1Contrastive topics (underlined in the following examples) have typically been studied on the basis of question-answer pairs as in [1]-[2] (Büring, 2016; Wagner, 2012).
[1] |
A: Who ate what? B: Fred ate the beans and Mary ate the spinach. |
|
(adapted from Wagner, 2012: 3) |
[2] |
A: Who do they want to kick out? B: She wants to kick me out. |
|
(adapted from Büring, 2016: 65) |
2These examples illustrate that contrastive topics always introduce alternatives, which can be mentioned explicitly (e.g., Fred versus Mary in [1]) or remain implicit (e.g., she versus “others who want to do some kicking out” in [2]). In this paper, we focus on cases with explicit alternatives, like [1], where there is an overt topic shift from the first topic (i.e., Fred) to the second topic (i.e., Mary).
3Several strategies exist to mark such contrastive topics. Cross-linguistically, most attention has gone to prosodic marking (e.g., Veselá et al., 2004 on Czech; Hedberg & Sosa, 2007 on English; Lee, 2007 on Korean and English; Ambrazaitis & Frid, 2012 on Swedish; Yasavul, 2013 on K’iche’; Sahkai & Mihkla, 2017 on Estonian; Zerbian et al., 2016 and Riester et al., 2020 on German; Gürer, 2020 on Turkish). In French however, preverbal subjects cannot be stressed (Lambrecht, 1994; Klein, 2012), which means that it is impossible to prosodically mark contrastive topics as le moineau domestique “the house sparrow” and le rouge-gorge “the robin” in [3].
[3] |
Le moineau domestique adore les graines de tournesol. Le rouge-gorge préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds. The robin prefers mealworms.’ |
4Instead, several other linguistic means can be used to explicitly express contrast on the subject, such as an explicit topic marker (e.g., quant à “as for”, de son côté “on his side”) (see [4a]) (Choi-Jonin, 2003; Lagae, 2007; Velghe & Lahousse, 2015), an emphatic pronoun (see [4b]) or a contrastive adverb (e.g., par contre “on the other hand”, en revanche “on the other hand”) (see [4c]).
[4a] |
Le moineau domestique adore les graines de tournesol. Le rouge-gorge, quant à lui, préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds. The robin, as for him, prefers mealworms.’ |
[4b] |
Le moineau domestique adore les graines de tournesol. Le rouge‑gorge, lui, préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds. The robin, him, prefers mealworms.’ |
[4c] |
Le moineau domestique adore les graines de tournesol. Le rouge-gorge, par contre, préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds. The robin, on the other hand, prefers mealworms.’ |
5Importantly, the three marking strategies in [4] can only mark contrastive topics in topic shift contexts, where the alternatives are explicitly mentioned (example [1]), and not in cases with implicit alternatives (example [2]). Moreover, they can only accompany the second contrastive topic (i.e., le rouge-gorge “the robin” in [4]), and not the first one (i.e., le moineau domestique “the house sparrow”) (Anscombre, 2006; Cappeau, 1999; Choi-Jonin, 2003; Fløttum, 1999, 2003; Lagae, 2007).
6Although there is a rich literature on the information-structural and semantic properties of contrastive topics in general (e.g., Wagner, 2012; Büring, 2016; Lee, 2017), the interchangeability of different marking strategies has not been extensively studied on the basis of corpus analyses. Moreover, as far as we can tell, little is known about the way in which register influences the choice for a certain marking.
7In this paper, we focus on the interchangeability of two non-prosodic marking strategies of contrastive topics in a topic shift context, emphatic pronouns (E-Pros) (see [4b]) and contrastive adverbs (C-Advs) (see [4c]), in spoken French (interviews) and two types of written French: rather informal extracts from a discussion platform and more formal texts from two newspapers. The goal is to analyse the choice between C-Advs and E-Pros in cases where they occur between the subject (S) and the finite verb (fin-V), as in [4b-c] above, depending on the type of corpus (i.e., register). We will show that the frequency of both marking strategies is register-dependent: E-Pros occur more often between the S and the fin-V in the informal written and spoken corpora, whereas C-Advs are more frequent in this position in the formal newspaper corpora. In addition, we will demonstrate that the use of E-Pros is influenced by pragmatic constraints, which is not the case for C-Advs: E-Pros typically modify subjects with a clearly identifiable – preferably human – referent.
8The paper is structured as follows. We first address the interchangeability of C-Advs and E-Pros (§2), by discussing previous work on these markers (§2.1), and by illustrating in which syntactic contexts (§2.2) and in which syntactic positions (§2.3) they can actually compete with each other. This is followed by a methodological section (§3), in which we present the corpora (§3.1) and the way in which the data were extracted from these corpora (§3.2). Next, we present and discuss the results of our corpus analysis on the choice between C-Advs and E-Pros (§4), by focusing on their frequency (§4.1), the syntactic type of subject (§4.2) and the semantic type of subject (§4.3).
9Before tackling the choice between C-Advs and E-Pros, it is important to briefly discuss previous research on these markers (§2.1) and to clarify in which cases exactly these two markers are interchangeable, i.e. in which syntactic contexts (§2.2) and in which syntactic positions (§2.3) E-Pros can be replaced by C-Advs.
