2. LINGUISTIC BACKGROUND

Dislocation is generally defined as a syntactic phenomenon whereby an element is detached from the main clause but is simultaneously represented within the main clause by means of a clitic pronoun (Ayres-Bennett et al., Reference Ayres-Bennett, Carruthers and Temple2001: 257), as illustrated in (4), where the dislocated NP and the co-referential clitic pronoun are bolded:

1. (4) L’auteur de l’Adonis, il ne peut être qu’un esprit singulièrement attentif (Valéry) (Blasco-Dulbecco, Reference Blasco-Dulbecco1999: 9; cited in McLaughlin, Reference McLaughlin2011: 2)

Research on this phenomenon is scattered throughout the literature, appearing under such discussions as word order (Ayres-Bennett et al., Reference Ayres-Bennett, Carruthers and Temple2001), emphasis (Riegel et al., Reference Riegel, Pellat and Rioul1994), and the left clause periphery (Rowlett, Reference Rowlett2007). In addition to being discussed under a range of topics, dislocation research has witnessed what Gadet termed a ‘foisonnement terminologique’ (Reference Gadet1991: 119); a variety of names (e.g., dislocation, detachment, segmentation) have been used to describe ‘dislocation’, and the semantic overtones of this terminology has been a source of contestation. Blanche-Benveniste (Reference Blanche-Benveniste and Brown2006: 483) asserts that ‘dislocation’ is a poor label that ‘hints about movements and pathological results’. Some scholars have also criticized the use of directional adjectives to specify the position of the dislocated NP relative to the main clause (i.e., ‘left’ and ‘right’ dislocation) due to their association with the written medium (ibid.; Blasco-Dulbecco, Reference Blasco-Dulbecco1999: 11). In order to avoid greater terminological confusion (as attested in discussions of borrowing, replication, code copying, transference, etc. in language contact research), I retain the term ‘dislocation’ due to its recognition in the field (Blasco-Dulbecco, Reference Blasco-Dulbecco1999: 43; McLaughlin, Reference McLaughlin2011: 210), and I refer to instances of ‘left’ and ‘right’ dislocation for the same reason.

The classification of examples necessitates a consideration of the distinction between dislocations featuring double marking as opposed to hanging topics. In the most recent book monograph on French dislocation—and 40 years after Cinque (Reference Cinque1977: 43) asserted that ‘in many cases, it is not simple to decide what construction we are observing’—De Cat (Reference De Cat2007: 108) notes that ‘[t]here is however considerable disagreement in the literature as to the precise characteristics of each construction’. She asserts that the only dependable method to identify hanging topic left dislocations (HTLD) has been the absence of dependency markers (also see Cinque, Reference Cinque1977; Larsson, Reference Larsson1979; Delais-Roussarie, Doetjes and Sleeman, Reference Delais-Roussarie, Doetjes, Sleeman, Corblin and de Swart2004): ‘The clearest case of HTLD would therefore be one in which the dislocated element could be introduced by a preposition’ (De Cat, Reference De Cat2007: 136) but is not. Applying this criterion, (a) would constitute a case of HTLD while (b) would represent left dislocation with double marking.

1. (a) Son photographe, elle lui en a jamais parlé.

2. (b) À son photographe, elle lui en a jamais parlé.

3. (examples reproduced from De Cat, p. 136)

This diagnostic was applied to the data for the present study in order to identify apparent examples of HTLD and exclude them from the analysis.

