Navigation – Plan du site

AccueilNuméros48Multiple retractions in spoken Fr...


Multiple retractions in spoken French and spoken German. A contrastive study in oral performance styles

Les allers-et-retours sur l’axe syntagmatique en allemand et en français parlés. Une étude contrastive de rhétorique orale
Peter Auer et Stefan Pfänder
p. 57-84


Si l’on en croit une idée répandue, la langue française permet d’exprimer les choses plus clairement que l’allemand. Nous sommes toutefois loin d’avoir tranché la question de savoir si la fameuse clarté du français est un mythe ou une réalité. Et pourtant, les linguistes se rejoignent sur l’argument principal pour justifier la clarté du français, à savoir la séquentialité fixe des éléments au niveau phrastique et textuel. Mais on n’a jusqu’à présent relevé que peu de témoignages empiriques mettant en évidence une grande différence entre l’allemand et le français dans leurs styles performés respectifs, en particulier chez des locuteurs non « professionnels ». Nous avons choisi dans cet article de nous consacrer aux rétractions doubles (ou multiples), un phénomène essentiel quant à l’organisation séquentielle d’énoncés oraux. La rétraction (retraction) se caractérise par le réinvestissement du même créneau paradigmatique. Une rétraction peut avoir lieu avant la clôture syntaxique (très souvent avant la formulation du rhème) et représente alors une forme de reformulation ou ré-édition du projet syntaxique émergent. Elle peut apparaître après complétion du projet syntaxique (et donc après le rhème), comme une expansion paradigmatique rétrospective. La présente analyse, qui se fonde sur des données semi-conversationnelles tirées d’interviews libres et informelles, montre clairement que, si les doubles rétractions en position pré- et postrhématique sont observables aussi bien en français qu’en allemand, les doubles rétractions postrhématiques manifestent un bon niveau stylistique en français — signes de cohésion, elles caractérisent les énoncés « bien formés » —, mais pas en allemand. Nos analyses font également ressortir que, tandis que les rétractions sont liées en allemand à des stratégies de changement de tour, elles permettent en français d’intégrer des matériaux sémantiques et syntaxiques hétérogènes pour former un tout cohérent. En français, les rétractions multiples servent alors à produire, chez un ou plusieurs locuteurs, des tours de parole perçus comme « bien formés » sur le plan rhétorique

Haut de page

Texte intégral


  • 1  See among others: Schmitt (2001) for German and Rühl (2002) for French.
  • 2  Cf. recently Geuen, Kimminich, Rappe & Pfänder 2007.

1There is a long tradition of research on French and German style, but this tradition is mostly concerned with written language, and certainly not with the conversational style of the average speaker of French or German, but rather with that of intellectuals and other public figures.1 The question of whether there are stylistic traditions that differ between Germans and French, and which are relevant in the language performance of people who are not particularly used to, or even trained, to perform in public, is therefore not easily answered. In research on grammar, it is often believed that spoken languages show more similarities than their written counterparts, since the culture-specific norms that may be imposed and enforced on the languages ‘from above’ are less likely to be followed in speaking than in writing. Therefore, oral language is believed to reflect the ‘real’ and ‘authentic’ structure of the language more accurately than written language with its artificial ‘distortions’ and ad-hoc norms, which have partly been formed by puristic notions of what a logical language should look like (cf. Auer 2004). One way of approaching oral (everyday) and written (formal) styles would be to assume, along the same line of reasoning, that the stylistics of the former are more ‘natural’ than those of the latter. They would respond to the same (partly universal) exigencies of talk-in-interaction in similar ways, so that oral styles across languages would be more similar than written styles. Another way of approaching oral style would be to pursue a kind of trickledown model (cf. Hans Naumann’s gesunkenes Kulturgut ‘sunken culture’; Naumann 1922); accordingly, one would expect attenuated and perhaps also distorted reflexes of high culture norms in the language behaviour of the uneducated or less educated classes. Finally, it is possible that oral languages have developed counter-styles or sub-cultural styles that have their own tradition and are time-honored reflexes of a non-official culture which have always been a subcurrent and therefore independent of the official, written ones and their standardizations.2 We are not claiming that any of these positions is the right one (and indeed, not even that one and the same answer holds for France and Germany), but rather believe that an answer to questions such as the ones raised here can only be found on the basis of a large body of empirical research, which we currently do not have at our disposal.

  • 3  See Caban, Kriegel & Pfänder 2007 for details.