10There is quite a rich literature on French C-Advs, but most studies are based only on formal language data and are restricted to one or two specific C-Adv(s) (e.g., Anscombre, 1983 on pourtant “yet”, pour autant “however”; Danjou-Flaux, 1980 on au contraire “on the contrary”, par contre “on the other hand”, en revanche “on the other hand”; Danjou-Flaux, 1983, 1984 on au contraire “on the contrary”; Hamma & Haillet, 2002 on par contre “on the other hand”, en revanche “on the other hand”; Jayez, 1982 on pourtant “yet”, quand même “nevertheless”; Lenepveu, 2009 on toutefois “however”; Lenepveu, 2007 and Mellet & Monte, 2005 on toutefois “however”, néanmoins “however”; Masseron & Wiederspiel, 2003 on au contraire “on the contrary”, par contre “on the other hand”; Mellet & Ruggia, 2010 and Moeschler & Spengler, 1981 on quand même “nevertheless”; Rivara, 2008 on pour autant “however”; Veland, 1998 on quand même “nevertheless”, tout de même “nevertheless”). As far as we can tell, the only linguists who consider a set of C-Advs in different registers are Csűry (2001) and Dupont (2019, 2020). However, Csűry (2001) does not systematically distinguish between the different registers in his corpus (novels, newspaper data, formal and informal speech) when presenting his results, and Dupont (2019, 2020) only focuses on two registers that are very similar in terms of their level of formality (newspaper editorials and research articles). In addition, previous research on French C-Advs mainly deals with their semantic properties and with the different types of contrastive discourse relation (concessive, corrective, etc.) they can express. These semantic analyses are not relevant for our study on the choice between C-Advs and E-Pros, because E-Pros always mark a simple contrast or antithesis.
11The use of E-Pros in French has less extensively been studied. Cappeau (2004) and Bahtchevanova & Van Gelderen (2016) provide a (non-exhaustive) overview of the different contexts in which E-Pros can occur (see §2.2 for more details), but they do not deal with E-Pros as markers of contrastive topics in a topic shift context. Other analyses focus on specific uses of E-Pros, for example in cases with clitic doubling (moi, je trouve que “me, I think that”) (Detges & Waltereit, 2014) or in constructions with a locative (je lui arrivais à l’épaule “I him came to the shoulder” [literally]) (Blanche-Benveniste, 2001). As far as we know, the only researchers who discuss the use of E-Pros to mark contrastive topics are Rocquet (2014), who analyses the syntactic and discursive properties of E-Pros versus prepositional phrases with quant à “as for” + pronoun (see example [4a] above), and Caddéo (2004), who compares clauses with an E-Pro followed by a noun phrase (NP) (lui, le propriétaire, trouve que “him, the owner, thinks that”) to clauses with an NP followed by an E-Pro (le propriétaire, lui, trouve que “the owner, him, thinks that”) (see §2.2 for more details). Caddéo (2004) illustrates that E-Pros can in the latter case only modify NP-subjects (and not NP-objects) and that these subjects can have a human or inanimate referent – but we will show that they have a preference for human subjects (see §4.3).
12Importantly, C-Advs and E-Pros have never been compared, although they can compete with each other when marking contrastive topics in a topic shift context, as we will illustrate in the next sections.
- 1 In some rare cases, the C-Adv is used as a complement or predicate, as illustrated in the following (...)
13Whereas the function of C-Advs is limited to the expression of a contrastive discourse relation between two clauses or constituents1, French E-Pros have a much larger use. As in most Indo-European languages, the emphatic forms of the pronoun are used after prepositions (e.g., J’achète des graines pour eux “I buy seeds for them”), but in French, the third person E-Pros can themselves also function as subject (Cappeau, 2004; Bahtchevanova & Van Gelderen, 2016). As subject of the clause, the E-Pro does not mark a topic shift, but signals topic continuity with respect to the topic of the preceding clause, i.e. le rouge-gorge “the robin” in [5].
[5] |
Le moineau domestique adore les graines de tournesol, mais le rouge-gorge est un véritable insectivore. Lui préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds, but the robin is a real insectivore. Him prefers mealworms.’ |
14In addition, E-Pros can function as a dislocated subject, with a clitic (underlined in the following examples) as the resumptive pronoun (De Cat, 2007) (see [6a]). The dislocated E-Pro assures topic continuity (lui “him” = il “he” = le rouge-gorge “the robin”) while also highlighting the contrast between the current topic (i.e., le rouge-gorge “the robin”) and the previous topic (i.e., le moineau domestique “the house sparrow”). As shown in example [6b], replacement by a C-Adv would be pragmatically infelicitous, because this marker is not able to signal topic continuity and contrast at the same time. The C-Adv tries to establish a contrastive relation between the clause le rouge-gorge est un véritable insectivore “the robin is a real insectivore” and the clause il préfère les vers de farine “he prefers mealworms”, which is impossible, since there is no contrast at the level of the subject (instead of contrast, there is continuity between rouge-gorge “robin” and il “he”) nor within the predicate (instead of contrast, there is continuity between insectivore “insectivore” and vers de farine “mealworms”).
[6a] |
Le moineau domestique adore les graines de tournesol, mais le rouge-gorge est un véritable insectivore. Lui, il préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds, but the robin is a real insectivore. Him, he prefers mealworms.’ |
[6b] |
Le moineau domestique adore les graines de tournesol, mais le rouge-gorge est un véritable insectivore. #Par contre, il préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds, but the robin is a real insectivore. #On the other hand, he prefers mealworms.’ |
15C-Advs can only appear in clauses with a pronominal subject signalling topic continuity when there is a contrast within the predicate, which is then emphasized by the C-Adv. This is illustrated in example [7a], where par contre “on the other hand” highlights the contrast between vers de farine “mealworms” and graines de tournesol “sunflower seeds”. In this type of context, the use of a dislocated E-Pro would be grammatically possible, but pragmatically unacceptable, since the E-Pro requires contrast at the level of the topic, and in [7b], there is no other discourse referent (such as le moineau domestique “the house sparrow” in the previous examples) with which le rouge-gorge “the robin” could be contrasted.
