Postposition of the Subject in Contemporary French: An Exploration of Medium, Register and Genre

Postposition of the subject has been widely explored in the French language but only rarely in relation to the oral medium. It is strongly associated with written discourse of a more formal variety. Through an analysis of a corpus of oral narratives, this paper analyses postposition of the subject in contemporary French syntax and its relationship to questions of medium, register and genre. The paper explores the extent to which postposition occurs in the oral medium, incorporat-ing findings from conversational narratives, traditional stories and contemporary storytelling. It analyses the typology of the most frequent types of example, the reasons why these types of postposition might occur in the discourse in question, and the extent to which questions of medium, register and genre play a role. The paper also discusses the implications for the future of this structure in oral varieties of French.


Research Context
Postposition of the subject in French has attracted the interest of researchers for generations. 1 As is widely acknowledged, the unmarked declarative word order in contemporary French places the subject before the verb (SV: anteposition); for example, je pense, ma mère dit, les politiciens ont décidé que and so on. To put it in typological terms, French is generally regarded as an 'SVO' language. 2 Instances where postposition of the subject is found in declaratives are therefore 'marked' contexts, restricted to a certain number of circumstances, several of which I shall discuss in this article. 3 A broad range of structures are mentioned in grammars of French and in the literature on postposition, demonstrating varying levels of productivity. For reasons explained in Section 1.3, below, the categories at the centre of this article are postposition after relative clauses (e.g. le châle épais que lui a tricoté sa grandmère), postposition after initial spatiotemporal elements (e.g. à Tahiti autrefois vivait un jeune guerrier), postposition involving (predominantly) clitic pronouns after a restricted set of initial adverbials (e.g. à peine a-t-il tiré dessus que…), and postposition in incises constructions (e.g. dit-elle). Other major types include ' absolute inversion', 4 postposition after indirect exclamatives and interrogatives, 5 and in fixed expressions. 6 Note that among these cases, there are examples of both main and subordinate clause usage; and cases where postposition is always nominal (after relatives, after spatiotemporal elements, in ' absolute inversion', and in indirect exclamatives and interrogatives), cases where it is almost always pronominal (after the restricted set of initial adverbials and in fixed expressions), and categories where it can be both nominal and pronominal (incises). 7 There are examples in which the level of constraint is high (e.g. after certain initial adverbials and in many examples after initial spatiotemporal elements) and others where the level of constraint is low and where we are dealing more with tendencies or affinities than with constraints (e.g. postposition after relatives; see Fuchs 1997b: 136).
A wide range of phenomena have been explored from a variety of theoretical perspectives. 8 Early approaches were often 'stylistic', focusing on the textual or pragmatic effects that postposition could have (see Ullmann 1952;Clifford 1973). Much recent work has concentrated on syntactic, semantic and informational factors, sometimes separately but more often in combination, with information structure generally viewed as particularly important, notably in cases where the subject is nominal (Lahousse 2011: 8). 9 Syntactic and semantic factors include analysis of the nature of the three key elements in cases of postposition after initial spatiotemporal elements -that is, the status and function of the initial spatiotemporal element, the properties of the verb (its semantic properties, its argument structure, the range of possible tenses), and the status and function of the subject, as well as the relationships between these three elements (Fuchs and Fournier 2003;Marandin 2003;Borillo 2006;Fuchs 2006). Discussion often centres on the constraints that operate on the order XVS (where X is an unspecified initial element), and on the contrasting properties of the initial element, subject and verb in cases where the order is XSV rather than XVS. In particular, Fuchs and Fournier (2003) argue that there is always a strong syntactic and semantic relationship between X and the verb where postposition is attested (e.g. the element X often forms part of the valency of the verb, as in example (14) below: à Tahiti autrefois vivait un jeune guerrier), as opposed to XSV where the initial element is more often autonomous and detached, relative to the rest of the phrase (e.g. un matin il partit avec pour seule compagnie celle d'un serviteur fidèle). 10 Gournay (2006) offers an analysis of features such as tense and verb type in the case of ' absolute inversion', arguing that there is a need to make a clear distinction between 4 E.g. Vint l'ère des grenades (Romains, cited in Le Bidois 1952: 20 andLahousse 2011: 77; see also Gournay 2006); Arrive la lettre d'un ami (Pourrat, cited in Le Bidois 1952: 20). 5 E.g. je me demande où est passé Paul (Le Querler 1997: 193); on se demande quand a lieu la prochaine pleine lune (Korzen 2014: 3). 6 E.g. 'paraît-il', 'semble-t-il'. 7 Note also that there are multiple ways of categorising postposition, some theorists taking as their starting point whether the subject is nominal or pronominal (e.g. Cappeau 2007;Korzen 2014), others the use of postposition in main versus subordinate clauses (Lahousse 2011), and yet others categorising by type of initial element and then subcategorising within that. The primary interest here is not categorisation, so I am simply listing some of the major types attested; the list is not exhaustive. 8 For an excellent detailed summary of research on nominal postposition, see Lahousse (2011: 11-59). 9 The most extensive recent study of postposition is Lahousse (2011) but see also Fuchs (1997a), Marandin (2003) and Korzen (2014). 10 See the extended discussion in Carruthers and Le Draoulec (2017), from which this example is taken (example 22 in their paper).
the properties of absolute inversion (where there is no initial element before the verb), and those involving an initial spatiotemporal element, as these are structures whose properties are often conflated in the literature. 11 In terms of information structuring, several theorists argue that in many instances of nominal postposition, a postposed subject fulfils the function of 'rheme' (known in some theoretical frameworks as the 'focus'), which is to give new information in the discourse, as opposed to the 'theme' or 'topic', which is the element that the phrase 'is about'. Lahousse sums this up: 'tous les auteurs étudiant VS en français semblent donc admettre, d'une façon ou d'une autre, que le sujet postverbal constitue la nouvelle information de l'énoncé, c'est à dire le rhème, le focus ou l'information nouvelle pour l'allocutaire ' (2011: 20, Lahousse's bold). 12 However, one of the weaknesses of this argument, as Marandin (2003: 357) and Lahousse (2011: 21) point out, is that it does not function straightforwardly in subordinate contexts where the notions of theme/rheme or topic/focus are not easy to apply. Ultimately, Lahousse demonstrates that it is necessary to use a multifactorial approach that takes account of information structuring, syntax and semantics in order to achieve a satisfactory analysis of nominal postposition in the full range of examples attested: 'la distribution du sujet postverbal en français moderne est un phénomène multifactoriel ' (2011: 265).