2We are well aware of the problems that generalized claims about differences between ‘the’ French and ‘the’ Germans run into. Both the French and (perhaps even more) the German speech community are far from monolithic and show considerable internal variation. There is no reason to believe that this should hold for stylistics less than for grammar. In fact, our aim is considerably less ambitious and restricted to the speech of a certain group of speakers in a well-circumscribed situation. The data we will compare are all taken from interviews with average French and German speakers of a middle to advanced age (around 60), who have no academic training and who are not used to speaking in public. In the German case, they were recorded in the framework of a project on intonational differences in the large urban vernaculars of Germany, in the French case within a bi-national project on migrant oral history.3

  • 4  The two cases cited from already published corpora (Ludwig 1988 and Blanche-Benveniste 1990) are c (...)

3The data were informal in the sense that there was no questionnaire and the interviewers asked open questions, participants engaged in exchanges other than question-answer sequences, and topics were pursued as long as they were of interest to both participants. They were formal in the sense that the interviewer and the interviewee knew each other only superficially or not at all, the interviewer was responsible for asking questions and the interviewee provided information about his or her life (life stories played an important part), and the interviewees were considerably older than the interviewers (the latter being around 30).4 With regard to the three approaches to oral styles outlined above, it is obvious that the interview situation makes a certain amount of dependence of these speakers’ styles on the official styles of the respective culture more likely than other speech activities would.

  • 5  Cf. Auer 1999, ch.7 on the notion of performance used here. It should not be confounded with perfo (...)

4Speech exchange systems such as conversational interviews (as one could call this hybrid genre) invite longer contributions by the interviewee in which s/he tells stories about life in the old times in general, and his or her life history in particular. These stories may have been told before, and they can be expected to exhibit features of a linguistic performance (more than, say, conversation over dinner among a couple).5 By this we mean that through an auto-reflexive monitoring of the way in which the stories are told, the formal delivery of the information becomes subject to evaluation, beyond the contents of what was being said. Performances invite an audience’s attention to their form, and they invite evaluating it as ‘good’ or ‘bad’, according to form-related criteria. In our case, it is the interviewer (and perhaps even the putative listeners of the tape-recording) that assume the role of the audience evaluating the performance.

  • 6  For further details on basic operations of spoken syntax, see Auer (2000, 2005, 2006).

5Style is, of course, an elusive concept, and a general description of stylistic differences as being more or less performance oriented is of little use as long as it cannot be shown that there are stable linguistic parameters that constitute this style. For this paper, we have singled out one specific phenomenon, which is tied to one of the basic operations of spoken syntax, i.e. retraction.6 In a retraction, a paradigmatic slot in an emergent syntactic unit is used twice, i.e. in formulating a next verbal component, the speaker re-uses a syntactic position which has already been filled by some other element before, and puts another element in this position. Retractions thus are characterised by a peculiar combination of forward-and backward looking formal features: something new is formulated by re-using something old. Syntactically speaking, retraction is the basis of repair, but not all retractions do repair work, let alone correct a previous item. Retraction is also the basis of list construction, and it is used for numerous other, non-repair functions, as we shall see below. While retractions are so basic to oral syntax that they occur in abundance in all our data, German or French, we will restrict our attention here even further to multiple retractions, i.e. to cases in which the same paradigmatic slot is used at least three times, or re-used at least twice. The quantity and quality of these multiple retractions is, so we claim, an important stylistic feature which differentiates German and French performance styles.

1. The structure of multiple retractions

  • 7  See appendix for a list of transcription conventions.

6Here is an example for a double retraction:7

Ex. (1): cil_pfaender_marco 14, 20-25

((The interviewer—S—just asked Madame Marco—MM—about her and her husband’s relationship to the sea))


mais nous sommes des gens, (-)

but we are people

qui aimons la mer; (0.5)

who love the sea

pour le paysage qu’elle nous offre,

for the landscape it offers us




pour tout ce qu’elle nous apporte en bruit:,

for everything it offers us in terms of sounds

en en odeur, (o.5) euh:

of of smells uh

pour s’y baigner, (1.0)

for taking a bath in

mais:: on n’aime pas aller sur les bat(h)eaux.

but we don’t like to go by boat

7It is easy to see that through the repetition of the highlighted pour-phrase, the speaker creates some kind of orderliness in her contribution. How this is done can be better seen if we replace the conversation analytic transcript by a more structural, grille-inspired one:

8The speaker produces a complete syntactic structure (nous sommes des gens qui aimons la mer) to which she attaches a continuation in the form of a causal prepositional phrase introduced by pour (pour le paysage) and expanded by a relative clause (qu’elle nous offre). Again, a syntactic completion point is reached. But instead of finishing her contribution at this point, the speaker re-uses the paradigmatic slot created by the prepositional phrase and formulates another causal prepositional phrase introduced by the same preposition and structured very similarly (the noun following the preposition is, in this case, a quantifying pronoun—tout—which is once more expanded by a relative clause: ce qu’elle nous apporte en bruit), i.e. she retracts to the position of the preposition and “starts again” at this point. Another possible syntactic completion is reached now. After a subordinate retraction internal to the main retraction (en bruit en odeur), the speaker retracts a second time to the paradigmatic slot established by the first and second prepositional phrase and re-uses it a third time. This time, she puts into this slot the same preposition pour but the preposition is now followed by an infinitival clause. (More about this shift from Prep + NP to Prep + S later.)