McLaughlin (Reference McLaughlin2011) highlights another definitional problem that linguists studying dislocation have faced: the relationship between the dislocated element and the clitic pronoun with which it co-occurs. The type of dislocation under study features clitic doubling, or double marquage, whereby two grammatical categories occupy a single syntactic position (Blasco-Dulbecco and Caddeo, Reference Blasco-Dulbecco and Caddeo2001): the clitic pronoun and NP exist in a relationship of co-reference (Blanche-Benveniste et al., Reference Blanche-Benveniste, Deulofeu, Stéphanini and van den Eynde1987: 83; Blasco-Dulbecco, Reference Blasco-Dulbecco1999). In order to underscore this relationship, the pronoun is referred to as ‘co-referent clitic ça’. In cases of double marking, Blasco and Cappeau (Reference Blasco and Cappeau1993) assert that the clitic pronoun is first linked to the verbal construction, noting that a secondary, co-referential relationship exists between the clitic and the dislocated lexical item. Blasco-Dulbecco (Reference Blasco1997: 2; Reference Blasco-Dulbecco1999: 99) contends that strict co-referential relationships in terms of gender and number between a dislocated element and a clitic pronoun are not always respected, as evidenced by examples such as (5), in which agreement is not realized:

1. (5) Les gosses, ça se lève tôt le matin (Queneau, Zazie dans le metro: 26, cited in Carlier, Reference Carlier1996: 134)

De Cat (Reference De Cat2005; Reference De Cat2007) argues that French subject clitics do not function as agreement markers, concluding that the minimal similarities in behaviour between subject clitics and affixes or morphemes should be viewed as an ‘accident’. Examples such as (5) do, however, prompt consideration of the particular type of reading that they impose.

Jones (Reference Jones1996) notes that ‘ça (and more rarely cela) can be used, even though gender is evident, when a generic interpretation is intended (i.e., when the statement applies to an entire class of entities)’ (p. 260). The examples that he provides illustrate the generic effect produced by ça, namely:

1. (6) Le café, ça empêche de dormir [generic reading, ça]

2. (7) Le café que j’ai bu hier soir, il m’a empêché de dormir [specific reading, il]

Jones comments that when this construction is extended to human beings, as in ‘les étudiants, ça ne travaille pas’ (his example), it frequently communicates a ‘pejorative attitude’ towards the dislocated NP and may be regarded as ‘patronizing or sarcastic even when the comment is complimentary’ (p. 260). While such statements may be understood as advancing a criticism about a semantic category (e.g., les étudiants in Jones’s example), represented by ça in the main clause, Blasco-Dulbecco (Reference Blasco-Dulbecco1999: 122–123) highlights the possibility for ambiguous readings in terms of genericity when a specific group constitutes the dislocated NP, as in example (8).

1. (8) mais je crois que ces envaisseurs ça a jamais existé (Jeanjean, Reference Jeanjean1985: 242–243).

In this example, the NP is marked by a demonstrative determiner, ces. The specificity expressed by the demonstrative coupled with the construction’s frequent association with generic readings call into question the pragmatic function (Ashby, Reference Ashby1988; Barnes, Reference Barnes1985) of this dislocation as an expression of genericity.

In addition to asserting that dislocation provides a means to provoke a generic reading, De Cat (Reference De Cat2007: 83) notes that dislocated NPs that express the subject of a generic sentence are ‘obligatorily dislocated in spoken French’. De Cat relates the avoidance of dislocation to the written medium, to which she attributes greater formality. In response to the examination of dislocation as an oral structure, as posed by Gadet (Reference Gadet1991: 110, i.e., ‘Le détachement: une structure de l’oral?’), McLaughlin (Reference McLaughlin2011) contends that orality has been oversimplified. Her research nuances the link between dislocation and orality by proposing three sub-types of orality: ‘primary’ orality as found in the oral code, ‘representative’ orality as manifested in journalistic prose and ‘constructed’ orality as imagined in fiction. Secondly, McLaughlin challenges the notion that the exclusive stylistic function of dislocation in French fiction is to express ‘orality’.