In the case of postposition after the restricted group of adverbials including ainsi, à peine, sans doute and so forth, there is also considerable emphasis in the literature on the relationship between contexts where there is postposition and those where there is not -in other words, where the adverbial is positioned, for example, postverbally (e.g. 'il a peut-être tort'). As we shall see, this discussion often revolves around questions of medium and register, but Guimier (1997) has also raised the possibility that pragmatic-semantic factors may play a role. In the case of relative clauses, debate often centres on the contextual factors that are particularly compatible with postposition, be these syntactic, semantic, informational, prosodic or stylistic (see, for example, Fuchs 1997a).

The question of medium, register and genre
The substantial body of research on postposition is almost exclusively based on corpora of written texts. Indeed, within the written medium, literary texts dominate, with only relatively small quantities of other forms of written French featuring, such as journalism. One of the reasons for this is the assumed higher level of productivity and therefore the possibility of gleaning larger quantities of data to use as the basis for analysis of the range of syntactic, semantic and informational phenomena mentioned in Section 1.1, thereby gaining a ' description maximale' of postposition (Lahousse 2011: 57). Lahousse further argues that the more limited contexts in which postposition occurs in non-literary written texts and in oral French are a subset of those in which postposition occurs in literary French: 'nous pensons que l'inversion en français écrit non littéraire et à l'oral s'observe dans un sous-ensemble des contextes où elle apparaît dans des textes littéraires ' (2011: 57). There is thus an assumption behind the published research, supported by the examples of postposition quoted in grammars of French, that postposition is a written phenomenon occurring mainly in formal registers. 13 As Cappeau puts it: ' c'est presque exclusivement l'écrit, essentiellement littéraire, qui a nourri la réflexion des linguistes et grammairiens et conditionné notre vision ' (2007: 11 See also the discussion in Korzen (2014: 11), which includes cases where connectors such as ' alors' are used in absolute inversion contexts (e.g. alors arriva le général; arriva alors le général). 12 Note that there are many types and subtypes of topic and focus. See the discussion throughout Lahousse's monograph (2011 139-40). That said, large-scale corpus-based research on the productivity of postposition in different text types, genres or registers of written French has not really been undertaken; it is in fact not clear to what extent there might be variation across different registers and genres within the written medium. Given these core assumptions, research on postposition in the spoken medium is extremely rare, with the notable exception of Paul Cappeau's work (1999Cappeau's work ( , 2007. Cappeau argues that even the typology of structures involving postposition in French has been shaped by how they are used in literature, pointing out that this ' conduit à raisonner sur des exemples qui semblent, quelquefois, à la marge des usages reels […] certains exemples littéraires avec sujet postposé ne peuvent être considérés comme offrant une image fidèle de la langue ' (2007: 140). 14 His research suggests that postposition is very rare but is nonetheless attested in spoken French: of the 132 mini-corpora in the Corpus de Référence du Français Parlé (CRFP), 27% contain no examples of postposition, 26% contain one example, and 47% more than one, with a total of 299 examples across the full corpus of 450,000 words. Blanche-Benveniste (1990: 193) summarises this as roughly one occurrence every half hour of conversation. Coming from the theoretical perspective of the Approche Pronominale, 15 Cappeau categorises the examples in three major classes: those involving nominal subjects, those involving pronominal subjects and those that are 'mixed' and can take both pronominal and nominal subjects. He finds that particular types of postposition are more frequent than others in conversational oral French, notably in certain types of relative (2007: 145ff.), after que in certain constructions clivées (e.g. ' c'est là qu'est toute la difficulté', p. 148) and after comme with the verb ' dire' (e.g. ' alors donc comme l'a dit Danielle…', p. 148). Cappeau also invokes the question of genre, arguing, as we shall see, that one particular type of relative might be a 'genre marker'. I shall return to Cappeau's work at various points in the analysis below.

Research questions and data
The point of departure for this article is the observation, in the light of previous work and comments on register, medium and genre, that relatively speaking, there appears to be a noticeable presence of postposition in a recent corpus of French oral narrative -the French Oral Narrative Corpus. 16 Where Cappeau's corpus contains approximately one postposed subject every half hour, this corpus contains 292 examples in 16 hours 40 minutes, so roughly nine per half hour. The French Oral Narrative Corpus, unlike Cappeau's, is not of vernacular conversational French but of performed stories, including fairy tales, anecdotes, animal tales, fantastic tales, myths, legends and other types. It contains eighty-seven stories which have been recorded in authentic performance contexts, transcribed and annotated using the Text Encoding Initiative (TEI), for a range of structures, including postposition. These are performances of stories for live audiences by professional and semi-professional storytellers, sometimes known as 'new storytellers' to distinguish them from traditional storytellers who learn their stories through an intergenerational oral tradition. New storytellers glean their storylines from written texts (reflecting a wide range of periods and story types) but transform them each time that they are recounted into an oral performance. With the exception of certain key phrases at specific moments in the story, it is very much the storyline rather than the text that is memorised and, in that sense, each performance is 'spontaneous'. 17 However, there is a clear sense in which these stories are more formal in terms of register and more self-consciously 'performed' than conversational stories and there is a connection with written texts through the sources, eclectic though these are. Indeed, this somewhat 'borderline' status of new storytelling between oral and written media makes it a particularly interesting site for the analysis of a phenomenon such as postposition that is traditionally found in formal written varieties but for which there is some, albeit limited, evidence in oral French. The French Oral Narrative Corpus will therefore be at the centre of this analysis, as the springboard for an exploration of medium, register and genre; and it will be referred to in the article as the 'new storytelling' corpus. However, given the highly specific nature of the corpus, I will present, where possible, some comparisons with three other small corpora representing different registers and media: i. a subset of twenty conversational narratives taken from the CRFP corpus: these are informal conversational stories from the same source as Cappeau's data and will be known as the 'mini-CRFP' corpus in order to distinguish the subcorpus used for this paper from the full CRFP corpus; 18 ii. a set of seventeen traditional stories taken from the Aubrac corpus (Tenèze 1975): the Aubrac corpus consists of storytellers who are part of an authentic oral tradition and who learnt their stories entirely orally as part of that tradition, rather than through written texts. It was recorded in the 1960s by a group of anthropologists and is largely in Occitan: this mini-corpus uses seventeen French stories from the collection and will be referred to as the 'Aubrac' corpus; 19 iii. a corpus of twenty published versions of contes of a range of story types (see Mills 2014), which will be referred to as the 'published' corpus.