9Example (1) presents a prototypical version of a double retraction. However, there are various variants which can be described as follows:


10The double retraction may operate with or without an anchor. An anchor is the word to which the speaker retracts and on which s/he restarts her contribution. In ex. (1), the anchor is the preposition pour. The anchor marks the paradigmatic slot in which the retraction is produced. However, retractions can occur without an anchor. Take, for instance, the following German example:

Ex. (2): (talk about two dogs)


aber die lieben sich heiß und INnig.

but they love each other so much

[<<più f>geben sich KÜSSchen, >

            kiss each other


[die SCHMUsen;

they cuddle


[SCHMUsen und‘

cuddle and


[is=a HERRlich; (0. 75)

that’s great

11Simplifying again to make the paradigmatic structure of MUB’s contribution better visible, we can represent the example as follows:













heiß und innig

so much



12In this case, the speaker has also reached a syntactic completion point after the first line and retracts to a syntactic position within this clause. The re-used slot begins with the finite verb in the left brace of the sentence, i.e. lieben (sich). The speaker replaced the reflexive verb by another verb—geben—followed by the homophonous reciprocal pronoun and an object phrase. Again, a possible syntactic completion is reached, but the speaker retracts a second time to the same slot, in which he inserts the verb schmusen which does not require a complement but leads on to a third possible completion point without further elements. In this example, the double retraction does not rely on an anchor, since the first word in the paradigmatic slot is not the same.


13The retraction may occur before or after a syntactic completion point. In the two examples discussed so far, the retraction sets in after syntactic completion, i.e. the retraction occurs at a point in which no syntactic projection is in play. However, retractions may also occur at an earlier point in the emergence of a syntactic construction, before the rheme has been produced. Consider the following example:

14Ex. (3): (Ludwig 1988, 38, 6-10)


quand on voit des pays

when you see countries

j’sais pas combien (ils sont)

I don’t know how many they are

quand on voit des pays comme la Russie que les gens/

when you see countries like Russia where people

tiens j’lisais ton œ le truc que tu m’as passé là

by the way I read your uh the thing you gave me

quand on voit ces pays-là

when you see those countries

que les gens … acceptent des situations inacceptables …

where people accept unacceptable situations

j’peux pas comprendre qui qu’i’s osent parler

I can’t understand that that they dare talk

15Or in a more schematic representation:

quand on voit des pays

when you see countries

(j’sais pas combien ils sont)

I don’t know how many there are

quand on voit des pays comme la Russie…

when you see countries like Russia…

(tiens je lisais ton truc...)

by the way I read your thing…

quand on voit ces pays

when you see these countries

j’peux pas comprendre...

I can’t understand

16In this case, the first line of the tripartite structure we are interested in (quand on voit des pays) is not a self-contained syntactic unit, but first requires (projects) a continuation which qualifies the indefinite object phrase, and then a main clause on which the temporal clause depends. Instead of providing these projected syntactic continuations, the speaker self-interrupts and inserts a parenthetical meta-comment first (j’sais pas combien ils sont); after that she retracts for the first time to the beginning of the syntactic structure, repeats her first words and indeed provides the beginning of a semantic qualification of the indefinite noun phrase, i.e. an example (quand on voit des pays comme la Russie…). The emerging structure, however, is interrupted once more, another parenthesis is inserted (tiens je lisais ton truc…), and the speaker retracts a second time to the beginning of her unfinished syntactic project, repeating the subordinated temporal clause in a slightly altered way (quand on voit ces pays-là…). Only now does she finally proceed to the main clause in order to bring her turn to a syntactically well-formed completion. In this case, the emerging syntactic structure grows step by step (incrementally), with each retraction. A syntactic closure is only reached after the last retraction. The example follows Bilger & Blanche-Benveniste’s observation (1999, § 1. 3.) that repetitions produce a kind of “rhythm” in oral French:

L’analyse de productions orales suivies, non préparées à l’avance, nous a permis d’observer que les locuteurs semblent utiliser une sorte de rhétorique fondamentale qui s’appuie sur des figures rudimentaires comme la répétition, la symétrie et la rupture. Les locuteurs répètent plusieurs fois de suite un même type syntaxique et rompent cette répétition en intercalant un type syntaxique différent, comme dans une disposition métrique.


