, in the present study, only 24 per cent of the data had to be excluded due to the ambiguity of forms, a much better rate of retention than reported for Laurentian French.

3.2 The subjunctive morphology in Baie Sainte-Marie Acadian French

In the spoken variety of Baie Sainte-Marie Acadian French, the subjunctive mood can be expressed in three different tenses as shown in Table 1: the present subjunctive (7), the past subjunctive (8) and the imperfect subjunctive (9).

1. (7)

   

2. (8)

   

3. (9)

   

It must first be noted that Laurentian French retains only the present subjunctive forms (Poplack, Reference Poplack, Laeufer and Morgan1992), with infrequent use of the past subjunctive and no tokens of the imperfect subjunctive. In fact, the imperfect subjunctive is largely lost from spoken varieties of French aside from a few notable exceptions (e.g., Guernesiais French, as reported by Jones, Reference Jones2000, and in some varieties of Acadian French, as reported by King, LeBlanc and Grimm, Reference King, LeBlanc and Grimm2018). Table 1 shows the conjugations for these three tenses in the Baie Sainte-Marie variety. It also reveals that the present and imperfect subjunctive forms show syncretism with other tense forms found in this variety of French. However, as is the case in Laurentian French, the present subjunctive forms for singular –er verbs are syncretized with the present indicative forms (as shown in 10) and are excluded from further analysis.

1. (10)

   

However, unlike in Laurentian French and less conservative varieties of Acadian French, the use of je…–ons and ils…–ont entails that all plural forms can be retained for analysis, as shown in (11).

1. (11)

   

In (11), it is precisely the addition of [j] (orthographically represented with a <i>) which indicates that this token is in the subjunctive mood and not the indicative (in which case, it would be allons).

Another potential case of ambiguity between forms involves another conservative linguistic feature, the simple past tense. As Table 1 shows, all forms of the simple past tense show syncretism with the imperfect subjunctive forms.Footnote ¹⁵ Despite this syncretism, I argue that it is possible to disambiguate between the two. In a Reference Comeau, King and Butler2012 article, Comeau, King and Butler demonstrate that the simple past tense is highly restricted to narrative contexts in this variety, occurring rarely in conversation (i.e., non-narrative discourse). Furthermore, it has been established that the subjunctive mood is restricted to embedded clauses governed by a particular set of main clause elements (verbal and non-verbal). Therefore, it is only at the intersection of these two contexts (i.e., subjunctive-selecting tokens in a narrative) that there is potential ambiguity between the two. As such, these tokens (i.e., those which occur following a main clause subjunctive-selecting element and within a narrative discourse) were excluded from further analysis.

4.1 The variable context for the French subjunctive

Since Poplack’s (Reference Poplack, Laeufer and Morgan1992) article on mood choice in Ottawa-Hull French, the usual approach in subsequent variationist studies of the French subjunctive has been to first extract all tokens of unambiguous subjunctive forms in order to then establish a list of subjunctive-selecting verbal and non-verbal main clause elements. Once the elements which have selected a token bearing subjunctive morphology at least once in the data have been established, then all tokens which occur within a finite embedded clause following these elements are extracted. This formal definition of the variable context provides an account as to where the subjunctive mood can occur based on community norms rather than arbitrary prescriptive norms, as detailed in Section 2.1. While I maintain that this approach has a clear advantage over the prescribed norms approach to the variable, I will argue that such an approach is inadequate to answer the question as to whether the subjunctive can function to express modality, which will be discussed in greater detail in Section 5.

4.1.1 The subjunctive mood governed by main clause verbs in Baie Sainte-Marie Acadian French

As per the method proposed by Poplack (Reference Poplack, Laeufer and Morgan1992) and adopted in most subsequent variationist studies of the French subjunctive (Poplack, Lealess and Dion, Reference Poplack, Lealess and Dion2013; Grimm, Reference Grimm2015), I first extracted all tokens of subjunctive forms in the Baie Sainte-Marie Acadian French data to determine which main clause elements (verbal and non-verbal) can occur with the morphological subjunctive mood. This allowed me to establish the list of main clause subjunctive-selecting verbal elements, shown in Table 2.