[7a] |
Le rouge-gorge adore les vers de farine. Par contre, il n’aime pas les graines de tournesol. |
|
‘The robin loves mealworms. On the other hand, he does not like sunflower seeds.’ |
[7b] |
Le rouge-gorge adore les vers de farine. #Lui, il n’aime pas les graines de tournesol. |
|
‘The robin loves mealworms. #Him, he does not like sunflower seeds.’ |
16Finally, when E-Pros occur after a nominal subject (underlined in the following examples), they function as markers of contrastive topics in a topic shift context. For instance, in example [8a], the E-Pro highlights that the sentence topic changes from le moineau domestique “the house sparrow” to le rouge-gorge “the robin”. In this topic shift context, E-Pros are fully interchangeable with C-Advs, as shown in [8b]. Hence, in the remainder of this paper, we will focus on examples like [8a-b], where E-Pros and C-Advs compete with each other.
[8a] |
Le moineau domestique adore les graines de tournesol. Le rouge-gorge lui préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds. The robin him prefers mealworms.’ |
[8b] |
Le moineau domestique adore les graines de tournesol. Le rouge-gorge par contre préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds. The robin on the other hand prefers mealworms.’ |
17From a syntactic point of view, both C-Advs and E-Pros with a topic shift interpretation are mobile elements (e.g., Altenberg, 1998, 2006; Caddéo, 2004; Csűry, 2001; Dupont, 2015, 2019, 2020; Lenker, 2014). As shown in examples [9a-b], they can appear in different positions in the clause.
[9a] |
Le moineau domestique adore les graines de tournesol. [*Lui] le rouge-gorge [lui] préfère [lui] les vers de farine [lui]. |
|
‘The house sparrow loves sunflower seeds. [*Him] the robin [him] prefers [him] mealworms [him].’ |
[9b] |
Le moineau domestique adore les graines de tournesol. [Par contre] le rouge-gorge [par contre] préfère [par contre] les vers de farine [par contre]. |
|
‘The house sparrow loves sunflower seeds. [On the other hand] the robin [on the other hand] prefers [on the other hand] mealworms [on the other hand].’ |
18There is however one important exception: E-Pros with a topic shift interpretation cannot occur before a nominal subject, as indicated in example [9a]. It is possible for E-Pros to appear before an NP, but in those cases, the E-Pro itself functions as the subject (see example [5] above) and it does not give rise to a topic shift interpretation. For instance, in [10], the E-Pro lui “him” signals topic continuity with respect to the sentence topic of the preceding clause, i.e. Jean. It is followed by the NP le grand frère de Pierre “Pierre’s older brother” (underlined), which is not a contrastive topic, but functions as an apposition to the E-Pro (see Caddéo [2004] for an analysis of examples with an E-Pro followed by an apposition).
[10] |
Sabine a deux enfants: Pierre, qui a cinq ans, et Jean, qui a douze ans. Pierre joue au foot et il aime aussi beaucoup le tennis. Pendant les vacances d’été, il a participé pour la première fois à un camp sportif. Mais Jean n’aime pas du tout le sport. Lui, le grand frère de Pierre, préfère la musique et la peinture. |
|
‘Sabine has two children: Pierre, who is five years old, and Jean, who is twelve years old. Pierre plays football and he also likes tennis very much. During the summer holidays, he took part in a sports camp for the first time. But Jean does not like sports at all. Him, Pierre’s older brother, prefers music and painting.’ |
19This means that, except for clauses where a C-Adv precedes a nominal subject, the two markers are interchangeable, as illustrated in examples [9a-b] above. In our corpus study however, we focus on the interchangeability in one specific position, namely cases where the C-Adv or E-Pro occurs between the S and the fin-V, as in [11].
[11] |
Le moineau domestique adore les graines de tournesol. [Le rouge-gorge]S [par contre/lui] [préfère]fin-V les vers de farine. |
|
‘The house sparrow loves sunflower seeds. [The robin]S [on the other hand/him] [prefers]fin-V mealworms.’ |
20This decision is based on the hypothesis that this syntactic position is the most ideal one to mark contrastive subject topics, allowing to clearly “isolate” the entities that are being contrasted and to highlight the topical shift (Altenberg, 2006; Cappeau, 1999; Dupont, 2019; Lenker, 2014). Yet, with respect to French C-Advs and E-Pros, this position between S and fin-V has not yet been studied in detail. Moreover, by focusing on this one specific case, we avoid that syntactic position as a factor interferes with our findings regarding the syntactic and semantic properties of the subject (see Brysbaert & Lahousse, 2020a on the choice between C-Adv and E-Pro depending on the syntactic position).
21We study the choice between C-Advs and E-Pros using data from six corpora: Le Monde (LM) (Le Monde, 2000), Est Républicain (ER) (Gaiffe et al., 2018), Yahoo Contrastive Corpus of Questions and Answers (YCC) (De Smet, 2009), Corpus oral de français de Suisse romande (OFROM) (http://www11.unine.ch/) (Avanzi et al., 2012-2020), Corpus de français parlé parisien des années 2000 (CFPP) (http://cfpp2000.univ-paris3.fr/) (Branca-Rosoff et al., 2011, 2012) and Corpus de référence du français parlé (CRFP) (Équipe DELIC, 2004).