The corpora are not remotely comparable in terms of size. The main focus here is the new storytelling corpus; the aim with the comparative corpora is to offer three small corpora of around twenty stories each, albeit of very different lengths. This is possible in the case of the mini-CRFP and the published corpora, but only seventeen suitable stories in French are available from the Aubrac corpus. In any case, there is no claim that the corpora can be used for meaningful quantitative comparisons. 20 As outlined in Section 1.1, the article will use the four most common types of postposition attested in the corpus of new storytelling as the basis for analysis: postposition after relative clauses (Section 2), postposition after initial spatiotemporal elements (Section 3), postposition involving mainly clitic pronouns after a restricted set of initial adverbials (Section 4) and postposition in incises constructions (Section 5). 21 The numbers in all cases are small, and in some cases extremely small. Quantitative statistical analysis is simply not possible beyond basic percentages, as the numbers dictate that significance tests would not be meaningful. This is, therefore, essentially a qualitative analysis of postposition in the new storytelling corpus, with basic statistics such as percentages where appropriate and qualitative comparison with the other corpora where possible.
The questions posed in this article are: • Why might these particular types of postposition be found in new storytelling? To what extent might questions of medium, register and genre play a role in the case of this type of discourse? What might their use tell us about the productivity of postposition in this variety of oral French? • More globally, what does their use in this corpus, and the comparison with other registers and media, tell us about use of these types of postposition in different registers and media? Does genre play a role? • And what are the implications for the future of this structure in oral varieties of French?

Relative Clauses
As noted above, Cappeau's analysis of the full CRFP confirms that relative clauses involving postposition are indeed attested in informal oral French. In terms of types of relative, those introduced by que and où are the most frequent, in that order. 22 However, given the size of the full CRFP (450,000 words), the number of examples of postposition in relative clauses is nonetheless very low, at just eighty-eight examples. Cappeau notes that these are dispersed throughout the corpus and are not limited to a small number of subcorpora (2007: 147); they are thus not indicative of idiolectal use by particular individuals. In the twenty conversational narratives forming the mini-CRFP used for this study (i.e. a small portion of the CRFP corpus), there are no relative clauses involving postposition, which is not entirely surprising given Cappeau's statistics.
The number of relatives involving postposition is, comparatively speaking, more substantial in the new storytelling corpus: there are forty-two examples across eighty-seven narratives/142,000 words, spread across a range of storytellers and types of story, plus four examples of ' que' in a construction clivée: 23 (1) c'était justement là qu'était la tortue (Le Corbeau, le rat et la tortue).
As with the full CRFP informal spoken corpus used by Cappeau, que and où are by far the most frequent relatives attested in the new storytelling corpus (accounting for 47% and 40% respectively), with only five examples of other types of relative, mainly involving lequel, desquels and such like, for example: (2) elle tendait ses deux bras en avant à l'intérieur desquels se trouvait le fils d'Ounamir (Ounamir).
The vast majority of the relatives attested occur in the discourse of the narrator: only four appear in the context of direct discourse, that is, in the speech of one of the characters in the story, again probably reflecting their rarity in 'speech'.
As far as relatives involving que are concerned, the distribution with different verbs looks quite distinct from patterns in Cappeau's full CRFP, where a small set of the most common verbs dominates. Relatives with a nominal antecedent in the full CRFP tend to involve the verbs avoir, être and faire, while those with a preceding ce tend to involve faire, être and dire. In the new storytelling corpus, there is a wide range of verbs after nominal antecedents, with no suggestion of clear patterns of co-occurrence, for example: (3) ils vont courir ils courent jusque sur le chemin qu'empruntera le s-le chasseur (Le Corbeau, le rat, la tortue et le daim); (4) elle s'est enveloppée dans le châle épais que lui a tricoté sa grand-mère (Les Babouches d'Aboukacem).
Not surprisingly, given that most events and descriptions relate to the past, both imparfait and passé composé are widely attested, with other tenses also deployed, including the future, as in (3) above. In other words, while small in number, relative que with a nominal antecedent appears nonetheless to have a productive status in new storytelling, in that it is used with a range of verbs and tenses.
In the case of preceding ce, as Cappeau found in the full CRFP, over half the examples are made up of a combination of three of the most common verbs -faire, être and dire: (5) un mari a le droit de savoir ce que fait sa femme le samedi (La Légende de Mélusine); (6) je ne sais pas ce que sera la fin de siècle à Paris (Nouvelle de Alphonse Allais); (7) voilà ce que vous dira la terre (Histoire de femmes [Desnouveaux]).
However, we need to be cautious in positing any significant link with these verbs, as the numbers are extremely small and several other verbs are also attested, such as devenir, savoir, croire, expliquer, désirer: (8) c'est ce qu'expliquait le moine Vrihatsvic à son ami (Iragnaka 2).
Again, although numbers are small, it is clear, broadly in line with Cappeau, that the 'ce que + dire' construction tends to occur in examples of characters' direct speech (9) or in sections where the narrator addresses the audience directly (10): (9) tu entends ce que me dit le chat? (Grand-mère mensonge 3); (10) prêtez vos oreilles à ce que vous dira la terre (Histoire de femmes [Desnouveaux]).