17It is possible that the speaker retracts to a certain slot in his or her utterance just in order to repeat the (yet unfinished) utterance fragment produced so far. For instance, he or she may repeat an anchor element various times, a form of retraction usually functioning as a hesitation marker. The incremental type may combine with this verbatim repetition, as in the following example:

18Ex. (4) : MU05 2172-78


weil ICH hab a: n beKANNten

because I have uhm a friend

wo also der HUND (0.53)

of-which the dog

die die (.) die [PIA.

the the the Pia ((=the dog’s name))




die FREUNdin vom OS[kar ((etc.))

the girlfriend of Oskar ((the speaker’s dog))

19Or again schematically:

wo also der hund

of-which PART the dog











Freundin vom Oskar

girl-friend of Oscar’s

20The speaker introduces a new referent (the dog of a friend) and may be searching for the dog’s name. Whatever the reason, he displays hesitancy by repeating the definite article once before he retracts to this anchor again to incrementally add the name “Pia”. Finally, the now complete syntactic (sub-)structure (noun phrase) is followed by another retraction to the anchor, and another noun phrase which replaces the dog’s name by a descriptive noun phrase.

21We finish this short discussion of the structural variants of multiple retractions by a note on list construction, a topic that has found considerably interest in the literature on conversational syntax (cf. Müller 1991, Selting 2003, Jefferson 1990). In a wide and quite loose sense, double retractions can be equated with lists. We prefer a more restricted usage though according to which lists are a special case of multiple retractions. We will only speak of a list if the retraction involves constituents which stand in a strict relationship of syntactic equivalence (and could therefore be conjoined by the conjunction und/et), and which, in addition, are members of the same semantic category. Thus, the double retractions discussed so far would not qualify as lists, since they are not always of the same syntactic category (as in ex. (1) and the incremental cases), and in all cases, the=listed= elements would not be part of the same semantic category. A list, however, is achieved in the following example by a double retraction:

Ex (5): (Interview with former railway employee)


.h aber BILder und äh

but photos and uhm

.h eisenbahnsouveNIRS: äh

railway souvenirs uhm

und dergleichen

and things of that type

sammle ich also HEUte noch

I still collect today




22The slot of the object noun phrase is re-used twice here, and the emerging list contains two elements from the same semantic category (that of collectibles) and a list-filler.

2. The structure and functions of retractions in German

23After this short overview of the structure of multiple retractions, we now turn to differences between the German and the French data. The German interview data contain many examples of retractions constituting lists (such as in example 5), and of retractions marking hesitations (such as in example 4). We also find examples in which the retraction serves to integrate parenthetical comments paralleling the French example 3:

24Ex (6): (topic is I’s new house)


.h wir ham jetza SCHÖne:’ (.) SCHÖne::-

we now have beautiful beautiful

müssen=s ma kommen ANschaun;

you have to come and see

.h also schöne: (.) TÜren drin:,

well beautiful doors in the house

also wenn=se mal ZEIT ham am vormittag (…)

so if you have time one morning (…)

wir ham jetza

We have now





(müssens mal kommen anschaun)

you have to come and see





TÜRN drin

doors inside

25The retraction sets in before a syntactic completion point, and before the rheme of the utterance has been produced. The first retraction on the adjective schöne may be a forward marker for the inserted parenthetical remark which is produced between the first and the second retraction. The second retraction resumes the superordinated sentence after the parenthetical comment is over.

26Examples of double retractions such as in (1) are rare, however. One of the few German examples which at first sight appear to be similar to the structure of the French example (1) is the following:

Ex. (07):


des WIChtigste bei eire haustür is (0.78)

the most important thing for your front door is

dass sie auch äh a tür is die ma net AUShebeln kann.

that it is uhm a door which you can’t lift out of the frame

die also mehrere (0. 40) [äh (0.26) verBINdungen (    )

which has more than one I mean uhm links (    )





und dass der stock au richtich beFEStigt [is

and that the door frame is fastened properly




richtich stark verSCHRAUBT [is

is screwed in very tightly



27The example shows an amount of “fuzziness” which is typical of German multiple retractions whose structure often is not easy to see. However, the grid representation makes it clear that what appears to be a case of double retraction at closer inspection turns out to be a collection of binary structures (simple retractions) which are nested into each other.