22Table 1 summarises the properties of the different corpora. Three of them consist of written French (LM, ER and YCC), while the others contain spoken French (OFROM, CFPP and CRFP). In the remainder of this paper, we will group together the results for the three spoken corpora, given their relatively small size.
Table 1 – Properties of the six corpora
- 2 The full ER corpus contains data from 1999-2003 and 2006-2011, but we only consulted the most recen (...)
|
LM |
ER |
YCC |
OFROM |
CFPP |
CRFP |
Mode |
written |
written |
written |
spoken |
spoken |
spoken |
Level of formality |
+ formal |
± formal |
- formal |
± formal |
± formal |
± formal |
Type |
national news-paper |
regional news-paper |
online discussion platform |
interview |
interview |
interview & public speech |
Year |
1998 |
2010-20112 |
2006-2009 |
2012-2020 |
2005-2019 |
1998-2004 |
Number of words |
25.7 million |
74 million |
6.1 million |
1.01 million |
0.7 million |
0.44 million |
23These corpora are representative of different levels of formality. Based on the continuum between “language of distance” (formal) to “language of immediacy” (informal) described by Koch and Oesterreicher (1985, 2007), the three written corpora can be ranged from most formal (LM) to least formal (YCC). The articles in LM clearly correspond to what is typically characterised as “language of distance”: they deal with public events and topics, relevant to a large audience, and are written in a neat and thoughtful style, with low to zero emotional involvement of the journalist. The data in the ER corpus are similar to those in LM as far as the text type is concerned. However, the content of this regional newspaper is mainly restricted to more local events and topics (e.g., sport competitions, weddings, births, politics on the level of the municipality, etc.), and is therefore of interest only to a smaller audience. As a consequence, journalists of ER also adopt a much more personal and involved writing style. The ER corpus should thus be situated somewhat further away from the “language of distance” pole. The third written corpus, YCC, comes closest to the “language of immediacy” pole. Users of this online discussion platform often write as they think, without editing their posts, which leads to a very direct and engaged writing style. The topics discussed on the platform (e.g., health issues, relational problems, etc.) are to some extent also more private than those in the two journalistic corpora, even though the users of the forum are anonymous and do not know each other.
24The three spoken corpora (OFROM, CFPP and CRFP) cannot be further distinguished based on the level of formality. They all represent more or less informal spoken French, consisting of conversations between an interviewer and one or more interviewees who are not familiar with each other – except for CRFP which also contains some public speech. However, due to the interview setting, the conversations are less spontaneous than if they took place with the same people without being recorded. Interviewees might feel inhibited to tell details, and they are supposedly not always comfortable with the topics addressed by the interviewer.
25As indicated in Table 1, there are also some differences between the corpora with respect to their year(s) of collection and their size, but we believe that these differences do not have an effect on the results of our study.
- 3 For example, C-Advs such as néanmoins (similar to English “however”) and pourtant (similar to Engli (...)
26With respect to the C-Advs, we extracted examples containing one of the following adverbs: en revanche (similar to English “on the other hand”), par contre (similar to English “on the other hand”), au contraire (similar to English “on the contrary”), à l’opposé (similar to English “to the contrary”), à l’inverse (similar to English “to the contrary”), inversement (similar to English “conversely”) or a contrario (similar to English “conversely”). These seven C-Advs were selected because they express a simple contrast or antithesis, which is the type of contrast that corresponds best to the contrastive interpretation conveyed by E-Pros3.
27As for the E-Pros, we focused on clauses with the masculine singular form lui “him” and the masculine plural form eux “them”. The female forms of the E-Pro elle “her” and elles “them” were disregarded, since they cannot be formally distinguished from the female forms of the clitic pronoun, i.e. elle “she” and elles “they” (Cappeau, 2004). Consequently, as shown in example [12a], it is difficult to set apart clitic left dislocations (with elle(s) as a clitic resuming the dislocated subject) from topic shift constructions (with elle(s) as E-Pro), unless relying on other cues such as prosody or punctuation, which are not always available, especially in informal written French. This ambiguity does not exist for the masculine pronouns, because their emphatic forms (lui “him” and eux “them”) are different from their clitic ones (il “he” and ils “they”) (see [12b]).
[12a] |
Le moineau domestique adore les graines de tournesol. La mésange bleue [elleclitic/elleE-Pro] préfère les cacahuètes. |
|
‘The house sparrow loves sunflower seeds. The blue tit [sheclitic/herE-Pro] prefers peanuts.’ |
[12b] |
Le moineau domestique adore les graines de tournesol. Le rouge-gorge [ilclitic/luiE-Pro] préfère les vers de farine. |
|
‘The house sparrow loves sunflower seeds. The robin [heclitic/himE-Pro] prefers mealworms.’ |
- 4 We thank Piet Mertens and Serge Verlinde for their extractions.
28In total, our dataset consists of 12,411 cases with a C-Adv and 5,252 with an E-Pro. With respect to the LM corpus, examples of the C-Advs and E-Pros were extracted by Piet Mertens (KU Leuven) and Serge Verlinde (ILT KU Leuven)4. As for the five other corpora, we performed the data extraction ourselves using the concordancer AntConc (Anthony, 2018).
29In this section, we present and discuss the results of our corpus study on the choice between C-Advs and E-Pros in different registers of French. We start with an overview of their frequency in the six corpora (§4.1), followed by an analysis of the syntactic (§4.2) and semantic (§4.3) properties of the subject in one of the positions where the two marking strategies are interchangeable, i.e. in cases where the C-Adv or E-Pro occurs between S and fin-V.