Blasco (Reference Blasco1997: 17) observes that types of dislocation are multiple and diverse in French: ‘il n’y a donc pas une dislocation en français mais plusieurs cas fort différents’. Although occasionally touched upon in discussions of non-prototypical cases of dislocation (Lambrecht, Reference Lambrecht1981; Blasco-Dulbecco, Reference Blasco-Dulbecco1999; Jones, Reference Jones1996), this article offers an initial corpus-based examination of a ‘rare syntactic form’ (Carruthers, Reference Carruthers1999): dislocation with co-referent clitic ça. In drawing on both an oral and a written corpus, the investigation builds on McLaughlin’s (Reference McLaughlin2011) refinement of ‘orality’ by illustrating the value of considering written data sources for the study of the dislocation under investigation as well as other syntactic constructions.

3. CORPORA AND METHODOLOGY

The French Oral Narrative Corpus (Carruthers, Reference Carruthers2013), a collection of audio recordings and accompanying transcriptions of 87 stories told to an audience by 18 storytellers (almost 1000 minutes of ‘spontaneous’ speech) was initially selected as the corpus for the present investigation. This preexisting corpus is conveniently tagged for instances of detachment where ‘the replacing pronoun is not totally straightforwardly co-referential with the dislocated element, as in les oignons ça fait pleurer’ (Carruthers, Reference Carruthers2013). In order to identify such cases of dislocation within the corpus, a search algorithm was createdFootnote ⁴ and an XML editor was used to search for these instances in each of the performed stories that comprise the corpus. Most of the results yielded by this search did not constitute examples of the syntactic construction under investigation and were eliminated accordingly (e.g., ces cris-là (ils) ne comprenaient pas ce que c’était). In total, the search produced only eight examples, which is unsurprising due to the focus of the present study on a rare syntactic form.

Carruthers (Reference Carruthers1999) suggests that researchers investigating rare syntactic constructions would benefit from a combination of two techniques, namely, a Labov-style interview, in which the researcher attempts to elicit the form being investigated, and oral and written questionnaires (e.g., asking whether participants use the construction in question, what sort of language it represents in their view, and with whom they associate it). These techniques were not feasible under the constraints of the present investigation. Consequently, it was deemed necessary to conduct a search in a broader corpus in order to generate a larger data sample for analysis, and FRANTEXT was selected.Footnote ⁵ Given the association between avoidance of dislocation and the written medium (De Cat, Reference De Cat2007), the selection of FRANTEXT could at first be regarded as an unconventional choice since this written corpus would presumably contain fewer instances of dislocation than an oral corpus. Nevertheless, with the premise that orality has been oversimplified, this investigation seeks to further contribute to its refinement through an exploration of the third sub-category put forth by McLaughlin (Reference McLaughlin2011): ‘constructed’ orality imagined in fictional written texts. The presence of written attestations of the dislocation under investigation in the FRANTEXT corpus highlights the importance of broadening our conception of orality in our research practices, as McLaughlin suggests.

FRANTEXT was selected as an appropriate corpus due to its size and the wide range of texts that it contains. So as to limit the scope of the project and allow for greater depth in analysis, the search was restricted to the nineteenth and twentieth centuries.Footnote ⁶ A preliminary search for all instances of ça in the entire period under examination, namely, 1800–1999, yielded 77,474 occurrences (Figure 1). These attestations were stratified into four half-century periods (i.e., 1800–1849, 1850–1899, 1900–1949, 1950–1999) according to search parameters available in the FRANTEXT database. The first fifth of the results for ‘ça’ were reviewed for each 50-year period. However, this approach did not yield a significant number of NP (,) ça [verb] or ça [verb] (,) NP dislocation examples. Consequently, it was deemed appropriate to instead search for ‘, ça’: (NP) ‘, ça’ [verb]. Although this could cause certain examples without a comma (NP ø ça [verb]) as well as instances of right dislocation (ça [verb] (,) NP) to be excluded from the present study, this represented the only feasible way to search and analyze thousands of instances of ça with the aim of encountering the dislocation variant under examination.Footnote ⁷ Even so, examples of right dislocation could still be encountered due to the way in which some dislocations were embedded in the surrounding narrative discourse, as in example (9).