28In structural terms, it is of some interest that German multiple retractions do not necessarily imply the use of an anchor. Consider the following list construction:

Ex. (8): (interview with a former railway employee)


so ham se also mit ALlem zu tun=so

therefore you have something to do with everything

mit persoNAL,

with sta

mit EINstellung,

with recruitment

und entLASsungen,

and redundancies




[wie] also mit der konTROLle .h äh: des ZUGverkehrs,

as with the regulation uhm of the train circulation

mit aufstellen von FAHRplänen,

with the setting up of time tables

un: d dergleichen;

and so on

und überWAChung des verkEhrs

and the supervision of the traffic

29The interviewee, a former railway employee, uses various retractions. In the first part of his contribution, he begins with a generalised description of his duties by stating that one had to do with everything at the railways. After that, the noun phrase mit allem is further specified by a list, the first two elements of which copy the preposition mit (=with=) which had been used in the introductory generalised description. The last list element (Entlassungen), however, is added without a retraction to the anchor preposition:

30In the second part of the turn, the speaker switches to another list construction, again based on the already established pattern of a preposition (mit) & (complex) noun phrase. In semantic terms, it is a continuation of the previous list, with more elements now added from the domain “railway”:

31As before, the components of the list do not follow exactly the same pattern. While the first and second phrase start with the preposition mit, the list filler in the third position of the list has no anchor, and neither does the repair-like addition of the fourth list item (supervision of traffic). The example shows, then, on the one hand, that list construction is a particularly frequent function of retractions in German, but also, on the other hand, that retractions in German are highly variable in their structure, and are often not, strictly speaking, a form of parallelism. The non-use of an anchor in some of the cases contributes to this picture.

32In addition to lists, German speakers use multiple retractions in the context of simultaneous speech. As is well-known from previous research, repetitions of items are a frequently employed technique to ward off interruptions and hold the floor in competitive environments. An example is the following:

Ex. (9): (conversation is again about I’s new house and the security doors that have been fitted in the front)


und da[nn äh

and then uhm


           [aber dann brAucht=s a=

           yes then you of course need

=von hinten her brAuch[t=s ja A

from the backside you of course need


                                       [i glaub da bricht ma dann bei uns

                                       I think somebody would have an easier time

leichter dann durchs <<laughing> MAUer[werk[durch>

breaking into our house through the wall


                                                                  [jaja [NA aber

                                                                  yes yes no but

[von HINten her (0.2) braucht=s a geNAUso äh [sichere tÜrn.

from the backside you of course need equally secure doors


[<<pp>>als (                )>                                         [((clears throat))

ja KLAR.

yes of course

33The retraction starts before the rhematic component (that equally safe doors are needed in the back, in addition to the front door). In the first retraction, an element is inserted in front of the verb (von hinten her), in the second, the construction is brought to an end by adding the rhematic component genauso sichere Türen. Withholding the rhematic part of the utterance, and thus arguably the most central part of the speaker’s statement, is obviously linked to the turn-taking problems surrounding this utterance. While speaker MUA is about to formulate his slightly critical remark that front door security is useless unless the backdoors are secured as well, the owner of the house wants to bring home her hyperbolic praise of the front door (which is so secure that burglars would have a better chance of breaking in through a stone wall than opening the door by force). This remark is clearly positioned competitively, interrupting MUA’s emergent construction. The competitive nature of this stretch of talk is underlined by MUA’s acknowledgement of I’s joking remark in the first retraction (which is introduced by=yes yes no but=).

34In sum, German double retractions are mostly used for marking hesitancy, for list constructions and for securing speakership in turn-taking turmoils. They are often composed in a way that avoids strict parallelism, for instance by alternating between retractions with and without anchors.

3. Retractions in French

35What are the similarities and what are the differences between French and German double retractions? In French, retractions occur before and after a syntactic closure just as in German. However, we observe differences both in form and in function in our data.

3.1. Symmetry in form

  • 8  See Clinquart, 2000 : 332 : "Les répétitions stylistiques au sein d’énoncés assertifs permettant l (...)

36The French data show a tendency to achieve parallelism in multiple retractions. An important evidence for this tendency is found in the consequent use of an anchor (often a preposition, a determiner or a relative pronoun).8 A first type of retraction in French, which is illustrated in examples 10, 11 and 12, often seems to be related to hesitations or lexical access problems and is as wide-spread in French as it is in German. Let us first consider the following case of a double retraction of the incremental type.

 (10) Poli 2006, 424-426


elle a trouvé du travail à la

she got a job at the

à la gare de :

at the station

à la gare de charles de marseille [= la gare de Saint-Charles]

at the charles station in marseille

elle a trouvé du travail

à la

à la gare de :

à la gare de charles de marseille

37In this case, the retraction occurs before a turn unit closure, i.e. within a sentence before the rheme, which is only reached in the last line. What is striking in this example is the consequent repetition of the prepositional anchor à. This persistent use of an anchor is not restricted to double retractions (ex. 10), but can also be found in multiple retractions; cf. the following extract of an interview with an army officer who talks about the history of Patagonia. The interviewer asked the older man about initiatives aiming to maintain Patagonian identity. The interviewee underlines the importance of one of these initiatives by a certain John Ross:

