From speaker to subject. The obligatorization of the Old French subject pronouns.

35
. Ulrich Detges (Tübingen) From speaker to subject. The obligatorization of the subject pronouns in Old French in a pragmatic perspective. The obligatory use of the subject pronouns is a typological feature which distinguishes Modern French not only from most of the other Romance languages (Bossong 1979: 66), but also from its ancestor Old French. To illustrate this point, a short extract from the Quatres Livres des Rois ((1a)) is contrasted with its translation into Modern French (see (1b)). (1) a. Old French (OF): Nadab i , le fiz Jeroboam, regnad sur Israel el secund an Asá, le rei de Judá, é dous ans ø regnád i , Málement ø uverad i vers nostre Seignur é ø sewíd i les males tráces sun pere j é le pechied par unt il j fist pecchier cez de Israel. (Quatres Livres des Rois, p. 153, XV, 25–6, cf. Herman 1954 : 370) b. Modern French (MF): Nadab i le fils de Jeroboam régna sur Israel en second roi avec Asa, le roi de Juda, et il i régna deux ans. Il i agit mal envers notre Seigneur et ø suivit i les mauvaises traces de son père j et le péché par lequel celui-ci j avait induit dans le péché ceux d’Israel. c. Modern English translation Nadab i, the son of Jeroboam reigned over Israel as second king with Asa, king of Judah, and he i reigned for two years. He i behaved badly against our Lord and ø followed i the bad traces of his father j and the sin through which the latter j had induced to sin those of Israel. The comparison between (1a) and (1b) brings to light considerable differences concerning the textual function of the subject pronouns as well as their respective semantic value. The main topic of the narrative is the biblical character of king NADAB, which is passed on from proposition to proposition. In such a case of topic continuity, Modern French (1b) normally 1 makes use of the unstressed subject pronoun il. 1 A frequent exception to this rule is the case represented by et ø suivit i in (1b), where we find a coordination of two predicates with co-referential subjects within the same sentence. Histori- cally, the omission of the second subject in this specific case

Transcript of From speaker to subject. The obligatorization of the Old French subject pronouns.

.

Ulrich Detges (Tübingen)

From speaker to subject. The obligatorization of the subject pronouns in Old French in a pragmatic perspective.

The obligatory use of the subject pronouns is a typologicalfeature which distinguishes Modern French not only from mostof the other Romance languages (Bossong 1979: 66), but alsofrom its ancestor Old French. To illustrate this point, ashort extract from the Quatres Livres des Rois ((1a)) is contrastedwith its translation into Modern French (see (1b)). (1) a. Old French (OF):

Nadabi, le fiz Jeroboam, regnad sur Israel el secund an Asá,le rei de Judá, é dous ans ø regnádi, Málement ø uveradi versnostre Seignur é ø sewídi les males tráces sun perej é lepechied par unt ilj fist pecchier cez de Israel. (Quatres Livresdes Rois, p. 153, XV, 25–6, cf. Herman 1954 : 370)

b. Modern French (MF): Nadabi le fils de Jeroboam régna sur Israel en second roiavec Asa, le roi de Juda, et ili régna deux ans. Ili agit malenvers notre Seigneur et ø suiviti les mauvaises traces deson pèrej et le péché par lequel celui-cij avait induit dans lepéché ceux d’Israel.

c. Modern English translationNadabi, the son of Jeroboam reigned over Israel as secondking with Asa, king of Judah, and hei reigned for two years.Hei behaved badly against our Lord and ø followedi the badtraces of his fatherj and the sin through which the latterj hadinduced to sin those of Israel.