1. (9) Voyez-vous, me dit Joseph, ça mange beaucoup, ces sortes d’animaux (Dumas, Le Capitaine Pamphile)

Figure 1. Delimitation of the FRANTEXT Data Set

The first fifth of the results for each 50-year period (15,495 occurrences in total for the four half-centuries) were examined for homogenous coverage of both centuries, and all instances of dislocation in which a dislocated NP co-occurred with ‘, ça’ were extracted for analysis (Table 1).

Table 1. Left Dislocations by Half-Century (FRANTEXT)

The data sample obtained from the FRANTEXT search initially yielded 149 examples of apparent dislocations marked by clitic ça. Two of these results were found to represent apparent cases of hanging topics according to the aforementioned diagnostic.Footnote ⁸ Unsurprisingly, almost all of the examples of dislocation (143 out of 147 examples, or 97 per cent) represent instances of left dislocation, which is likely a consequence of the adopted methodology. Despite the expected predominance of left dislocations, one instance of repeated dislocation (McLaughlin, Reference McLaughlin2011) was encountered:

1. (10) une femme qui est mariée… une femme qui est mariée… ça peut tout entendre, ça peut tout voir, une femme qui est mariée… (Goncourt, Sœur Philomène)

Additionally, three instances of right dislocation were obtained from the search, as in (11).Footnote ⁹

1. (11) Eh bien, ça parle, un avocat! (Sand, L’Homme de neige).

Since previous syntactic research has demonstrated the purely surface-level homogeneity of left and right dislocations (Lambrecht, Reference Lambrecht1981, Reference Lambrecht, Haspelmath, König, Oesterreicher and Raible2001; Fradin, Reference Fradin1988), the present study focused exclusively on the 143 examples of left dislocation.

4. RESULTS AND DISCUSSION

The examples from the oral Carruthers (Reference Carruthers2013) corpus and written FRANTEXT corpus are first presented separately. After analysing the data, I reinforce the value of employing oral and written corpora in syntactic research and encourage the consideration of written data in future investigations of rare syntactic forms.

4.1 Carruthers (Reference Carruthers2013)

Carruthers’ (Reference Carruthers2013) corpus yielded only eight examples, example (3) and examples (12)–(18), consisting of seven left dislocations and one right dislocation (example 18). Table 2 summarizes the grammatical number (i.e., quantifier, or indefinite, partitive, definite article) of the dislocated NPs, and Table 3 summarizes the use of demonstrative or possessive determiners in the eight examples. The latter two types of determiners were categorized separately since they are not associated with expressions of genericity.

1. (3) les petites vieilles ça peut courir très vite (Cevin, ‘Le Marchand’Footnote ¹⁰ )

2. (12) une femme en colère ça marche vite (Kiss, ‘Les Baies d’amour’)

3. (13) une vache ça coûte cher quand même (Kiss, ‘Le Paysan et la paysanne’)

4. (14) une femme aussi belle ça n’existe pas (Kiss, ‘Titete et Ticorps’)

5. (15) ah des fleurs roses des fleurs blanches ça embaumait! (Kiss, ‘Les Baies d’amour’)

6. (16) les oignons ça faisait autant pleurer (Kiss, ‘Le Paysan et la paysanne’)

7. (17) ses larmes ça donnait du courage aux gens qui en manquaient (Cevin, ‘Grand-Mère Mensonge - 1’)

8. (18) ça sent longtemps dans les casseroles l’odeur de brûlé (Kiss, ‘Le Paysan et la paysanne’)

Table 2. Grammatical Number of Dislocated NPs (Carruthers)

Table 3. Demonstrative and Possessive Adjectives Modifying Dislocated NPs (Carruthers)