Table 2. Main clause verbs which can select the subjunctive in Baie Sainte-Marie Acadian French

As Table 2 shows, there are six main clause verbs which occur with the subjunctive in the Baie Sainte-Marie Acadian French data. Once these main clause verbal selectors were identified, it then became possible to extract all finite subordinate clauses which contained either a subjunctive, indicative or conditional verb. Following this, it became possible to calculate the rate of subjunctive usage per main clause verb, shown in Table 3.

Table 3. Distribution of subjunctive mood per verbal subjunctive-selecting elements

As Table 3 shows, with main clause verbal elements, the overall rate of subjunctive usage is quite high at 94 per cent. However, with the exception of one main clause verb, the negated form of the verb of opinion croire ‘to believe’, all main clause verbs occur with the subjunctive categorically, at a rate of 100 per cent.

The near lack of variability is somewhat surprising when we compare with studies of other varieties of French which have adopted the same variable context. For Laurentian French, all studies report variable subjunctive usage, from 76 per cent in Ottawa-Hull French (Poplack, Reference Poplack, Laeufer and Morgan1992) to 78 per cent in Ontarian French (Grimm, Reference Grimm2015).Footnote ¹⁸ In addition to Laurentian French, the Acadian French spoken in northeastern New Brunswick shows variable subjunctive usage (Roussel, Reference Roussel2016) with very low rates of the subjunctive mood (16%).Footnote ¹⁹ In their overview of Acadian communities in other provinces, King, LeBlanc, and Grimm (Reference King, LeBlanc and Grimm2018) report rates ranging from 73 per cent to 90 per cent, with the exception of the low 32 per cent for L’Anse-à-Canards. Taken together, results from Laurentian varieties as well as some varieties of Acadian French suggest that the subjunctive mood is indeed subject to inherent variability. However, the Baie Sainte-Marie results stand in stark opposition to these patterns: the subjunctive mood isn’t subject to inherent variability in most contexts in this variety of French.

With the exception of negated croire (discussed in greater detail in Section 4.1.1.1), the remarkable finding here is that the subjunctive is categorical in these contexts. This result has repercussions for broader claims that the subjunctive is subject to inherent variability in French (Poplack, Lealess and Dion, Reference Poplack, Lealess and Dion2013) or to those espousing claims that the subjunctive mood is being lost in certain varieties of French (Laurier, Reference Laurier, Mougeon and Beniak1989; Neumann-Holzschuh, Reference Neumann-Holzschuh, Brasseur and Falkert2005). At least for the Baie Sainte-Marie variety, there is near absence of variation, an established precursor to linguistic change.

4.1.1.1 Accounting for other moods: The expression of doubt

In Baie Sainte-Marie Acadian French, the only context following a verbal main clause element which shows variability is with the negated main clause verb croire ‘to believe’ as shown in (12).

1. (12)

   

Unlike other subjunctive-selecting main clause verbs (e.g., falloir ‘to be necessary’, aimer ‘to like’, vouloir ‘to want’), negated croire is used to express doubt, which is in fact one of the original functions of the subjunctive argued to be operational in Old French (Martineau, Reference Martineau1994). While other main clause verbs show a remarkable lack of embedded clause mood variation, a closer analysis of the tokens with negated croire suggests that the subjunctive mood is in fact not simply lexically constrained in this variety of French, but serves to express modality.

A first piece of evidence to understand why subjunctive use with negated croire cannot be relegated to a lexical construction is the fact that it is specifically the lexical item croire which functions as a verb of opinion in this variety rather than the lexical item penser. In contrast, studies of other varieties of French (e.g., Poplack, Reference Poplack, Laeufer and Morgan1992; Grimm, Reference Grimm2015) report that the subjunctive occurs with the negated verb of opinion penser. Within the Baie Sainte-Marie data, there are certainly tokens of penser ‘to think’ which literally means ‘to think’, as shown in (13), but they do not function to express an opinion or belief.

1. (13)

   

While there are tokens of penser to literally mean ‘to think’ (even negative tokens of penser), there are no tokens of it as a verb of opinion in this variety, which is expressed by the verb croire ‘to believe’, as shown in (14).