The comparison between (1a) and (1b) brings to lightconsiderable differences concerning the textual function ofthe subject pronouns as well as their respective semanticvalue. The main topic of the narrative is the biblicalcharacter of king NADAB, which is passed on from propositionto proposition. In such a case of topic continuity, Modern French(1b) normally1 makes use of the unstressed subject pronoun il.1 A frequent exception to this rule is the case represented by

et ø suiviti in (1b), where we find a coordination of two predicateswith co-referential subjects within the same sentence. Histori-cally, the omission of the second subject in this specific case

2 Ulrich Detges

At some point, however, the topic NADAB is replaced by NADAB’SFATHER, introduced in (1b) as a rheme by son père, and taken upin the following as topic by the pronoun celui. In ModernFrench, celui indicates that the referent in question is notidentical with the topic of the preceding proposition – celuimarks a referential contrast. If we compare the Modern French(MF) translation (1b) with the original Old French (OF)version (1a), we find that whenever MF uses il, there is ø inthe OF original. The sole occurrence of il in OF correspondsto celui in the modern translation. Even though the generalconditions for the use of celui in MF are very different fromthe ones which govern the use of il in OF, the comparison of(1a) and (1b) brings to light an essential fact: in OF, theSP do not express, like in MF, that their referent is identicalwith some other referent already introduced in the universeof discourse. Rather, they are used to mark a referential contrast(Buridant 2000 : 429, Herman 1954 : 370). Correspondingly,they are used much rarer in OF than in MF.The diachronic process which transformed the SP of OF ( see

(1a)) into their MF successors (see (1b) comprised twocomplementary aspects. On the one hand, it consisted in aconsiderable frequency increase of the SP. On the other hand,it entailed a loss of their original contrastive effect. Bothaspects – increase in frequency as well as semantic weakening– justify to consider this process, which started around theend of the 12th century, as a case of grammaticalization inthe sense of Lehmann (1995: 9).2 In what follows, both terms,obligatorization and grammaticalization, will be employedsynonymously.

1.The subject pronouns in Old French – somegrammatical particularities

The contrastive effect of the OF subject pronouns had alreadyweakened in certain contexts long before the end of the 12th

century. This holds especially for impersonal constructions,where SP are attested since the 11th century (Jensen 1990:137).3 Another particularity of OF from the first textsonwards, is the extreme frequency of SP in subordinate

is a relic of the Old French technique of marking topiccontinuity.

2 More precisely, it is a case of secondary grammati-calization (seeDetges 2003).

L’obligatorisation des pronoms sujets 3

clauses, especially in preverbal position (Franzén 1939,Kattinger 1970, Buridant 2000: 432). The reason for this wasthe canonical OF sentence structure, which was T(opic) V(erb) X,with the position of the topic being open to all sorts ofconstituents, under the condition, however, that these had tobe tonic elements. This structure was the same for both mainclauses and subordinate clauses. Whereas in main clauses therealization of the SP was rather rare, they were almostobligatory in subordinate clauses if no other tonic elementpreceded the verb (see (2a, b)).4 In the opposite casehowever, they were normally omitted ((3)). (2) a. Ainz priet Deu quet il le lur parduinst (Alex 269)

‘But pray :3s:Pres God that he it to:them forgive:3s:Subj:Pres’ ‘But he prays to God that he may forgive it to them.’

b. Siet el cheval qu’il claimet Barbamusche (ChRol 1491) ‘Sit:3s:Pres on:the horse which he call:3s:Pres Barbamusche.’ ‘He sits on a horse which he calls Barbamusche.’

(3) Oliver sent que a mort est ferut. (ChRol 1951)Olivier sense:3s:Pres that to death is:3s:Pres struck‘Olivier senses that he is struck to death.’

This tendency has traditionally been explained by prosodicconsiderations: according to this view, the SP were necessaryin subordinate clauses in order to prevent conjunctions andrelative pronouns, which were always unstressed, fromappearing in the preverbal position, reserved for tonicelements only (Wartburg 31970: 66).5 However, this hypothesisis inconsistent with the fact that precisely in the initialposition of subordinate clauses, the SP had already turnedinto unstressed elements (Franzén 1939: 34), probably as aconsequence of their frequent realization in this position. Afunctional explanation has been proposed by Thun (1989: 207-10, see also Buridant 2000: 347). As we have seen in (1a, b),an essential function of the OF SP consisted in taking upnon-topical protagonists and to establish them as new topics.3 For the diachronic process lead to the emergence of

impersonal SP see Koch (1994).4 Object pronouns, e.g. as le and lur in (2a), are generally

considered as atonic elements.5 In the traditional view, this rule is only a special case of

a more general syntactic regularity. See Thurneysen (1892: 303-4).