Examples (15)–(17) deviate from expectations for generics in two respects. Generic statements are habitually expressed in the present tense since they communicate generalizations—truths from the perspective of the speaker—about a particular class of objects or animate beings. The examples in question involve incongruous use of the past tense with imperfect aspect. Further, the use of a possessive adjective (example 17) is not typically associated with generic readings due to its expression of specificity. These observations motivated attention to grammatical determiners as well as tense in investigating the expression of genericity in the larger FRANTEXT corpus. The gender of human referents is also noteworthy: all of the human NPs are women in these examples. This prompted further exploration of whether men also featured as NPs in the FRANTEXT data and, if so, whether a pejorative attitude was expressed towards them with similar frequency. That all examples from the Carruthers corpus were produced by only two storytellers also reinforced the necessity of consulting a larger corpus to eschew generalizations on the nature of dislocations with co-referent ça based on the speech of only two informants.

4.2 FRANTEXT

The analysis of the 143 left dislocations identified in the FRANTEXT data offered insights for discussions of genericity (Section 4.2.1), the differential expression of a pejorative attitude according to gender (Section 4.2.2) and the use of a particular register in the construction under study (Section 4.2.3).

4.2.1 Genericity

Generics are generally defined as statements that refer to an entire class or subclass of objects or individuals (Galmiche, Reference Galmiche1983: 30; Riegel et al., Reference Riegel, Pellat and Rioul1994: 284; Jones, Reference Jones1996: 260). Generic readings can be achieved with or without recourse to dislocation. For example, both example (19) and example (20) impose generic interpretations, with the dislocated element referring to a semantic category (in this case, les maris).

1. (19) Les maris durent toujours (my example)

2. (20) Les maris, ça dure toujours (Scribe, Le Mariage de raison)

Galmiche (Reference Galmiche1983: 30–31) defines ‘les articles génériques’ (i.e., articles used to introduce generics) as un(e), le/la, les, and des. He emphasizes that only phrases presenting ‘un syntagme générique’ can elicit a generic reading (p. 30). The presence of an ‘élément restrictif’ limits the scope of the dislocated NP and results in an ambiguous, ‘ungrammatical’ statement (p. 31), as in:

1. (21) *Les gosses de mon voisin, ça se lève tôt le matinFootnote ¹¹ (cited in Carlier, Reference Carlier1996: 134).

Non-generic determiners that constitute an ‘element réstrictif’ (i.e., demonstrative and possessive adjectives, e.g., ses gosses) violate properties of generic statements since they restrict the scope of the dislocated element that they modify and denote specificity (Blasco-Dulbecco, Reference Blasco-Dulbecco1999: 123). In order to test the association in the previous literature between dislocation featuring co-referent clitic ça and genericity, the data were analysed with regard to the types of determiners modifying dislocated NPs (Table 4).

Table 4. Determiners for Dislocated NPs (FRANTEXT)

The number of dislocated NPs introduced by an indefinite article (51 examples, or 36% of 143) is similar to the number introduced by a definite article (56 examples, or 39% of 143). Most striking is the high presence of demonstratives (24 examples, or 17% of 143), as illustrated by examples (22) and (23), due to their restrictive quality (Blasco-Dulbecco, Reference Blasco-Dulbecco1999: 123).

1. (22) Ces gosses, ça avale des gobelets d’eau froide, c’est détestable pour la santé (Colette, Claudine à l’école)

2. (23) ces gens-là,Footnote ¹² ça n’a ni famille ni patrie, ça n’a de commerce qu’avec le démon, et ça pille, et ça tue pour rien, pour le plaisir de faire le mal… (Zola, Vérité)Footnote ¹³

In such instances, the construction represents a commentary on a more restricted group of individuals—not a ‘class’ of individuals in a broad sense—and consequently prompts us to question the notion that ça always functions as a marker of genericity in this type of dislocation.

Although researchers agree that references to a particular object or individual (or group of objects or individuals) do not constitute generics (Galmiche, Reference Galmiche1983; Riegel et al., Reference Riegel, Pellat and Rioul1994; Carlier, Reference Carlier1996; Jones, Reference Jones1996; Blasco-Dulbecco, Reference Blasco-Dulbecco1999), example (24) convincingly demonstrates that ça may be used to refer to both a specific, restricted group and a particular individual.