1. (14)

   

This is a notable difference between the Baie Sainte-Marie variety and all of the Laurentian varieties studied to date.Footnote ²¹ If the subjunctive morphology were to be simply lexically conditioned with no regard to semantics, we could not account as to why the semantic equivalent of penser in the Baie Sainte-Marie variety, croire, is the subjunctive selector in this variety, but penser is not.Footnote ²² Furthermore, it is not simply the presence of the lexical verb croire that can occur with the subjunctive; it must occur in a negated context. There are tokens of affirmative croire in the corpus (as in 14), but none of these occur with the subjunctive (and were not retained as part of the variable context). If the selection of the subjunctive was due to mere lexical conditioning, we would not be able to account as to why negation is a necessary requirement for the subjunctive mood to surface in the embedded clause for this main clause verb, but not the other verbs.Footnote ²³ The fact that it is only in negative contexts in which croire can express doubt further suggests that it is not the lexical item itself which selects the subjunctive, but that it is triggered by the dubitative semantic reading resulting from the negated verb of opinion. In this way, the subjunctive morphology can co-occur with negated croire in constituting a formal expression of modality, namely the subjunctive mood.

Another piece of evidence which supports an analysis whereby the subjunctive in the context of negated croire expresses doubt can be seen once we consider the tense concordance effect between the main clause and the embedded clause, as shown in Table 4.

Table 4. Tense concordance for negated croire (shown in number of tokens)

As the grey area in Table 4 shows, the subjunctive mood only occurs in contexts where the main clause verb is in the present indicative tense. When the main clause verb is in the past (either imperfect indicative, present perfect or past conditional), other forms are used in the embedded clause, such as the imperfect indicative, the present perfect and the past conditional. This suggests that in order to express doubt using the subjunctive, the expression of doubt must pertain at the moment of speech (e.g., je crois point que… ‘I don’t think that…’ vs. je croyais point que… ‘I didn’t think that…’). To account for this tense concordance effect, I argue that the expression of doubt must be relevant at the moment of speech since in the case where negated croire is conjugated in the past (the final three columns in Table 4), the expression of doubt no longer pertains to the speaker at the moment of speech and, crucially, we are left with an assertion (as opposed to doubt) regarding the embedded clause event. This results in other moods being selected in the embedded clause, but not the subjunctive, and so does not constitute a potential context of variation. However, as Table 4 shows, there is still variation even when the negated croire is in the present tense (the first column). Within the context of present tense negated croire (arguably, the true variable context for negated croire), the subjunctive mood occurs at a rate of 47 per cent (N = 8/17). In addition to the eight subjunctive tokens, we are left with nine tokens of negated present tense croire, which occur with a mood other than the subjunctive. How might we account for these nine tokens? The formal theoretical literature on the subjunctive mood can shed light on this apparent case of variation.

While it may be tempting simply to attribute these nine tokens to inherent variability, a closer inspection of these tokens suggests otherwise. Two of these nine tokens are used in interrogatives, as shown in (15) where the embedded clause verb is in the present conditional and (16) where the embedded clause verb is in the periphrastic future.

1. (15)

   

2. (16)

   

While there are different theoretical analyses of mood choice, some researchers have posited that the difference between indicative and subjunctive mood is related to the speaker’s assertion of the realization of the embedded clause event. For instance, Abouda (Reference Abouda2002) argues that such a semantic feature, assertion, can account for differences in mood choice. The fact that other moods occur in these two tokens is in line with formal theoretical accounts of the French subjunctive which argue that interrogation and negation cancel each other out since they both express non-assertion (Abouda, Reference Abouda2002). Evidently, it’s difficult to draw clear conclusions from two tokens, but nevertheless the pattern is consistent with the formal literature on the French subjunctive and the predictions are borne out in the data. Likewise, the so-called ‘polarity subjunctives’ in the formal literature, such as negated verbs of opinion, are described as a context which can vary based on the intentions of the speaker.Footnote ²⁴ Evidently, it is all too easy to attribute differences in mood morphology to differences in intention on the part of the speaker. From a variationist perspective, we simply cannot measure and correlate intentionality on the part of the speaker based on the presence of the subjunctive morphology. The argument is clearly circular. We leave this open question to future research.

4.1.2 The subjunctive mood governed by non-verbal main clause elements in Baie Sainte-Marie Acadian French

Beyond the set of subjunctive-selecting verbs, a small number of non-verbal main clause elements are also found to occur with subjunctive morphology in the Baie Sainte-Marie data set. These are outlined in Table 5.