- 5 These studies only distinguish between written and spoken French and do not take into account diffe (...)
30Before focusing on the number of occurrences of C-Advs and E-Pros in the position between S and fin-V (§4.1.2), it is interesting to have a look at their overall frequency in the different registers. The frequency of the seven C-Advs (see §3.2) is provided in Table 2. The results show that there is a clear register difference: the C-Advs are much more frequent in the informal written corpus (YCC) and the spoken corpora (OFROM, CFPP and CRFP) (between 259 and 460 occurrences per million words) than in the two formal journalistic written corpora (LM and ER) (between 81 and 131 occurrences per million words). This finding nuances the claims in the linguistic literature that the frequency of C-Advs differs between spoken and written French (e.g., Bilger & Cappeau, 2003; Csűry, 2001; Danjou-Flaux, 1980; Masseron & Wiederspiel, 2003)5: our data show that the level of formality (formal versus informal) has more influence than the mode (written versus spoken).
Table 2 – Overall frequency of C-Advs per corpus
|
LM |
ER |
YCC |
OFROM |
CFPP |
CRFP |
Absolute frequency |
3,377 |
5,995 |
2,323 |
280 |
322 |
114 |
Relative frequency |
131 / million words |
81 / million words |
381 / million words |
277 / million words |
460 / million words |
259 / million words |
31Interestingly, no such register difference is attested for the E-Pros, which occur almost as frequently in all corpora (between 40 and 73 occurrences per million words), as is represented in Table 3.
Table 3 – Overall frequency of E-Pros per corpus
|
LM |
ER |
YCC |
OFROM |
CFPP |
CRFP |
Absolute frequency |
1,881 |
2,938 |
309 |
51 |
43 |
30 |
Relative frequency |
73 / million words |
40 / million words |
51 / million words |
50 / million words |
61 / million words |
68 / million words |
32We hypothesise that this quantitative difference between C-Advs (register effect) and E-Pros (no register effect) is due to differences with respect to specific adverbs. The set of C-Advs is more heterogeneous, and the results abstract away from individual frequency differences between the adverbs. For example, the C-Adv par contre “on the other hand” is especially frequent in informal written and spoken French, whereas en revanche “on the other hand” occurs most often in formal written French (Brysbaert & Lahousse, 2020b).
33Also note that the C-Advs (Table 2) are overall more frequent than the E-Pros (Table 3): the relative frequencies of the C-Advs range from 81 to 460 occurrences per million words, whereas those of the E-Pros vary between 40 and 73 occurrences per million words. This can be partly explained by the fact that C-Advs have a larger distribution than E-Pros. As was shown in examples [9a-b] above, E-Pros with a topic shift interpretation cannot occur before nominal subjects. Hence, when users of French want to add contrastive marking before the nominal subject, they can only opt for a C-Adv.
- 6 The C-Advs occur most frequently before the subject (Brysbaert & Lahousse, 2020a).
34If we zoom in on the cases with a C-Adv or E-Pro occurring between the S and the fin-V, as in example [11] above, it becomes clear that E-Pros are much more frequent in this position than C-Advs6. The results in Table 4 also show that the frequency of C-Advs and E-Pros in this specific syntactic position is register-dependent. The C-Advs occur most often between S and fin-V in the formal written corpus LM (14%), with only a few rare cases in spoken French (1%). The E-Pros, on the other hand, appear more frequently in this position in the informal written corpus (83%) and spoken corpora (82%), compared to the two journalistic corpora (54% and 61%). This finding suggests that, with respect to their frequency in different registers, these two non-prosodic marking strategies are more or less “complementary”: as the level of formality decreases, the C-Advs appear less often between S and fin-V, while the E-Pros occur more often in this position.
- 7 The relative frequencies were calculated with respect to the total number of analysed examples of C (...)
Table 4 – Position between S and fin-V: frequency of C-Advs versus E-Pros7
|
LM |
ER |
YCC |
OFROM, CFPP and CRFP |
C-Adv between S and fin-V |
14% (365) |
6% (276) |
3% (47) |
1% (8) |
E-Pro between S and fin-V |
61% (1,093) |
54% (1,519) |
83% (235) |
82% (67) |
35In what follows, we focus on these examples with a C-Adv or E-Pro between S and fin-V, to determine which language-internal factors play a role in the choice between the two marking strategies, more specifically the syntactic (§4.2) and semantic (§4.3) properties of the subject.
36Table 5 provides an overview of the syntactic type of subject per register for cases with a C-Adv occurring between S and fin-V (see example [11] above). With respect to the spoken register, it is not possible to make any generalisations due to scarceness of data (see Table 4 above). We will therefore not consider the examples from the spoken corpora in the discussion of the results for C-Advs.
Table 5 – C-Adv between S and fin-V: syntactic type of S per register
|
LM |
ER |
YCC |
OFROM, CFPP and CRFP |
Total |
Noun |
79% (289) |
75% (206) |
72% (34) |
62.5% (5) |
77% (534) |
Relative pronoun |
6% (23) |
12% (32) |
4% (2) |
25% (2) |
8% (59) |
Other pronoun |
13% (49) |
13% (37) |
17% (8) |
12.5% (1) |
14% (95) |
Other |
1% (4) |
0.4% (1) |
6% (3) |
0% (0) |
1% (8) |
Total |
100% (365) |
100% (276) |
100% (47) |
100% (8) |
100% (696) |
- 8 We did not include the examples from the column “OFROM, CFPP and CRFP” (8 in total) and the row “Ot (...)
- 9 We present the examples as they occur in our corpora, without correcting any typographical, spellin (...)