1. (24) ces hommes… celui-là surtout, ça ne se doute de rien, ça ne pense à rien (Scribe, Le Mariage de raison)

The singling out of a particular individual (i.e., celui-là surtout) from an already restricted group (i.e., ces hommes) increases specificity. ‘Ces hommes’ could only support a generic interpretation if the NP were read as ‘ce genre d’hommes’. As previously mentioned, De Cat (Reference De Cat2007: 83) asserts that dislocated NPs expressing the subject of a generic sentence are ‘obligatorily dislocated in spoken French’. The presence of ça in this line, taken from Scribe’s play, could therefore be understood as a feature of the oral code that is represented graphically (i.e., ROG, Representation of the Oral in the Graphic, see McLaughlin, Reference McLaughlin2011: 7). The assertion made in (24) that the men being referred to, and that one man in particular, ne se doute de rien, ne pense à rien would seem to support the notion that detachment featuring co-referent ça may convey a pejorative attitude and may represent spoken French but need not express a consistent degree of genericity.

Carlier (Reference Carlier1996: 142) asserts that ça cannot refer to a particular individual, however an individual is singled out by name in example (25).

1. (25) le riche banquier Leuwen, avec Mademoiselle Des Brins, ça ne triche pas… (Stendhal, Lucien Leuwen)Footnote ¹⁴

A generic reading could only be produced if ‘le riche banquier’ were used to refer to a class of bankers (e.g., le riche banquier, ça ne triche pas), but this is not the case. For this reason, the use of detachment with ça can be interpreted as characterizing spoken French.

A different ‘generic particle’ introduces a member of a particular family in example (26):

1. (26) mais un Mauprat, vois-tu, ça sait lire et écrire, et ça n’en est que plus méchant. (Sand, Mauprat)

This example further attests to the extent to which the scope of a NP was found to vary within the data. Here, a particular family represents a ‘class’ of individuals, and generalizations are made about members’ characteristics and abilities. As with all generics, the judgement made about ‘un Mauprat’ is expressed as an unequivocal truth (Galmiche, Reference Galmiche1983: 29).

Galmiche (Reference Galmiche1983: 32) points out that, paradoxically, while generic statements express ‘truths’, the dislocated elements cannot be introduced by quantifiers expressing ‘universality’ (i.e., tout(e)(s), n’importe quel(le), chaque). In other words, although generic statements can be made about a Mauprat, they cannot be made about all Mauprats. He provides example (27) as an illustration of this point.

1. (27) * Tous les canaris, ça chante bien. (Galmiche’s example)

Nevertheless, one example expressing universality was identified in the written corpus, namely (28).

1. (28) tous les journaux, ça ne sert à rien (Murger, Scènes de la vie de Bohème)Footnote ¹⁵

Similarly, Galmiche notes that other quantifiers (i.e., *beaucoup de, quelques, plusieurs, certains, deux [nombre]) are equally unacceptable in generic statements, but example (29) from the FRANTEXT corpus contains the quantifier ‘cent’.

1. (29) Cent francs , même à la condition d’en rendre cent vingt, ça ne se lâche pas comme un chien dans une église… (Balzac, Les Petits Bourgeois)

Given the specificity of an amount of money, in this instance ça could hardly be interpreted as a marker of genericity. While this example could be understood as another instance of ROG (McLaughlin, Reference McLaughlin2011: 7), one could also posit that the dislocated element is resumed by ça to compensate for the interjection of a prepositional phrase.