Table 5. Distribution of subjunctive mood per non-verbal subjunctive-selecting elements

Upon initial inspection, Table 5 shows there to be a wider range of variability with non-verbal main clause elements than was the case for main clause verbs. However, a closer inspection of the data reveals that the overall rate of 66 per cent masks the subjunctive rates per individual non-verbal main clause elements (some are variable, others are not). There is a clear difference between the last three items in Table 5 (i.e., mais que, pour que and the infrequent subjunctive-selectors) in that they are near categorical.Footnote ²⁵ It is with the first three non-verbal elements (après que, avant que and jusqu’à tant que) where we do find variation in the data. However, as was the case for negated tokens of croire, the data in these three contexts also appear to be motivated by semantic considerations. If we first take the data with après que, we see that there is only a 5 per cent rate of subjunctive use (this rate is made up of two tokens). A closer inspection of the data with après que reveals that all the data are situated in the past or involve generic or atemporal reference, as in (17). The only two tokens which are temporally situated following the moment of speech are the two subjunctive tokens, which occur together, as shown in (18).

1. (17)

   

2. (18)

   

Therefore, in the case of tokens with après que, the fact that all non-subjunctive tokens are situated prior to the moment of speech (or generic) and that the two subjunctive tokens are situated following the moment of speech lends support to a semantic interpretation of mood choice.

Variation in mood choice is also evident with the other main clause elements, avant que ‘before’ and jusqu’à tant que ‘until’. As was the case with après que, the data in these contexts also appear to be motivated by semantic considerations. All of the indicative tokens are within a context in which the event has already been realized at the moment of speech, as in (19), while subjunctive tokens are unrealized, as in (20).

1. (19)

   

2. (20)

   

In (19), as in the other indicative tokens, the event expressed by the embedded clause (i.e., the act of having gone to Boston) is already realized at the moment of speech. Conversely, in (20) we see that the event of having gone to the nearby town of Yarmouth has not yet been realized at the moment of speech and we find use of the subjunctive. Thus, mood choice, again, appears to fall out from semantic functions rather than simply inherent variability, much as we saw with negated croire.

5.1 Mood choice as a linguistic variable in French

One of the major findings of this article is the near lack of variation with regard to mood choice. Within the context of main clause verbs, we saw that variation is restricted to the negated verb of opinion croire. Within the set of non-verbal main clause elements, we do find some variation, but this variation falls out from semantic differences related to modality rather than pure inherent variability. These findings suggest that, unlike most spoken varieties of French, including other varieties of Acadian French, mood choice in this variety is simply not a linguistic variable. This begs the question: Why did it not become a linguistic variable as it did in most spoken varieties? While we may be tempted to ascribe the subjunctive’s status as a non-variable in Baie Sainte-Marie to prescriptive efforts, I would argue that it’s highly unlikely that the Baie Sainte-Marie speakers were influenced by the prescriptive sources which seek to account for inherent variability. The older speakers in the sample have limited levels of education: most have a grade 8 level of education or less. In fact, historical low levels of education are a feature of Acadian communities more generally.Footnote ²⁶ Thus, with limited education and limited exposure to the standard, it’s highly unlikely that Baie Sainte-Marie speakers are influenced by French grammarians attempting to account for variation in mood choice. Rather, the more likely scenario is that the subjunctive simply hasn’t become a linguistic variable in this variety.Footnote ²⁷

If we take as a starting point that the subjunctive in Baie Sainte-Marie never was variable, then we are left with the inverse question as to why it became a linguistic variable in other varieties (i.e., the actuation problem of Weinreich, Labov and Herzog, Reference Weinreich, Labov, Herzog, Lehmann and Malkiel1968). To answer this question, I argue that a clue may be found in some findings of other varieties of French where the subjunctive is a linguistic variable, such as in Laurentian and in other Acadian varieties of French. In the case of Laurentian French, Poplack, Lealess and Dion (Reference Poplack, Lealess and Dion2013) report a strong tense concordance effect in nineteenth-century Quebec French in that both imperfect and especially conditional main clause verbs highly disfavour the subjunctive mood. In fact, it was the strongest constraint operating on mood choice in that data. However, in their twentieth-century data (the Ottawa-Hull corpus), they report a similar effect, although it had weakened by this time. In their most recent data (twenty-first century), the constraint was no longer statistically significant. In his analysis of four communities in Ontario French, Grimm (Reference Grimm2015: 146) reports that the Hawkesbury 2005 corpus (for which he has more data) does not show a tense concordance effect, although he advances that ‘[i]t is likely, however, that it did once exist, as the phenomenon is reported in other Laurentian varieties’.