4 Ulrich Detges

Following Thun, this function, which regards the paratacticconcatenation of different sentences within the text, had beentransposed onto the hypotactic organization within the complexsentence – with the difference that in this latter context, itbecame a grammatical procedure and hence was largelymechanized. This would explain why, in OF, we do not onlyfind SP in cases of topic discontinuity as in (2a) but alsoin constellations with clear topic continuity such as (2b). However, both hypotheses fail to explain why the SP started

to become increasingly frequent in main clauses around theend of the 12th century. As we have seen, the high frequencyof the SP in subordinate clauses was linked to a technique ofsubordination. Therefore, it is quite unlikely that their usein this particular context could have triggered a frequencyincrease in main clauses. Recall that in oral communication,main clauses are far more frequent than subordinate clauses.Consequently, major grammatical changes are normallytransmitted from main clauses to subordinate clauses and notvice versa (Givón 1976: 170-1).6 What is more, the highfrequency of the SP in subordinate clauses followed aregularity which had remained stable since the 11th century(Franzén 1939: 29). Therefore it is reasonable to assume thattheir sudden frequency increase in main clauses from the endof the 12th century onwards had no direct connection with thefacts described so far.

2. The obligatorization of subject pronouns – apolygenetic process

There are two established explanations for theobligatorization of the SP in late OF. The first one arguesthat this change was due to the influence of a Germanicsuperstrate or adstrate (Kuen 1970 [1957], Hilty 1975). Thisposition can claim a striking coincidence as an argument inits favour: the Romance varieties which have obligatory SPeither developed in regions invaded by Frankish or Lombardpopulations – as in the case of French and of the dialects ofNorthern Italy7 – or in territories where the Romancepopulation lived in close contact with Germanic neighbours –

6 This regularity does, of course, not apply to thegrammaticalization of subordination markers (cp. Raible 1996:76).

L’obligatorisation des pronoms sujets 5

e.g. in certain Rhaeto-Romanic areas (Linder 1987).8 However,a hypothesis based on a specific superstrate influence wouldhave great difficulties to explain why the SP are currentlybecoming obligatory in contemporary Brazilian Portuguese orin Puerto Rican Spanish. The second position proposes a functional explanation.

According to this view, the obligatorization of the subjectpronouns was an answer to the progressive loss of the verbalinflectional endings in Middle French (see Foulet 1935/36, v.Wartburg 31970: 72-6, Vance 1989: 296). However, verbalinflections have stayed intact in a good number of NorthernItalian as well as in Rhaeto-Romanic varieties withobligatory SP (Mair 1992: 355, Wanner 1993). In these cases,the hypothesis of a “morphological distress responsible forthe obligatorization of the SP can definitively be ruledout.9 Discussing the relative chronology of the two changes

7 This influence may perhaps help to explain why the medievaldialects of Northern Italy show the same prominent tendency asOld French to realize preverbal SP in subordinate clauses (cp.Vanelli 1987: 182, 184).

8 Likewise, Silva-Corvalán (1993: 36-8) observes that incertain Spanish varieties spoken in the USA the conditions ofuse of the SP change under the influence of English.