In addition to considering restrictions in the scope of NPs imposed by particular determiners, it is important to examine limitations on generic readings resulting from tense selection. Given the association between generic statements and truths, we would expect to encounter verbs in the present tense in the main clause of examples. However, the attestation of examples in the past tense in the Carruthers corpus motivated an examination of the FRANTEXT examples with attention to tense (Figure 2).Footnote ¹⁶

Figure 2. Tense of Left Dislocations (FRANTEXT)

Unsurprisingly, most of the FRANTEXT examples contain a verb in the present indicative (132 or 92% of 143). Although the use of the simple past closely approximates the effect of examples in the present, namely, the communication of a constant state or ‘truth’ (example 30), the inclusion of the indicative pluperfect in example (31) expresses specificity on two levels: the employed tense and the presence of a demonstrative adjective illustrate that dislocations featuring co-referent clitic ça do not systematically impose generic interpretations.

1. (30) Les femmes, ça n’a jamais su prendre les hommes. (Zola, Travail)

2. (31) ces jeunesses, ça s’était parlé sans se rien dire. (Lamartine, Geneviève)

This observation is reinforced by the two examples in the future tense, which are also characterized by the presence of a demonstrative, as in:

1. (32) ces beaux cheveux blonds, ça tirera l’oeil. (Sand, Le Marquis de Villemer)

and a dislocated NP modified by a possessive adjective:

1. (33) mes vers, ça durera toujours! (Claudel, Poésies diverses)

Although four of the statements including a present conditional support a generic reading, there are two exceptions in which such an interpretation is less straightforward: although to a lesser degree than in cases involving a demonstrative, specificity is expressed through a comparison (i.e., a particular individual is positioned as representative of a kind of individual) in example (34). Specificity is also communicated by means of a possessive adjective in example (35):

1. (34) un gaillard membré comme vous , ça gagnerait de l’or dans not[re]’pays. (Moselly, Terres Lorraines)

2. (35) Vos boniments au mariage, ça pourrait bien faire du vinaigre. (Aymé, Clérambard)

4.2.2 Pejorative attitudes

As previously mentioned, the presence of exclusively-female animate NPs in the oral corpus prompted further investigation into differences in the use of the dislocation under study according to gender.

In the FRANTEXT data, animatesFootnote ¹⁷ are seen to outnumber inanimates (81 animate examples, or 57% of 143). Of the animate referents encountered, 73 are human (90% of 81),Footnote ¹⁸ and it would seem important to explore Jones’s (Reference Jones1996: 260) observation that ‘the generic use of ce and ça extends to human beings’ and ‘often conveys a pejorative attitude towards the persons concerned’. Examples expressing a pejorative attitude are classified as ‘critical references’ in this analysis.

In order to investigate the frequency with which different human referents are criticized, these NPs were divided into four sub-categories: children, men, women and ambiguous mixed-gender groups. The number of examples for each group is summarized in Figure 3.

Figure 3. Representation of Animate Sub-Categories (FRANTEXT)

Twenty-one examples involved NPs of ambiguous or mixed gender. Since the masculine grammatical gender is employed in French when referring to an exclusively male group or a mixed-gender group, potentially mixed-gender adult NPs were not subject to gender analysis. For instance, it is impossible to determine the presence of women in NPs such as les riches:

1. (36) les riches, ça a toujours du temps à perdre. (Bazin, Le Blé qui lève)

Each example was examined within its surrounding context in order to determine whether it represented a criticism (Table 5). The difference in the percentage of critical references towards female versus male referents was found to be statistically significant.Footnote ¹⁹

Table 5. Critical References to Animate NPs (FRANTEXT)

As Harris (Reference Harris and Fisiak1980: 154) observes, the distinction between proximity versus distance as conveyed through demonstrative determiners ‘is at times pressed into service with a secondary function, e.g. to add a pejorative overtone’. That four of the 12 criticisms of women involve a NP that is modified by a demonstrative determiner (example 37) again calls into question the notion that dislocations characterized by co-referent clitic ça always and only impose generic readings.