If we consider other varieties of Acadian French, the article by King, LeBlanc, and Grimm (Reference King, LeBlanc and Grimm2018) lends support to the tense concordance effect being operational in Acadian French as well. While their analysis focuses only on mood choice with falloir (the most frequent main clause verb), they report a striking pattern across varieties: rates of the subjunctive appear to be much lower when falloir is in the conditional. In Saint-Louis, Prince Edward Island, while the overall rate of subjunctive use is fairly high (85%, total N=170), rates plummet to 18 per cent if falloir is in the conditional. The pattern repeats in the other varieties. In Abram-Village, also in Prince Edward Island, rates of the subjunctive (overall 73%, total N=327) drop to 13 per cent. In Ⓘles-de-la-Madeleine, Quebec, rates of the subjunctive (overall 90%, total N=193) drop to 21 per cent.

By way of comparison, I show the effect of tense concordance in the Baie Sainte-Marie data in Table 6 for the verb falloir to ensure comparability with the results from King, LeBlanc and Grimm (Reference King, LeBlanc and Grimm2018).

Table 6. Tense concordance with falloir in Baie Sainte-Marie Acadian French

Recall, that with the exception of negated croire, mood choice doesn’t vary with main clause verbs in Baie Sainte-Marie Acadian French. As Table 6 shows, the rate of subjunctive is 100 per cent across different main clause tenses, even with the conditional. This is a striking difference between the Baie Sainte-Marie variety and the other Acadian French varieties reported in the King, LeBlanc and Grimm study. The varieties they study show dramatically low subjunctive rates if falloir is in the conditional. Likewise, they report absence of the imperfect subjunctive with the exception of Saint-Louis, Prince Edward Island. Baie Sainte-Marie also retains the imperfect subjunctive and it is in fact the most frequent subjunctive tense used when falloir is in the conditional as shown in Table 6. However, in Saint-Louis, it is the conditional which is mainly used (14/17) in the embedded clause when falloir is in the conditional.Footnote ²⁸ Thus, it may be that, collectively, the loss of the imperfect subjunctive and the strength of the tense concordance effect (especially for having the conditional in both main and embedded clauses) play a role in triggering the subjunctive in becoming a linguistic variable. Therefore, it may be that the concomitant loss of imperfect subjunctive with the tense concordance effect represents a weak point in the grammar, a point which can allow for the subjunctive to then vary with other moods. In this line of reasoning, Saint-Louis may then represent an intermediary stage with regard to mood choice becoming a linguistic variable; an intermediary stage between, on the one hand, Baie Sainte-Marie, where mood choice is not inherently variable (i.e., not a linguistic variable), and the other varieties of French where it does display clear inherent variability.

5.2 The subjunctive mood’s semantic productivity

The other outstanding issue relating to the French subjunctive mood is its semantic role. Previous variationist research has debated whether the French subjunctive mood expresses semantic meaning or whether it is simply lexically conditioned and thus devoid of a semantic role. One of the key arguments of Poplack, Lealess and Dion’s (Reference Poplack, Lealess and Dion2013) article in support of this interpretation is that the subjunctive is constrained not by semantic classes of verbs, but rather by a small set of main clause elements (verbal and non-verbal). While the Baie Sainte-Marie results likewise show that the subjunctive is confined to embedded clauses following particular main clause elements, there are a number of pieces of evidence which suggests that the function of the subjunctive extends beyond simply belonging to fixed lexical constructions. These pieces of evidence will be explored in turn.