9 It must said however that this debate is far from beingclosed. If we believe some of the recent synchronic studies ondifferent varieties of American Spanish, there exists nostatistical significant correlation between the realisation ofthe SP and the cases of ambiguity in the verbal morphology ofSpanish (Meyer-Hermann 1996: 294-6, Bentivoglio 1987: 45, Barre-nechea, Alonso 1977: 346-9). Others however – among them Silva-Corvalán (1993) and Hochberg (1986) – claim the existence ofsuch a connection. In her works on the evolution of theBrazilian Portuguese SP, Duarte (2000: 19, 1996: 109) explainstheir progressive obligatorization by certain morpho-syntacticchanges which, according to her, have contributed to a weakeningof the verbal inflection. This explanation, however, is incontradiction with the data given in Duarte (1995) – shown abovein (4) (see also note 11)–, since the inflectional ending of thefirst person has stayed intact in the new system of BrazilianPortuguese. What is more, Barme (2000: 192) shows that, neitherin peninsular Portuguese nor in Brazilian Portuguese, themorphological ambiguity of the verbal forms play a role in therealization of the SP. In the same research, Barme comes to theconclusion that in cases in which neither the verbal endings northe context allow to identify the referent of the subject areextremely rare (2,4% in his corpus of Brazilian Portuguese, 2,6%

6 Ulrich Detges

in French, Harris (1978: 113) points out that the loss of theverbal inflectional endings only occurred after theobligatorization of the SP. If this view is correct, then theobligatorization of the SP was not a consequence of the lossof verbal inflectional but rather a (necessary butinsufficient) condition of this change.The obligatorization of the SP is a polygenetic phenomenon,

that is, a type of process which has affected many languagesin different parts of the world at different epochs. Giventhat each of these languages developed under historicallydifferent conditions and that each of them possessed, atevery moment of its existence, an idiosyncratic grammaticalsystem of its own,10 it follows that polygenetic change cannotbe explained either by grammatical necessity or by specifichistorical circumstances. In this paper, I am going topropose an alternative hypothesis according to which theobligatorization of the subject pronouns is the result ofcertain rhetorical strategies. These strategies, which arepart of the universal features of spoken language, aremotivated by the speakers’ desire to assure themselvesimmediate communicative advantages. Put more briefly, theobligatorization of the SP in OF as well as in otherlanguages is triggered by strong pragmatic factors.Even though polygenetic changes cannot be explained by

grammatical factors, this does not imply that they are notsubject to any kind of regularities. In the obligatorizationof SP, such a regularity concerns the leading role of thefirst person in this process:

in the one of peninsular Portuguese).10 For a systematic description of the considerable differences

between Old French, Modern Spanish, Modern Portuguese andRumanian concerning the use of third person SP, see Thun (1989).

L’obligatorisation des pronoms sujets 7

(4) The proportional distribution of SP by grammatical person (of Oliveira 2000: 39)11

1st pers. 2nd

pers.3rd pers.

BrazilianPortuguese

82% 78% 45%

Italian 51% 44% 27%

The percentages given in (4) for the grammaticalization ofthe SP which is currently taking place in BrazilianPortuguese show that this process is most advanced in thefirst person. When we compare Brazilian Portuguese withstandard Italian, a “conservative” pro-drop language withoutthe slightest indication of an obligatorization of its SP, wefind that the relative proportion of grammatical persons isexactly the same in both languages.12 In fact, the use of theSP in progressive obligatorization like in BrazilianPortuguese differs only gradually from what we find in pro-drop languages. The percentages given in (4) raise animportant question: what purposes do the SP - andparticularly first person subject pronouns - serve in pro-drop languages? The following section will be dedicated tothis question.

3.Discourse functions of SP in pro-drop languages

For Spanish, a pro-drop language which is as conservative asItalian, Rosengren (1974: 225-30) points out that certainverbs systematically favour the use of the SP. In the firstplace, these are verbs of opinion (e.g. pensar ‘to think’,creer ‘to believe’, see also Bentivoglio 1987: 53). Moreover,it seems that in the context of these verbs, the SP are nolonger exclusively used with their original contrastivevalue.