37As becomes clear from Table 5, there are no significant differences between the registers (Fisher exact test (4, N = 680), p = 0.14)8. In all corpora, the subject modified by the C-Adv most frequently takes the form of a noun (between 72% and 79%) (see [13])9. The C-Adv can also modify a relative pronoun (between 4% and 12%) (see [14]) or another type of pronoun (between 13% and 17%), such as ceux-ci “these”, d’autres “others” or rien “nothing”, illustrated in [15]-[17]. In some rare cases (between 0.4% and 6%), the subject is neither a noun nor a pronoun, but for example a verb phrase (VP) (see [18]).
[13] |
Ce sont surtout les actifs qui lisent des livres, mais aussi des journaux, et ce sont eux aussi qui utilisent un ordinateur et écoutent la radio; les retraités, en revanche, lisent de moins en moins, préférant la télévision. (LM) |
|
‘It is mainly working people who read books, but also newspapers, and they are also the ones who use a computer and listen to the radio; pensioners, on the other hand, read less and less, preferring television.’ |
[14] |
De toute évidence, la rigueur scientifique n’est pas le principal apanage de ces ouvrages qui, en revanche, flattent l’imagination. (LM) |
|
‘Obviously, scientific rigor is not the main prerogative of these works which, on the other hand, flatter the imagination.’ |
[15] |
Les Auriverde ont appliqué la même recette qu’en demi-finale pour mettre les Cubains dans le rouge. Ceux‑ci, en revanche, ont multiplié les bourdes, payant clairement leur manque d’expérience. (ER) |
|
‘The Auriverde applied the same recipe as in the semi‑finals to drive the Cubans into a corner. These, on the other hand, multiplied the blunders, clearly paying for their lack of experience.’ |
[16] |
Si quelques‑unes de ces affiches paraissent avoir assez mal vieilli, d’autres, en revanche, frappent par leur caractère prémonitoire. (LM) |
|
‘If some of these posters seem to have aged rather badly, others, on the other hand, strike by their premonitory character.’ |
[17] |
Si la récupération du pont de l’Ascension peut être admise, rien en revanche ne justifie le deuxième samedi imposé. (ER) |
|
‘If the recuperation of the “bridge” of Ascension Day can be admitted, nothing on the other hand justifies the second imposed Saturday.’ |
[18] |
Mélanger les ingrédients avec une spatule, ça c’est fastoche. Râper les carottes, en revanche, fait mal aux bras. (ER) |
|
‘Mixing the ingredients with a spatula is easy. Grating carrots, on the other hand, hurts your arms.’ |
38The results for the examples with an E-Pro between S and fin-V are summarised in Table 6.
Table 6 – E-Pro between S and fin-V: syntactic type of S per register
|
LM |
ER |
YCC |
OFROM, CFPP and CRFP |
Total |
Noun |
90% (984) |
84% (1,283) |
54% (126) |
22% (15) |
83% (2,408) |
Relative pronoun |
10% (107) |
16% (236) |
45% (106) |
78% (52) |
17% (501) |
Other pronoun |
0.2% (2) |
0% (0) |
1% (3) |
0% (0) |
0.2% (5) |
Other |
0% (0) |
0% (0) |
0% (0) |
0% (0) |
0% (0) |
Total |
100% (1,093) |
100% (1,519) |
100% (235) |
100% (67) |
100% (2,914) |
39In contrast to the C-Advs (Table 5), the syntactic type of subject in examples with an E-Pro is clearly influenced by the register. In the two journalistic corpora (LM and ER), the subject is a noun in the majority of the cases (between 84% and 90%) (see [19]), whereas in the spoken corpora (OFROM, CFPP and CRFP), the subject predominantly takes the form of a relative pronoun (78%) (see [20]). The results for the informal written corpus (YCC) are situated in between, with nominal subjects being slightly more frequent (54%) than relative pronouns (45%).
[19] |
Le conducteur n’est pas parfaitement bien assis, mais les passagers installés à l’arrière, eux, sont à l’aise. (LM) |
|
‘The driver is not perfectly seated, but the passengers in the back, them, are comfortable.’ |
[20] |
sinon à côté de ça on a des des gens comme euh Éric Orsenna un autre euh académicien qui lui veut défendre la grammaire (OFROM) |
|
‘otherwise besides that we have people like uh Éric Orsenna another uh academician who him wants to defend grammar’ |
- 10 We did not include the examples from the row “Other pronoun” (5 in total) in the statistical test, (...)
40These register differences are statistically significant (χ2 (3, N = 2,909) = 348.37, p < 0.001, with a moderate effect size, Cramer’s V = 0.35), but not straightforward to explain10. It could be the case that relative pronouns are in general more often used in the informal written (YCC) and spoken corpora (OFROM, CFPP and CRFP), but it seems unlikely that this would be the only reason for the register difference, especially because no similar tendency is attested for the C-Advs (Table 5).
41A comparison of the results for the C-Advs (Table 5) and the E-Pros (Table 6) reveals that the use of E-Pros is more restricted than the use of C-Advs. First, E-Pros cannot appear when the subject is a VP as in [21] (compared to [18] above).
[21] |
Mélanger les ingrédients avec une spatule, ça c’est fastoche. Râper les carottes, *lui, fait mal aux bras. (ER) |
|
‘Mixing the ingredients with a spatula is easy. Grating carrots, *him, hurts your arms.’ |
- 11 The same pattern applies to clitic left dislocations, where in case of an event, the resumptive pro (...)