Since the structure is deployed to convey forms of reported speech, both the present tense and the future appear as frequently as past tenses. Cappeau (2007) posits a possible link with register in the case of 'ce que + dire' and 'ce que + être' examples: he notes that in the full CRFP corpus, these are more frequently associated with more formal speech, such as public and political discourse, than 'ce que + faire'. To test this hypothesis, he analyses a corpus of political discourse, where he finds 'ce que + dire' to be particularly frequent. He then goes a stage further, arguing that 'ce que + dire', given its capacity to introduce the speech of others, is in fact a genre marker in political discourse (2007: 147), where conveying the discourse of others is a common feature. The new storytelling corpus, however, suggests no evidence of difference in terms of register between uses of faire, être and avoir in this context. In fact, it is even difficult to speak of 'ce que + postposition' as a productive construction in this type of oral discourse and it is worth noting that some occurrences are almost formulaic in nature and/or repeated, for instance (7) and (10) above, which both occur in the same story.
As with postposition after ce que, some examples are repeated in a formulaic fashion, not least (13), which is repeated twice in the same story.
Postposition after relatives in the written corpus of published contes follows very similar patterns to new storytelling, again in relatively low frequencies: there are only eight examples across the twenty stories, five of which follow que and three follow où, the latter all imperfects marking states, as we might expect. While the five examples following que involve five different verbs and three tenses, the numbers are too low to make arguments about productivity. The same observation regarding quantity applies to the Aubrac corpus of traditional stories, where there are only two examples across seventeen stories, although the particular examples are restricted to one type of relative -'ce que + dire' -thereby aligning more closely with Cappeau's findings for conversational French.
In short, postposition after relatives is rare in the oral medium, with only the new storytelling corpus showing a marginally raised level of frequency. Even here, only two types of relative are attested with any degree of frequency, i.e. with que and où. While there is some evidence of productivity in terms of the range of verbs and tenses attested, other examples are almost formulaic in nature. There is no obvious association with particular storytellers or types of story: the examples are scattered throughout the corpus. Finally, there is no suggestion of any of the relatives attested representing a marker of genre; their relative rarity is probably simply related to medium and register, with only the published and highly performed oral stories in the new storytelling corpus demonstrating any sense of productivity.
In practice, the spatiotemporal elements co-occurring with postposition vary in terms of their relationship with the verb and the nature of this relationship is, in turn, closely linked to the likelihood of postposition of the subject. Fournier (1997: 99) sums up these relations as follows: 'plus le complément antéposé est étroitement articulé au verbe, plus l'affinité avec la postposition du sujet est forte et plus la relation sémantique entre le complément et le verbe est serrée, plus l'affinité avec la postposition du sujet est forte'. In other words, if the spatiotemporal element is a compulsory complement of the verb as in (14) and (15) above, and (17) and (18) below, postposition is effectively compulsory and the order is XVS (Fournier 1997: 100): (17) de sa matrice a jailli Ouranos le Ciel Etoilé (Conte de la création [Buléon]); (18) de la fontaine jaillissaient des pieds charmants (La Légende de Mélusine).
However, in many examples, the relationship between the spatiotemporal element and the verb is not so clear cut. There has been considerable theoretical debate around the analysis of different types of postposition following initial spatiotemporal elements, in particular concerning the status of the initial spatiotemporal element. In cases such as (14), (17) and (18), the spatiotemporal element is very clearly part of the valency of the verb: 'un complément intraprédicatif, fortement régi, sous-catégorisé par V (c'est-à-dire participant de sa valence)' (Fuchs and Fournier 2003: 91) and postposition is, as noted above, effectively compulsory. In other cases, the relationship to the verb is less clear: the spatiotemporal element is, strictly speaking, not part of the valency of the verb but is equally not entirely autonomous, as in (20) and (21), cited by Fuchs and Fournier (2003: 94): (20) A côté de nous consommaient des Arabes, réfugiés par paquets sur les banquettes et qui somnolaient (Céline, Voyage au bout de la nuit); (21) Le 22 décembre dernier, dans un hôpital de Nice, mourait plus discrètement qu'il n'a vécu André B., dit 'Java', qui fut l'un des héros du Journal du voleur (Le Monde).
There are two different, but interconnected, theoretically problematic issues here. The first concerns whether or not the spatiotemporal element is extrapredicative or intrapredicative. Fuchs and Fournier (2003) argue that the spatiotemporal element X in these cases is extrapredicative (largely because it is not part of the verb's valency), while Borillo contends that in spite of being external to the valency of the verb, such elements are intrapredicative, having 'le statut de complément adjoint, externe à la structure argumentale du verbe même s'il reste intraprédicatif ' (2006: 31). But whichever view one takes on this, including the possibility that there is a continuum of predicativity in such contexts, most relevant for this discussion is that this is a key context for examples of postposition of the subject. The second theoretically problematic issue concerns whether or not the spatiotemporal element in any of these cases of postposition can act as a 'frame introducer' in the terms of Charolles (1997) -in other words, whether it can introduce sections of discourse (rather than single phrases) in relation to which it acts as a frame. 28 Fuchs and Fournier (2003) argue that frames require the spatiotemporal element to have a high degree of autonomy, and autonomy is effectively rendered impossible with XVS because of the tight relationship between the spatiotemporal element and the verb. This article does not intend to take up the debate around framing versus non-framing, which is explored in detail in Carruthers and Le Draoulec (2017). Rather, the key point is that, given the relative rarity of subject postposition in the oral medium, it appears to be found fairly frequently after initial spatiotemporal elements in the new storytelling corpus, whether some of these instances constitute frames or not. There are sixty-six examples in total, across a range of storytellers and types of story: forty-two of these are spatial, fifteen are temporal and nine are both. The function of postposition after initial spatiotemporal elements is related to the positions in the conte where it is attested. In short, examples are found both in descriptive sections of stories (where they often occur with present and imperfect tenses) and also in distinctly structural positions, such as the beginning of key sections of the narrative (where they occur with imperfect, passé composé and narrative present). 29 In terms of productivity, it could be argued that since a considerable number of examples are attested with a range of tenses and verbs, there is reason to believe that postposition in this context is fairly productive. On the other hand, there are some indications of fixed and recurring patterns. For example, a recurring type of structure involves the introduction of a new character or idea into the discourse, often a new protagonist -a particularly useful device for an oral storyteller. The subject in these cases is the 'rheme' or 'focus' in the structure, bringing new information, appearing in the contes in this specific context with verbs of appearance and disappearance, most frequently arriver, sortir, naître, jaillir, as in (17) and (19) above. As Marandin puts it: 'le protagoniste apparaît ou disparaît dans une situation qui doit avoir été préalabrement introduite ' (2003: 385). This type of structure thus appears with a restricted set of verbs in the new storytelling data, as it does elsewhere. Indeed, the structure can be embedded in a formulaic construction, as in: (22) des meules sortent des flammes sortent des armes sortent des guerriers sortent des bateaux hérissés de lances comme des oursins (Histoire du Roi Freudi).