One of the key issues in assessing whether the subjunctive mood expresses meaning results from how the variable context is circumscribed. While Poplack’s (Reference Poplack, Laeufer and Morgan1992) proposal to analyse mood choice following the set of main clause elements which can select the subjunctive based on community norms was a notable advance, I argue here that it does not allow us to consider the range of meanings expressed by the subjunctive. Namely, the finding that the subjunctive is favoured under a set of main clause elements is certainly revealing in terms of a constraint operating on mood choice, but this does not capture the possibility that the subjunctive could be used to express a range of meanings. By limiting the analysis to those contexts which can select the subjunctive (the main clause verbal and non-verbal elements), we do not clearly see whether these elements belong to larger semantic classes in the language or whether they are the sole members of such classes. To accurately assess the semantic aspects of these main clause elements, one would have to consider all main clause elements which select a finite embedded clause within a corpus. By considering whether there are other main clause verbs which fall into the same semantic classes (volition, doubt, etc.), then we might be able to see whether it is the larger semantic class of verbs which can express a similar meaning that selects the subjunctive or only a particular subset of this class (namely, those verbs which have been found to select the subjunctive).Footnote ²⁹ Such an undertaking is beyond the scope of the present article and remains an open question in terms of the semantic role of the subjunctive. Despite this, there are other pieces of evidence which would argue for its semantic productivity.

Within the context of main clause verbal elements, the facts surrounding the negated verb of opinion croire suggests that it is highly unlikely that the subjunctive mood is simply lexicalized, resulting from a pairing of the subjunctive morphology with a main clause element, in this case, croire. It is specifically the fact that croire functions as a verb of opinion in the Baie Sainte-Marie variety and not penser, as in most other varieties of French, which suggests that it’s not simply the lexical item itself which governs mood choice, but the fact that it is a verb of opinion. Furthermore, the fact that the subjunctive is only possible when the verb is negated (functioning then to express doubt) lends support to the fact that the subjunctive works, along with the main clause element, to express a specific semantic meaning (in this case, doubt). Conversely, if croire was simply lexically conditioning the subjunctive, this would not account as to why negation is required for this main verb, but not others (e.g., falloir, vouloir). Further to this, the fact that negation is expressed not with one negative marker, but two in this variety of French (i.e., pas and point), further argues against a purely lexical-based analysis.

As another example of its semantic productivity in this variety, we do find two subjunctive tokens under an English-origin main clause element, next time, as shown in (21).

1. (21)

   

The fact that the subjunctive is possible under an English-origin governor such as next time suggests that it is not simply lexically conditioned by a similar set of main clause elements as found in other varieties, but that it can be productively used to express certain modalities, even if the governor is not historically part of the French language.

One other aspect of the lexical argument relies on the fact that there are few embedded verbs which can show explicit subjunctive morphology. These are reported to largely be être ‘to be’, avoir ‘to have’, aller ‘to go’, and faire ‘to make, to do’. However, this is not surprising considering the fact that studies of Laurentian French typically have to exclude approximately half of all tokens due to ambiguous forms (e.g., within the class of –er verbs, only second-person plural tokens can be retained for analysis). Therefore, I argue that the association between embedded verbs and subjunctive use should be characterized as epiphenomenal rather than as support of a lexical interpretation: by being restricted to analyse those forms which are unambiguous, it is clearly the suppletive forms (which are a large part of the unambiguous data which could be retained for analysis) which are found to favour subjunctive mood. This problem is not as serious in varieties with a richer verbal morphology, such as in Baie Sainte-Marie Acadian French, where only 24 per cent of the data had to be rejected due to ambiguous forms.

Finally, it may make better sense to adopt a functional definition of the variable context (such as a concept-oriented one, as per Edmonds, Gudmestad and Donaldson, Reference Edmonds, Gudmestad and Donaldson2017), rather than a formal one. That is, we may better assess its productivity in the language by considering its multiple uses much in the same vein as Grimm’s (Reference Grimm2015) analysis of variation between falloir and devoir in order to express necessity in Ontario French. By taking such an approach to the variable context, we may then better understand how the subjunctive mood is productive across the language in expressing modality. Poplack and her colleagues (Reference Poplack, Lealess and Dion2013) argue that a defining feature of spoken language, form-function asymmetry, is at odds with the prescriptive enterprise which seeks to impose form-function symmetry. With regard to the subjunctive mood, I argue that the subjunctive morphology does serve to express different meanings (e.g., doubt, necessity and volition). By analysing one of these at a time (e.g., how necessity is expressed, how doubt is expressed), we would be better placed to understand the role of the subjunctive in relation to other variants in each of those systems.