42This is due to the fact that the E-Pro lui “him” requires a subject referring to a specific discourse referent, which is not the case with VP-subjects, since these describe an event. For the same reason, pronominal subjects such as cela “that” and ceci “this” cannot be modified by the E-Pro lui “him”. In these cases, speakers of French would have to use an E-Pro that is specialised in referring to events, i.e. ça “that”11. However, this results in a topic shift construction which is formally indistinguishable from clitic left dislocation (see [22]), as we explained for the female form elle “she/her” above [12a].
[22] |
Mélanger les ingrédients avec une spatule, ça c’est fastoche. Râper les carottes [çaresumptive/çaE-Pro] fait mal aux bras. (ER) |
|
‘Mixing the ingredients with a spatula is easy. Grating carrots [thatresumptive/thatE-Pro] hurts your arms.’ |
43Second, the use of an E-Pro is very rare in cases where the subject is a pronoun other than a relative one. Our dataset contains only five examples of this type, either with the possessive pronoun le tien “yours” or the demonstrative pronoun ceux “these/those” as the subject (see [23]).
[23] |
Question: Les idées vieillissent-elles? Réponse: Certaines oui, d’autres jamais. Mais un constat évident s’impose: ceux qui les émettent, eux, vieillissent! (YCC) |
|
‘Question: Do ideas age? Answer: Some yes, some never. But one obvious observation is clear: those who express them, them, are getting old!’ |
44Although our corpora do not contain such examples, E-Pros can also occur after pronouns such as certains “some” (see [24]). This shows that it is possible, but rather infrequent, for E-Pros to refer to an indefinite pronoun.
[24] |
Fessée pas fessées, privé de ses jeux favoris, le piquet, etc etc… De nombreuses punitions existent et sont un élément qui fait partie de l’éducation des enfants. Certains, eux, ne punissent pas leur enfant sous couvert de laisser son enfant de développer en liberté, ce qu’on appelle actuellement les enfants roi. (http://latiteorange.blogspot.com/2011/03/reflexion-sur-leducation-des-enfants.html) |
|
‘Spanking no spanking, deprived of his/her favourite games, the picket, etc etc… Many punishments exist and are an element that is part of the education of children. Some, them, do not punish their child under the guise of letting their child develop freely, what is currently called king children.’ |
45This finding is in line with the idea that E-Pros prefer to modify subjects referring to a specific discourse referent. The discourse referent of the subject certains “some” is quite vague – which makes it unattractive for E-Pros – but it is still specific enough to allow modification by an E-Pro. By contrast, with indefinite pronouns such as rien “nothing”, personne “nobody” or tout “everything”, which refer to a very general discourse referent, modification by an E-Pro is impossible as in [25] (compared to [17] above).
[25] |
Si la récupération du pont de l’Ascension peut être admise, rien *lui ne justifie le deuxième samedi imposé. (ER) |
|
‘If the recuperation of the “bridge” of Ascension Day can be admitted, nothing *him justifies the second imposed Saturday.’ |
- 12 It could however also be the case that C-Advs occur less often in subordinate clauses in general. W (...)
46The comparison between the results for C-Advs (Table 5) and E-Pros (Table 6) also reveals that in general, the E-Pros (17%) more often modify a subject relative pronoun than the C-Advs (8%). This difference is statistically significant (χ2 (1, N = 3,502) = 19.40, p < 0.001, with a very small effect size, Cramer’s V = 0.07), and supports our hypothesis that E-Pros preferably modify a subject that has a clearly identifiable discourse referent: relative pronouns have an antecedent which often refers to a clear discourse referent12.
- 13 In our corpora, the vast majority of the examples in this category belong to the type “organisation (...)
47The semantic properties of the subject in examples with a C-Adv/E-Pro between S and fin-V were coded based on the animacy annotation guidelines provided by Zaenen et al. (2004). Following Thuilier and Danlos (2012), we made a three-way distinction between subjects with a human referent (example [26]), an animate referent, i.e. an animal (example [27]) or an organisation/collectivity (examples [28] and [29])13, and an inanimate referent (example [30]).
[26] |
Alain et Martine décidaient de faire les vingt kilomètres, Claude et Mireille, eux, sortaient des sentiers battus pour se rendre à Saint-Germain des Prés. (ER) |
|
‘Alain and Martine decided to do the twenty kilometres, Claude and Mireille, them, went of the beaten track to get to Saint-Germain-des-Prés.’ |
[27] |
Si la guêpe germanique préfère les endroits confinés, le frelon asiatique, lui, est adepte des nids aériens. (ER) |
|
‘If the German wasp prefers confined places, the Asian hornet, him, is a fan of aerial nests.’ |
[28] |
Le premier [i.e. Goodyear] développe actuellement un pneu, l’EMT (Extended Mobility Tyre), aux vertus similaires à celles du PAV et pourrait donc être un partenaire idéal pour Michelin. Bridgestone, lui, ne semble pas s’intéresser à cette voie. (LM) |
|
‘The first [i.e. Goodyear] is currently developing a tire, the EMT (Extended Mobility Tyre), with properties similar to those of PAV and could therefore be an ideal partner for Michelin. Bridgestone, him, does not seem to be interested in this path.’ |
[29] |
Le pays de Galles, pays hôte de l’épreuve, est également qualifié d’office. L’Angleterre, l’Écosse et l’Irlande, en revanche, doivent passer par des tournois qualificatifs organisés cet automne. (LM) |
|
‘Wales, the host country of the event, is also automatically qualified. England, Scotland and Ireland, on the other hand, must go through qualifying tournaments organised this autumn.’ |
[30] |
La faiblesse est quelque chose d’intérieur, qu’on ressent en soi, par exemple on se sent faible quand on est malade. […] La fragilité au contraire est une caractéristique appliquée à quelqu’un après une évaluation faite de l’extérieur. (YCC) |
|
‘Weakness is something inside, that we feel in ourselves, for example we feel weak when we are sick. […] Fragility, on the contrary, is a characteristic applied to someone after an evaluation made from the outside.’ |
48Our goal was to code the animacy status of the discourse referent to which the subject is referring, and not of the linguistic expression itself. For example, in [31], the subject Louis was categorised as inanimate, since it is referring to “the name Louis” and not to “a specific person named Louis”.