Most striking of all, in terms of postposition following spatiotemporal elements, is the fact that one type of structure is particularly frequent in a specific context, that is, 'spatiotemporal element + vivre + subject' at the beginning of the Orientation, as in examples such as (14). Borillo notes that in cases like this, although the spatial element is not necessarily already known in the discourse (as is normally the case for the 'topic'), it is nonetheless ' conventionnellement posé(e) ' (2006: 38) in stories. For Borillo, 'la structure d'inversion locative a pour function d'introduire ou de ré-introduire une entité, de la mettre en scène en lui donnant comme cadre de présentation le lieu dans lequel elle se situe, dans lequel elle est visible, dans lequel elle vit, se manifeste etc. C'est cette entité qui, une fois introduite, a de grandes chances de devenir le topique dans le fragment de discours qui suit ' (2006: 38). In practice therefore, the subject in 'focus' position often becomes one of the protagonists of the story and the 'topic' of the subsequent section of discourse. This type of structure is, therefore, something of a convention in the conte: counterbalancing some evidence of productivity, there is also a formulaic dimension to certain examples of postposition, which would suggest that productivity may be limited.
In terms of the other registers and media analysed, there are no examples of postposition in this context in the twenty conversational narratives taken from the mini-CRFP corpus, and only nine in Cappeau's much larger CRFP conversational corpus. It can safely be concluded that the levels attested in new storytelling are indeed considerably higher than in conversation. There is only one example in the Aubrac corpus of traditional stories, a corpus where language is much less self-consciously performed and much less stylised than in new storytelling. Interestingly, there are also relatively few examples in the corpus of twenty published storiesonly seven, all of which are of similar types to those in the new storytelling corpus, for instance: (23) Sous eux défilait une campagne désolée (Story 6); (24) Il y a de ça bien longtemps vivait en Irlande une vieille femme pas très ordinaire (Story 13).
What tentative conclusions can be drawn about this type of postposition, in terms of register, medium and genre? There are certainly links to register: only the more formal varieties -the new storytelling corpus and to some extent the published stories -contain a critical mass of examples. Questions of genre and medium are more complex. Levels of postposition in this context are strikingly high in the new storytelling corpus but markedly low in both the conversational data (the mini-CRFP, backed up by Cappeau's data) and the Aubrac corpus of traditional stories. Taking the latter together with the relatively low levels in the published corpus, it certainly cannot be argued that postposition after initial spatiotemporal elements is a genre marker for oral narrative or for the conte. However, it may be possible to contend that it is a hallmark of new storytelling. Levels are certainly raised relative to in other types of oral discourse and examples are spread across the corpus in terms of different storytellers and different types of story. It is hard to know if there is enough evidence to speak of this as a clear 'genre marker' for new storytelling, in the way that Cappeau refers to 'ce que + dire' in political discourse, but raised frequency of postposition of the subject after spatiotemporal elements is doubtless a possible candidate.
In terms of medium and register, it would be interesting to investigate whether the raised quantities in new storytelling are attested in other types of oral performance involving high degrees of 'pre-paredness' and/or a strong relationship with written texts. Note, for example, the following use of postposition in a D-Day landings anniversary speech by François Hollande (Ouistreham DD 70 years on, 6 June 2014): (25) Sur ces plages de Normandie s'étend le souvenir d'un affrontement acharné, incertain, décisif. Sur ces plages de Normandie, plages paisibles, flotte encore l'âme des combattants qui ont donné leur vie pour délivrer l'Europe. Sur ces plages tranquilles souffle, quel que soit le temps qui passe, quel que soit le climat des saisons, un seul vent, le vent de la liberté.
Here, in an orally performed speech read from a written text, the president repeats the spatiotemporal element 'sur ces plages' three times, introducing three key elements as the focus or rheme -'le souvenir d'un affrontement acharné', 'l'âme des combattants' and 'le vent de la liberté'. What Hollande's speech has in common with new storytelling, to use Koch and Oesterreicher's terminology (2001), is that (i) in terms of medium, there is some transfer between 'réalisation écrite' and 'réalisation orale'; and (ii) in terms of ' comportement communicatif', both types of discourse share key properties, such as ' communication publique', 'interlocuteur inconnu', 'monologue', ' communication préparée'.

Initial adverbials triggering postposition
There is a series of adverbials that normally trigger postposition of the subject if they appear at the head of the phrase and which feature in the new storytelling corpus in small but sufficient numbers (thirty-three in total) to merit some exploration -these being à peine, peut-être, sans doute, aussi and ainsi. 30 Most occur in postposition contexts only with pronominal subjects, but ainsi can also occur with nominal subjects. It is important to note that all these elements can appear elsewhere in the clause, notably postverbally, where no postposition of the subject occurs: (26) il avait peut-être décidé de partir plus tôt; (27) elle veut sans doute vivre près de sa famille.