1. (37) Ces grandes dames, ça ne va pas souvent à l’église (Stendhal, Lucien Leuwen)

4.2.3 Register

While detachment featuring ça may appear to reflect orality, it is surprising that only six examples from the FRANTEXT data sample present salient features of non-standard French (i.e., via contraction or particular non-standard orthography). A contraction is found in examples (34) and (38).

1. (38) Les femmes, ça n[e]’doute de rien (Leclercq, Le Savetier et le financier)

These examples can be understood as manifestations of the intermediary code, ROG (McLaughlin, Reference McLaughlin2011: 7), and the absence of the schwa [Ó™] can be interpreted as a feature of spoken French. A marked register is also signaled through lexical choices (example 39) and eye dialect, the representation of colloquial speech through non-standard orthography (examples 40 through 42).

1. (39) Ces guiantres de noces, ça vous joue toujours ce tour-là (Leclercq, Le Savetier et le financier)Footnote ²⁰

2. (40) Non, mossiou.Footnote ²¹ L’Italien, ça rit toujours, mais ça ne se radoucit point pour ça! (Sand, La Daniella)

3. (41) Les miracles, ça se passe là-bas, loin, jamais cheux nous (Hamp, Marée fraîche; Vin de Champagne)

4. (42) Un petit mémento de tems en tems, ça amuse (Pixérécourt, Coelina, ou L’enfant du mystère)

Conclusions on register or features of orality in graphic attestations of the type of dislocation under study cannot be drawn based on the limited examples highlighted above.Footnote ²² Future syntactic research should explore and compare the register (as communicated on the lexical and syntactic level, e.g., the presence of ne-deletion) of the construction under investigation in both written and oral corpora.

5. CONCLUSIONS AND FUTURE DIRECTIONS

This study, which represents an initial corpus-driven exploration of a rare syntactic form, dislocation in which a dislocated element occurs with co-referent clitic ça, has contributed three main findings to French syntactic research. Firstly, in light of the varying scope of dislocated NPs encountered in the corpus data, the research questions the notion that dislocation with clitic ça always and only imposes a generic reading. Secondly, the written data reveals a significant difference in the number of women versus men featured in examples expressing criticisms. Although it is premature to draw any definitive conclusions, further research should continue to examine how this type of dislocation may be deployed more frequently when making pejorative comments about particular groups. Thirdly, this investigation nuances the association between dislocation and orality (McLaughlin, Reference McLaughlin2011) and consequently encourages methodological practices that consider both written and spoken texts as complementary sources of dislocation data.

Future corpus-based investigations focused on this construction should consider a broad range of spoken as well as written data, including but not limited to online forums or blogs that may generate discussions of perceived sociocultural differences,Footnote ²³ social media platforms, literary texts, and the press. Elicitation techniques can also be employed to collect examples of this construction. The elicited data could then be complemented by oral and written questionnaires aimed at assessing semantic overtones that speakers attribute to dislocation featuring co-referent ça versus other syntactic alternatives (see Carruthers, Reference Carruthers1999). Beeching (Reference Beeching1997) reminds us that ‘the relationship between form and function is not static’, and it is necessary to look to the surrounding discursive context—and, when possible, to speakers (or writers) themselves—in an attempt to identify the function of a particular construction embedded within a particular set of circumstances.

The data from this study compel us to interrogate the notion of dislocation as a ‘linguistic variant’. Coveney (Reference Coveney2003: 112) asserts that ‘les variantes doivent représenter “différentes façons de dire la même chose”’, but Carruthers (Reference Carruthers1999: 14) argues that ‘it is very difficult to argue that two syntactic constructions are ever truly semantically and pragmatically equivalent’, citing the case of dislocation. She holds that ‘it is not possible to argue that such constructions “mean the same thing”’ (p. 16). Surveys would help us to better understand the extent to which constructions ‘mean the same thing’ when ça substitutes a ‘traditional’ subject (e.g. il, elle) and would assist us in assessing the complex ways in which syntax contributes to nuanced meaning.