[31] |
Pour les prénoms masculins et féminins, les trois premiers ne changent pas, commente l’auteur, Stéphanie Rapoport. Louis, par contre, remonte en 4e position. (ER) |
|
‘For male and female first names, the first three do not change, comments the author, Stéphanie Rapoport. Louis, on the other hand, moves up to 4th position.’ |
49Table 7 provides an overview of the results for the C-Advs. It is clear that this marker does not have a preference for a certain semantic type of subject: C-Advs occur almost as frequently after human subjects (36%) than after inanimate subjects (38%). Modification of an animate subject (organisation or animal) is slightly less common (26%), which can probably be explained by a lower frequency of this type of subjects in general. There is one important difference between the corpora: in the two journalistic corpora (LM and ER), C-Advs combine more frequently with animate subjects than in the informal written corpus (YCC), where they combine more often with inanimate subjects. According to us, this difference is related to the type of text rather than to the level of formality. In the two journalistic corpora, articles often deal with topics facilitating the use of organisational (animate) subjects, such as le gouvernement “the government”, Arsenal or Danone. By contrast, on the discussion platform, there is for example a separate thread about computer problems, which could increase the chance of using inanimate subjects (e.g., the mouse, the software program, etc.).
Table 7 – C-Adv between S and fin-V: semantic type of S per register
|
LM |
ER |
YCC |
OFROM, CFPP and CRFP |
Total |
Human |
38% (138) |
34% (94) |
38% (18) |
37.5% (3) |
36% (253) |
Animate |
24% (89) |
30% (83) |
11% (5) |
12.5% (1) |
26% (178) |
Inanimate |
38% (138) |
36% (99) |
51% (24) |
50% (4) |
38% (265) |
Total |
100% (365) |
100% (276) |
100% (47) |
100% (8) |
100% (696) |
50In contrast to the C-Advs, the E-Pros have a pronounced preference for human subjects (between 60% and 84%) (Table 8). This difference between the C-Advs and the E-Pros is statistically significant (χ2 (2, N = 3,610) = 281.49, p < 0.001, with a small effect size, Cramer’s V = 0.28), and is in line with the hypothesis that E-Pros preferably modify subjects referring to a specific discourse referent (see §4.2). Human subjects have a more clearly identifiable – and active – discourse referent than (organisational) animate and inanimate subjects. As for the C-Advs, we argue that the differences between the corpora are due to the text type rather than to the level of formality. For instance, in the spoken corpora, E-Pros almost never modify an inanimate subject (only 1 occurrence), probably because the interviewees prefer to talk about the human and animate beings in their life rather than about inanimate things, which is also stimulated by the questions asked by the interviewer.
Table 8 – E-Pro between S and fin-V: semantic type of S per register
|
LM |
ER |
YCC |
OFROM, CFPP and CRFP |
Total |
Human |
72% (788) |
69% (1,047) |
60% (140) |
84% (56) |
70% (2,031) |
Animate |
14% (158) |
15% (225) |
16% (38) |
15% (10) |
15% (431) |
Inanimate |
13% (147) |
16% (247) |
24% (57) |
1% (1) |
16% (452) |
Total |
100% (1,093) |
100% (1,519) |
100% (235) |
100% (67) |
100% (2,914) |
51In this paper, we studied the choice between C-Advs and E-Pros, which are two competing marking strategies for contrastive topics in a topic shift context in French. We first showed that C-Advs and E-Pros are only interchangeable (i) when used as topic shift markers with a nominal subject and (ii) when occurring in a syntactic position after this nominal subject (i.e., immediately after the subject or within the predicate). Next, based on a corpus analysis of a subset of these cases, i.e. examples where the C-Adv/E-Pro occurs between the S and the fin-V, in six corpora representing different registers, we provided evidence for the hypothesis that the choice between C-Advs and E-Pros is influenced by (i) language-internal pragmatic constraints and by (ii) the language-external factor register. More specifically, the frequency of E-Pros and C-Advs is register-dependent: as the level of formality decreases, E-Pros appear more often between S and fin-V, while C-Advs are less frequently used in this position. In addition, the use of E-Pros is influenced by pragmatic constraints, in the sense that they typically modify subjects referring to a clearly identifiable – preferably human – discourse referent, whereas C-Advs easily appear in combination with any type of subject. More generally, this paper shows that, when analysing the “free” choice between two comparable linguistic markers, it is important to study the combined effect of language-internal and -external factors (see e.g., Engel et al., 2021 on English).
52Further research is needed (i) to examine whether similar effects can be found for C-Advs and E-Pros occurring in other syntactic positions (i.e., after the fin-V), (ii) to statistically assess the importance of the different factors at play in the choice between C-Advs and E-Pros, and (iii) to study the competition with other topic shift markers which can also appear in several syntactic positions, such as quant à lui “as for him”, de son côté “on his/her side” and pour sa part “for his/her part”.