In other words, these adverbials are by no means consistently linked to compulsory postposition of the subject; only where they appear at the head of the clause do they trigger postposition and there is no evidence that this position in the clause is preferable to any other. Indeed, everything suggests that an adverbial in initial position followed by postposition is strongly associated with more formal registers, not least because typologically the ' adverb + verb + subject' structure is viewed as a remnant of an older word order, commonly referred to in theoretical terms as TVX (topic-verb-other constituents). 31 Comments on contemporary French frequently note that the appearance of this group of adverbials without postposition, as in examples (26)-(30), is much more likely in the oral medium and in more informal registers (see, for example, the comments in Guimier 1997: 54, 70). In addition to cases where the adverbial is postverbal or where que is used, there are instances where the adverb appears at the head of the phrase and there is simply no postposition -that is, in precisely the context where postposition is normally triggered: (31) peut-être elle voulait plus me voir? (Guimier 1997: 53).
Referring to examples such as these, Guimier (1997: 53) comments that they are most common in 'un niveau de langue très familier voire vulgaire' and ' caractéristique de l'oral' but also explores the relationship between the adverb and the verb, arguing that some examples are semantically or pragmatically motivated and 'ne sont pas forcément la marque d'un style négligé ' (1997: 57). In the case of à peine in initial position without postposition, Guimier comments that 'il semble ici que la construction soit extrêmement rare et caractéristique de la langue populaire ' (1997: 80). So, in the majority of cases, structures that do not involve postposition, whether they involve postverbal positioning of the adverbial, use of ' que' or quite simply no postposition, are viewed as strongly associated with informal and oral usage, even if in some cases other semantic or pragmatic factors are also considered relevant.
Postposition most frequently occurs with a pronoun subject but also can appear with a nominal subject which is then repeated as a pronoun (usually known as ' complex inversion' or ' complex postposition'): (32) peut-être Pierre viendra-t-il (Guimier 1997: 43).
As Guimier notes regarding 'à peine temporel', in a structure taking the form 'à peine P que Q', 'une relation chronologique s'établit entre les deux propositions: l'événement de Q suit immédiatement celui de P dans le temps ' (1997: 80). This relation of chronological progression between events is, by wide consensus across a range of theoretical approaches, absolutely fundamental to the construction and interpretation of 'narrative' discourse. As Smith puts it, speaking of the discourse mode of 'Narrative': 'we interpret the events of a narrative as occurring in sequence, one after another ' (2003: 14). Chronological sequence is the default interpretation for juxtaposed event clauses, with alternative interpretations normally only coming into play if signalled explicitly (e.g. with an adverbial indicating a contradictory temporal relation, such as anteriority [e.g. auparavant] or simultaneity [e.g. en même temps]). 34 A peine is, as Guimier's comment makes clear, highly compatible with narrative sequence, reinforcing the sense of chronological progression that is already the default interpretation of two juxtaposed events in narrative discourse.
Aussi (19% of examples) is also highly compatible with narrative sequence: (36) il était un jeune homme et il allait lui succéder un jour ou l'autre donc il allait l'emmener aussi convoqua-t-il… (Le Marchand). Guimier (1997: 72) classifies aussi when in initial position as 'un signe de consécution', although he also makes clear, quoting Zenone (1983: 205), that sequence is not the only possible semantic component; this interpretation of ' consécution' may express ' cause-effet' and/or 'raison-conséquence' but it is nonetheless always present. Ainsi (10% of examples) is even more polysemic, but one of its major functions corresponds to what Guimier calls ' ainsi consécutif ' (1997: 85), where 'il introduit une suite logique pour le contexte avant'. Elsewhere it is usually compatible with sequence: 35 32 For a discussion of the typology of 'à peine', see Guimier (1997: 82). 33 In (33) and (34), the present tense in the main clause is a narrative present, i.e. a present marking a past event.
Note also that ' que il' is a direct transcription of the oral. 34 This does not mean that sequence is always made explicit: connectors that mark or reinforce sequence, such as ensuite or puis, may be found but it is also possible (indeed common) to find either elements where sequence is only one component of their semantic properties (e.g. et and alors) or parataxis, where no connecting element is attested. 35 For a typology of ainsi, see Lahousse (2011: 93). (37) ainsi fut consacré l'échange (Fionn Le Blanc). 36 In summary, adverbials that are compatible with temporal sequence, even if this is neither the only discourse relation they convey nor indeed their central semantic property, constitute the vast bulk of this category, that is, 84% of initial adverbials triggering postposition. This might help to explain their presence in this form of oral narrative discourse, especially given the broader context of extreme rarity in this position in the spoken medium. The overall number of examples is too low and the examples too scattered throughout the corpus to speak of any firm patterns in terms of storyteller or type of story.
In terms of the three other corpora, there are no examples at all of postposition in this context in either the Aubrac (traditional storytelling) or the mini-CRFP conversational narrative corpus. In the published written corpus, numbers are miniscule -although examples are very similar to those in the new storytelling corpus (one of ainsi and two of à peine), for example: (38) A peine avait-elle ouvert la bouche qu'elle lâchait le bâton (Story 6).
The low presence of these examples in the most informal oral variety (mini-CRFP) and the more informal of the two storytelling corpora (Aubrac) is not at all surprising and correlates with the comments alluded to above concerning register and medium. It also correlates with Cappeau's findings (2007), where this category is barely mentioned in relation to the larger CRFP corpus. What is somewhat more surprising is the fact that levels are, relatively speaking, slightly raised in the oral conte corpus relative to the written conte. It is possible that those elements, notably à peine, might serve, in the context of performance, as strong markers of sequence at the head of the phrase, or for the dramatisation of the rapidity of the second event (in the main clause) after the first, or simply as a stylisation device that separates out the language of the conte from everyday usage. But given the quantity of examples, this should perhaps not be overstated. There is also a question around whether certain storytellers might make more frequent use of such elements than others: six of the sixteen examples of à peine are articulated by the same storyteller, for instance.