(11) (Blanche-Benveniste 1990 : 28, 244, 246)

il faut euh
we have to
saluer des des initiatives comme celle de de John Ross
greet the initiatives like the one of John Ross
avec beaucoup de de cœur et avec beaucoup de euh de de
with a lot of commitment and with a lot of
comment dire avec euh
how should I say
avec beaucoup de d’envie de de de vivre des situations
with a lot of eagerness to live situations [that are]
assez rarissimes euh dans notre dans notre époque
quite extraordinary in our epoch
il faut euh saluer des des initiatives comme celle de de John Ross
we have to greet the the initiatives like the one of of John Ross

38In this example, a syntactic closure would be possible after avec beaucoup de cœur, but this possibility is not made use of, as the following expansion shows. The actual turn completion is reached only after several retractions. All the retractions go back to the prepositional anchor de, whose syntactic function changes, however. First, de is meant to introduce the nominal complement of beaucoup, later on, it introduces the infinitival complement of envie de.

39A frequently used anchor is the relativizer, as in the following case:

(12) cil_fra_zürbig_s10 (about a king who has a close relationship to his land)

c’était un roi qui était né au maroc (.)
he was a king who was born in Morocco
qui a été élévé (.) qui a été élévé au maroc (.)
who was brought up in Morocco
qui faisait ses études au maroc (.)
who studied in Morocco
qui a tOUt fait au maroc quoi (.)
who did everything in Morocco
c’est pas comme les rois qui ont fait ses études
he is not like those kings who studied
soit en europe soit euh: (.)
either in Europe or
dans d’autres pays quoi
in other countries you see

40Or in a grille version:

  • 9  The resumptive turn can occur before, after or before and after the list contruction, as in the fo (...)

41In (10) and (11) the retractions occurred within the emergent syntactic project, whereas in this case the retraction is used for “paradigmatic exploration” (Clinquard 2000), i.e. the expansion of a syntactically already completed structure. In expansions of this kind, the series of retractions often ends in a resuming component as in the above example: after having given some facts of the king’s curriculum vitae, the speaker sums things up by adding qui a tOUt fait au Maroc. Here, syntactic structure and prosodic pattern (focus accent on the resumptive adverb tOUt) converge in a conclusive move.9

42Using the relativizer as an anchor is not very common in our German data (although grammatically possible). The difference is on the stylistic level: French shows a clear preference for the formal regularity and parallelism, creating some kind of symmetry or “rythm” (Sabio 2006).

3.2. Rhetorical function

43While French shares the main functions of retractions with German (hesitation, lists, presumably also turn-holding), a close analysis of the data reveals a crucial, arguably exclusive function of retractions in French: multiple retractions are used to create cohesion in complex descriptions or argumentations. What the following examples have in common is that the anchor always has the same signifiant, but not always the same signifié. The cohesion, then, is only formal.

44Reconsider the following extract of example (12):

(13) qui a été élévé au maroc (.) qui faisait ses études au maroc (.)

45In both retractions, the relative pronoun qui is the syntactic subject of the corresponding sentence. But whereas the semantic role of the first qui refers to a patient, the second has to be categorized as the agent of the predicate faire. Thus, the syntactic parallelism hides away the diversity of semantic roles on the sentence level.

46What in French looks like a paradigmatic replacement may therefore in fact slightly change the syntactic pattern (or construction). These paradigmatic=cheats= are not at all rare in our data. Let us reconsider example (1) from this perspective:

  • 10  We are well aware of the fact that these two semantic relations have been interpreted as two sides (...)

47Here, the preposition pour introduces a noun (le paysage) which is then further specified by an appositive relative construction (qu’elle nous offre). Then, pour is re-used as preposition in the first retraction (pour tout ce qu’elle nous apporte…). Syntactically, there is only a slight change since the noun phrase le paysage is replaced by a quantifier which is specified by a relative clause. Semantically, both constructions are quite parallel, too, since both express a causal relationship. In the second retraction, however, pour is no longer used as a preposition, but as a conjunction introducing an infinitival clause. Not only is the syntactic function of pour different between the first and the second retraction; the semantic (or logic) relation expressed in the second retraction is one of finality rather than causality.10

48The same holds for the following example in which the anchor does not have the same syntactic function:

Ex. (14): (on French colonialism in the 17th century)

O :


it was

c’était le protectorat de toute façon (.)

it was in any case the=protectorat=

 (en) fin

I mean


it was

c’était l’époque du colonialisme quoi

it was the time of colonialism you know

tous les pays cherchaient la richesse un peu partout quoi

all countries were looking for wealth almost everywhere you know



et lui il a signé

and he signed


all the same

c’était aux français de prendre le contrôle de tout quoi (.)

the French hat to start ruling over everything you know

49Here, c’était opens an equative copula construction in the beginning (c’était le protectorat) and also in the first retraction, but it expresses deontic modality in the second retraction (c’est à X de faire Y). In spite of the hesitation signals (enfin, bon, bof ), the utterance sounds wellformed, although its semantic and syntactic cohesion is only superficial.