Incises
Incises are instances of verbs of speech (dire, répondre etc.) that follow or are inserted into sections of direct discourse (e.g. ' dit-il', ' a-t-elle ajouté', 'répondit Maman' etc.). In written texts, the environment of direct discourse is a very common site for postposition of the subject, which is deployed textually to signal 'un décrochage syntaxique et discursif' (Riegel,Pellat and Rioul: 137), separating the sections of direct discourse from the rest of the text. In practice, most examples cited in grammars of French are written and, indeed, usually literary.
In spoken varieties of French, postposition in this context is extremely rare and often limited to certain verbs (notably dire, faire and the auxiliary avoir) and to particular registers and genres. As Blanche-Benveniste puts it, postposition is found ' dans des situations de langage public et de langages professionnels [avocats, jurists, medias]' (2010: 63). However, it is not possible to argue that a different construction is used in spoken French in every instance where a written text might use an incise: in practice, although oral varieties of course contain forms of represented discourse, the distribution of these forms and their structures does not necessarily take the same form as in written varieties. 37 In informal spoken French, embedded 36 The figures for the remaining two adverbials, i.e. the two that are not in any way linked to temporal sequence, are extremely small, peut-être occurring in 14% of cases and sans doute in 3%: e.g. peut-être est-elle partie (Grandmère mensonge 1), sans doute avait-il été un endroit généreux en d'autres périodes (1001 Nuits 3). 37 For a full analysis, see Marnette (2005) and Rendulic (2015). See also Blanche-Benveniste (1997: 107-10). direct discourse is extremely common. Marnette's analysis (2005: 140) of a corpus of informal spoken French narratives (also taken from the CRFP) calculates that direct discourses occupy 27% of the text (the most frequent category is indirect discourse which occupies 59%, e.g. 'il ajoute que…', 'Maman disait que…', 'Pierre a répondu que…'). Direct discourse in these very informal registers tends to appear accompanied by a verb of speech (usually positioned before the direct discourse) with the order SV, as the following examples from the mini-CRFP demonstrate: 38 (39) ils arrivent -anglais + je lui dis non + je cherche un poste quoi + je lui dis peut-être il y avait euh + quelque chose comme ça j'ai dit; (40) une fois j'avais une femme qui avait les médicaments pour maigrir je sais pas si tu connais les amphétamines /ce sont, 0/ des médicaments qui sont utilisés pour couper l'appétit mais ça a énormément d'effets secondaires (L2 ah oui) et (L2/ce sont, c'est/ des excitants les amphétamines) ce sont des amphéta-ouais ce sont des excitants /les amph-, XX/ des propriétés anorexigènes et donc et je lui dis euh je vous écris sur la boîte elle me dit non ah je /lui dis, lui ai dit/ ah bon vous connaissez elle me dit non non elle me dit de toute façon ça servira à rien j'ai dit ah bon ça servira à rien elle /me, m'a/ dit oui.
The verb of speech is overwhelmingly dire but faire is also attested as a verb of speech in the most informal registers.
Again, the interesting corpus in our data in terms of postposition is that of new storytelling. The other three corpora conform to predictable patterns. The written corpus of contes is, as we should expect, rich in postposition in an incises context: even in a small corpus of twenty stories, there are 115 instances of postposition in this context, the majority with nominal subjects (74%), the remainder with pronominal (26%). The mini-CRFP conversational narrative corpus contains no examples of postposition (note that Cappeau 2007 found extremely few examples in the full CRFP corpus) and the Aubrac corpus of traditional stories contains only two. In both of the latter corpora, it is the other forms of reported discourse noted above with regard to the most informal varieties that dominate -that is, indirect discourse or direct discourse with verbs of speech but no postposition. Additionally, the Aubrac corpus contains examples of direct discourse where que is inserted before the verb of speech, again with the SV order preserved: 39 (41) alors qu'est-ce que tu fais là ma pauvre petite bête qu'il lui dit le maçon (Les Trois Poulettes).
In the new storytelling corpus, the situation is more complex. ) and, given the fact that these speech acts are ' events' on the narrative line, the examples occur mainly with the narrative present or the passé composé, or (in the small number of stories where the passé simple is the main narrative tense) with the passé simple. 40 As with the published corpus, most subjects are nouns rather than pronouns, with the statistics more extreme in this case at 91% nominal, 9% pronominal.
It is important to note that direct discourse is extremely common in the new storytelling corpus. All research on oral narratives -whether medieval or modern, conversational or performed -testifies to the importance of direct discourse: substantial quantities of direct discourse are in fact cited by Hymes (1974) and many subsequent scholars as a key performance feature, alongside gestures, sound effects, asides and so forth. 41 Given the substantial quantities of direct discourse involved in storytelling, it is therefore the case that postposition occurs in only a minority of direct discourse environments, unlike the published stories where it is the norm. In the new storytelling corpus, the vast majority of the time we find, as in other types of oral narrative, either direct discourse with subject-verb word order (usually positioned before the direct discourse but after is also possible), or examples where there is no verb of speech at all. Both phenomena are illustrated in the following example, where the central section contains no verbs of speech: 42 (42) les pauvres quarante africains ils se sont jetés par terre la tête dans les bras ils se sont dit notre dernière notre dernière heure est venue il va nous écrabouiller c'est terrible mais lui le génie il était content tout content il dit ah vous m'avez délivré ahaha ce que je suis heureux mais j'aurais pu rester trois mille ans dans cette bouteille ça s'est déjà vu des choses comme ça oh mais je suis trop content vous m'avez délivré non vraiment. je veux faire quelque chose pour vous je veux vous remercier tiens ben je vous accorde un voeu à chacun allez alignez-vous là. ah vous êtes quarante c'est beaucoup ça fait rien allez chacun aura un voeu qui sera exaucé toi le premier qu'est-ce que tu veux? ben moi j'aimerais bien être blanc il y a des avantages à être blanc quand même bon ben tu es blanc et toi qu'est-ce que tu veux? moi aussi je voudrais bien être blanc bon ben accordé et toi qu'est-ce que tu veux? moi aussi je voudrais bien être blanc bon ben tu es blanc et toi? moi aussi je voudrais bien être blanc et il va comme ça jusqu'au trente-neuvième alors il se tourne vers le quarantième il lui dit toi aussi tu veux être blanc comme les autres? non moi je veux qu'ils redeviennent tous noirs (Les 40 Africains et le génie).