50The same holds—now on a more textual level—for the following case:

 (15) cil_caban_jean_d_2006_787_792 (introducing a friend who is priest)


je parle d’un ami curé que j’avais ?

I’m talking about a friend of mine who is priest (that I had)

qui a prêché à budapest

who preached in Budapest

qui a prêché (.) à: perpignan

who preached in Perpignan

qui a prêché (.)

who preached

qui était en conflit avec son évêque ?

who was in conflict with his bishop




parce que lui était directeur de la (croix de la de tarne)

because he was director of the Croix de la Tarne

et il était en conflitavec monseigneur

and he was in conflict with the priest (bishop)

51In a grille-version:

52The conflict between the priest and the bishop is presented in the same way as all the elements of the priest’s curriculum vitae given just before. Thus, semantically non-parallel elements of the turn are formally parallelized, which clearly goes beyond the function of rhetorical parallelism known as the “iconic signalizing of functional equivalence” (Atayan 2000: 289).

53This type of rhetorical use of multiple retraction can also be observed in the next example, where a concessive construction is hidden under the parallelism:

Ex. (16): cil_caban_2006_jean_d_597-601 (D. is answering the interviewer’s question whether ‘all’ of his brothers were great singers)


vous saviez donc tous chanter

all of you thus were good singers


il y en a qu’un de euh=f (-)

there is one among uh

un de n de=n de nous cinq

among the five of us

mon frère francis qui ne s (-)

my brother Francis who didn’t kn (ow how to sing)

qui qui avait la voix juste mais qui:

who who had the voice just right but who

qui chantonnait quoi (--)

who could not sing loud enough, right?

54In the grille-version we get:

55Here, the interviewer suggests that all the five brothers were talented musicians, who were especially good at singing (the ‘X’ of the cardinal concessive pattern of Barth-Weingarten 2003, Couper-Kuhlen & Thompson 2000). The interviewee then acknowledges this X by giving an X’ (qui avait la voix juste), but contradicts the overall assumption by telling that one of the five was not able to sing accurately nor loud enough (Y: qui chantonnait quoi).

56In sum, double or multiple retractions in French may help to create the impression of a strong cohesion of an emerging complex argumentation even though the semantic or syntactic coherence is only weak.


  • 11  For frequency differences between written French and written German, see Atayan 2006: 314.

57French and German speakers are not so different in their use of double retractions in emergent syntactic projects in semi-formal interactive genres such as interviews. This is not surprising, given the expectation mentioned above that languages are more similar in their spoken form than in their written version since the former lacks the normative pressure exerted on the latter. Both in the French and German data, multiple retractions occur before and after a syntactic completion point and before and after the rheme. Despite this similarity in the overall strategy of emergent syntax, there seem to be major differences in form, frequency and function,11 all of them converging in a general hypothesis of a more stylistic-rhetoric function of retractions in French.

58From a syntactic point of view, there is less variation in the French data which quite consequently use the technique of retracting to and repeating the preposition, determiner or relativizer/conjunction as an anchor. Not only is the same paradigmatic slot re-activated again and again (as in multiple retractions in general), the speaker also re-uses the same word of the retraction over and over again.

59On the functional level, double retractions can be found in both languages in list constructions and as a display of hesitancy. But French speakers use retractions for organising their arguments much more than German speakers do, who on the other hand may use retractions for interactional purposes (such as dealing with simultaneous talk). It seems that complex arguments and even narratives are forced into the format of multiple retractions by the French speakers, as a kind of simplest structure-providing device which creates syntactic parallelisms and rhetorical effect even though the semantics of what is said may not render themselves easily to this kind of structuring.

60In the introduction, we promised not to suggest far-reaching interpretations to our empirical findings. Thus, instead of giving answers, we would like to raise a new question at this point.

61As early as 1961, Harald Weinrich showed that the discourse of the génie de la langue francaise is not based on expressivity or on tradition as other national language discourses are (cf. Eco 1993). Instead, the core of the French discourse is clarté achieved by a well-structured syntactic order on the phrase level. In the same contribution, Weinrich argues that syntactic clarity in French syntax is ‘a myth’. This myth might have given rise to an ethos or even a virtue of well-structured writing. Could it be the case that the same holds for oral language? I.e., that speakers try to create—at least formal—cohesion, even at the cost of cheating syntactically?

  • 12  This argument could help to overcome the old binary (internal vs. external) approach to language c (...)
  • 13  For the term of "francais ordinaire", see Gadet 1997.

62What we can hint at in this context are some differences in the emergence of the “bon usage”, which in France, more than in Germany (Barbour & Stevenson 1998, 145-151) is defined not only in relation to written but also to oral language usage. Vaugelas, often cited as the father of the=bon usage=, not only mentioned the example of the good writers, but also “la manière de parler” of the best men and women at the Royal Court. The specific technique of double retraction may reflect a tradition of oral language performance in France which differs from the German one. In French more than in other European languages, the normative (standard) language developed in close relationship with oral traditions and conversational maxims (cf. Bader 1988), at least in its early phase (Ludwig 1996: 1494). Even grammars written for schools and universities tended to allude to the virtue of well-formed oral conversation in French (see among others Dauzat 1947: 353-4). Since the writings of Rivarol and Voltaire, clarity and ease of understanding have been associated with regularity before variation (Meschonnic 2000). Thus, what Sabio (2006) calls “ce rythme particulier” in French, might in fact be a performance style. In part this rhythm might by a reflection of the often postulated “facilité de construction qui se prêterait […] à l’oreil” (Dictionnaire Général, cit. apud Saint-Gérand 2000 : 43).12 The French retraction style then might be part of what Blanche-Benveniste & Bilger (2000) call the rhétorique fondamentale des locuteurs non-professionnels in ordinary13 everyday French. This, however, is a claim that has to be substantiated by future research.

Haut de page


Appendix: Transcription conventions (following GAT, cf. Selting et al. 1998)

Sequential structure

[       ]

simultaneous talk







brief pause (0.1 sec)


pause of one second

segmental transcription

:, ::


glottal cut-off


laughter in talk

haha he he



akZENT, akzENT

primary (nuclear), secondary accent

< >

prosodic description of the passage in < >

, ?

rising intonation (boundary tone)

; .

falling intonation (boundary tone)


((          ))

para-/extralinguistic activities

(            )


Haut de page


1  See among others: Schmitt (2001) for German and Rühl (2002) for French.

2  Cf. recently Geuen, Kimminich, Rappe & Pfänder 2007.

3  See Caban, Kriegel & Pfänder 2007 for details.

4  The two cases cited from already published corpora (Ludwig 1988 and Blanche-Benveniste 1990) are carefully chosen for representing the same setting.

5  Cf. Auer 1999, ch.7 on the notion of performance used here. It should not be confounded with performativity in the sense of Austin or Butler.

6  For further details on basic operations of spoken syntax, see Auer (2000, 2005, 2006).

7  See appendix for a list of transcription conventions.

8  See Clinquart, 2000 : 332 : "Les répétitions stylistiques au sein d’énoncés assertifs permettant l’exploration paradigmatique peuvent concerner différentes catégories grammaticales : principalement des marqueurs grammaticaux […]"

9  The resumptive turn can occur before, after or before and after the list contruction, as in the following case (the interviewee is underlining her very strong feelings about the city of Marseille): j’ai tout ici: (--) j’ai le solEIl (.) j’ai la mEr (-) j’ai la mOntAgne (.) j’ai tOU:t (cil_fra_röder_05_08_18a_farida). As in example (12), the word accent on tOUt is quite strong.

10  We are well aware of the fact that these two semantic relations have been interpreted as two sides of the same coin. Cf. Raible (1992) on both uses of pour.

11  For frequency differences between written French and written German, see Atayan 2006: 314.

12  This argument could help to overcome the old binary (internal vs. external) approach to language change, as proposed very convincingly in Oesterreicher 2007.

13  For the term of "francais ordinaire", see Gadet 1997.

Haut de page

Pour citer cet article

Référence papier

Auer P. & Pfänder, S. (2008). Multiple retractions in spoken French and spoken German. A contrastive study in oral performance styles. Cahiers de praxématique 48. Montpellier : Pulm. 57-84.

Référence électronique

Peter Auer et Stefan Pfänder, « Multiple retractions in spoken French and spoken German. A contrastive study in oral performance styles »Cahiers de praxématique [En ligne], 48 | 2007, document 2, mis en ligne le 01 janvier 2010, consulté le 26 novembre 2020. URL :; DOI:

Haut de page


Peter Auer

Deutsches Seminar 1/Romanisches Seminar
University of Freiburg, Germany

Stefan Pfänder

Deutsches Seminar 1/Romanisches Seminar
University of Freiburg, Germany

Haut de page

Droits d’auteur

Tous droits réservés

Haut de page
  • Logo Praxiling
  • Logo Presses universitaires de la Méditerranée
  • Logo ERIH PLUS
  • OpenEdition Journals
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search