The fact that there may be no verb of speech and therefore no formal identification of the speaker (or 'locuteur' in énonciation terms) is, of course, unproblematic in oral narrative because the storyteller can use intonation or pitch to distinguish between characters' voices in the context of an oral narrative, especially where there is an element of performance, as is the case with this corpus.
In summary, in terms of our consideration of questions of medium, register and genre, the picture in relation to incises is not entirely surprising: medium is the primary determiner of postposition with verbs of speech, and register is also relevant. The case of new storytelling is interesting. It is a fundamentally oral form of discourse, where many features correlate strongly with other types of oral narrative, such as conversational or traditional stories. However, there are some features of this genre that show clear influence from conventional written patterns. On the basis of an analysis of temporal features (e.g. tense, connectors), I have argued (Carruthers 2005: 127) that new storytelling represents ' a borderline type of discourse, a type of oralité seconde which is fundamentally oral, but where written patterns exert an important influence'. 43 I would posit here too that the presence of postposition with verbs of speech is due to the borderline status of the new storytelling genre which in turn is, at least in part, linked to the role of written texts as the original source of many of the new storytellers' stories -a factor that clearly distinguishes new storytelling from traditional storytelling (see Section 1, above). There may also be a connection with the preferences of individual storytellers: of the eighteen storytellers, there are seven who use little or no postposition in this context and four where it occurs very frequently (together producing 79 of the 115 examples). It is impossible to make robust statistical claims here; two of these storytellers offer a substantial contribution in terms of volume of stories in the corpus but, on the other hand, the same is true of several of those who almost never use postposition with direct discourse. Moreover, the quantities of direct discourse vary across stories, although it is generally very frequent. Equally, there are no obvious links with story type or tense usage. I would thus hesitate to identify postposition with verbs of speech as a genre marker in new storytelling, especially as it is far from being the dominant form of represented discourse and its use varies between different storytellers. Nonetheless, the range of ways in which discourse is reported certainly reinforces the sense that new storytelling is a borderline genre in terms of linguistic patterns associated with different media, demonstrating a distinctive mixture of oral and written patterns.

Conclusions and further questions
To return to the research questions, four contexts for postposition have a significant presence in new storytelling and these have formed the basis for this paper. The raised levels of postposition after initial spatiotemporal elements and after à peine, aussi, ainsi and so on seem to be, at least in part, related to the fundamental properties of oral new storytelling and less exclusively due to medium and register. Both contexts are certainly more frequent sites for postposition in new storytelling than in other types of oral discourse. In the case of à peine specifically (and to some extent aussi and ainsi), clear marking of temporal sequence is particularly compatible with the use of these adverbs at the head of the phrase, although of course postposition can still be avoided through use of que. In the case of spatiotemporal elements at the head of the phrase, these are particularly important for the performance of the story in terms of its evolution in time and space. In other words, the spatiotemporal positioning of events and descriptions in the evolution of the story, and the progression of narrative events in sequence, both seem to play a role in increasing the presence of two particular contexts in which postposition might -and in a number of cases does -occur. Indeed, at times both types of usage of postposition appear to be somewhat formulaic, something which is also highly compatible with the performed oral story. It is difficult to know whether we can go as far as describing either of these contexts for postposition as a genre feature. Certainly, relative to the other types of oral data examined, and bearing in mind Cappeau's research, new storytelling seems to constitute an oral discourse genre where levels of postposition of the subject are raised and the reasons for this might be connected to other features of the genre, notably its narrative structure and its performance context.
In terms of broader questions of register and medium, when new storytelling is viewed alongside the other mini-corpora analysed, it is clear that the presence of postposition ranges from relatively rare to extremely rare and tends to be restricted to certain types of structure. Questions of register and medium are undoubtedly relevant. Incises, for example, are tightly linked to medium and register: they are widespread in the formal written medium of literature, almost non-existent in informal oral varieties and present in not insignificant quantities (albeit minority rather than dominant), in the more formal 'borderline' genre of new storytelling. Similarly, in the case of relatives, postposition seems also to be related to medium and register; it is certainly attested in the dataset analysed here, but productivity is very limited in even the most formal varieties of the oral medium, with particular types of relative dominating (où and que), as they do in other less formal oral varieties.
More work does need to be done on postposition in different genres, media, registers and text types in order to get a more accurate picture of the broader context. To what extent is it attested in a range of written text types? And which oral discourse types show raised levels of postposition of the subject, and why? To what extent is its presence or absence linked to usage of other features that are strongly associated with particular registers or media, such as retention of negative ne or the passé simple? At this stage, it is certainly not possible to say that the apparent raised levels of postposition in the performed oral story are indicative of future productivity of the structure more generally in French. Quite the opposite: it would appear that its use is closely connected to certain features of the genre, not least the fact that it is a very particular type of oral discourse which is performed rather than informal, spontaneous within certain conventions, and where written texts are an important source. It would be interesting to see to what extent other borderline or hybrid genres such as speeches or news reports exhibit raised levels of postposition. My hypothesis would be that the borderline nature of new storytelling in terms of medium and its particular narrative structures converge to raise the levels of postposition but that this particular mixture of factors will probably be found in few other types of oral discourse. However, while it is crucial not to suggest that postposition is productive in oral French, or that its use is likely to increase in the future, it is nonetheless important to point out that the combination of medium, register and genre can mean that certain types of oral discourse can produce patterns that are quite distinctive; and it is vital that these are explored so that we have a full picture of contemporary oral French. As Blanche-Benveniste puts it (her bold): ' c'est un artifice indéfendable de vouloir réduire l'ensemble de la langue parlée aux seules manifestations spontanées, et de ramener la compétence linguistique en langue parlée à cette seule dimension. L'existence de différents genres dans la langue parlée montre qu'il est bon de multiplier les angles d'observation, et d'envisager plusieurs sortes de compétences linguistiques ' (1997: 62).