11 The percentages regarding Brazilian Portuguese are takenfrom Duarte’s (1995) unpublished thesis. For the first and thesecond person, they are confirmed in Duarte (1996: 117, 2000:20), for the third person Duarte (2000: 20) gives a percentageof 48%.

12 This impression is confirmed by Rosengren (1974: 233). Thun(1989: 194) notes a strong reticence in Modern Spanish regardingthe use of the SP for inanimate objects.

8 Ulrich Detges

3.1. Topical self-reference and rhetorical devaluation

With respect to the contrastive value of the SP in peninsularModern Spanish, we have to distinguish two cases, which – aswe will see later on – represent two successive stages of adiachronic change. In the first case, the speaker uses aconstruction of the type yo creo/creo yo ‘I believe, I think’ toindicate that his opinion is in contrast with anotherpossible point of view. In this specific case, illustrated by(5), yo creo/creo yo marks the point of view which, in theopinion of the speaker, is the valid one. Within theconstruction yo creo/creo yo, it is the SP which materiallycarries the contrastive effect. In the case under discussionhere, i.e. when the construction expresses a real contrast,the SP is always emphasized by intonation. The contrastiveweight of the variant creo yo with the SP in invertedposition, is sensibly stronger than the one of yo creo wherethe SP is in anteposition.13

(5) Case 1: Real contrast – “legitimate” use of the SP (CorpusOral 1992)

S2 : Ya no hay pegas. ¡Huy! Después que se les echaba ahía los cerdos, 1

There are no more magpies. Oh! As soon as you gave the pigs [toeat], over there,

ahí en ese llano allí arriba, en el camino... después sellenaba eso de 2

in this plain, up over there, on the way ... well, all this wasfull of

pegas. Muchas, muchas. 3

magpies. Many, many.

S4 : El por qué no se sabe pero... 4

We don´t know why, but...

S2 : Que ¿por qué habrán desaparecido?5

The reasons why they have vanished?

13 What is more, creo yo is very often used as a comment at theend of an utterance: Los problemas que tenéis vosotros, como pareja, vienen defuera, creo yo ¿eh? (Corpus Oral 1992). Thus, it seems that the SP isbest placed at the beginning or at the end of an utterance if itis to deploy its contrastive effect at a maximum.

L’obligatorisation des pronoms sujets 9

S5 : Pues yo creo que son los insecticidas. <silencio> 6Well, I believe that it is due to the insecticides. <silence>

In (5), yo creo (line 6) signals that S5 disagrees with anopinion brought forward by S4 (line 4). At the same time,this example illustrates a secondary effect of thecontrastive use of yo creo/creo yo: the construction announcesthat the utterance introduced by yo creo/creo yo will containnew and hence unexpected information. Normally, utterancescontaining new information are considered as highlyinformative. Even more than normal utterances they correspondto the Gricean maxim of relation (Grice 1975: 46), “berelevant”.Every speaker has an interest in making his contributions

look maximally relevant. Introducing an utterance by aconstruing a contrast with something already said, cantherefore be a rhetorical procedure, designed to make the utteranceappear more informative than it actually is. This is a strongmotivation for speakers of a pro-drop language to use thepronoun of the first person more often than necessary,including in situations where no conflicting point of viewexists. This is the case in the following example.(6) Case 2: Rhetorical contrast – “over-use” of the SP (CorpusOral 1992)

Dialogue in the radio, S1 = female presenter, S2 = femalesinger

S1 : ¿Pero tú no estás segura de si podrías ser buenaactriz o no? 1

But you aren´t sure whether you could be a good actress or not?

S2 : No lo sé. A nivel de teatro, pues mira... hecomprobado que... que bueno, 2

I don’t know. As far as the theatre is concerned, look, Ibelieve, well,

más o menos me defiendo [...].3

I can cope, more or less.

S1 : Bueno... Pero yo creo que los jóvenes eh... artistas, hoyen día tenéis 4

Good... But I believe that the young people... artists, today, youhave to face

un reto muy duro, porque se pide a uno que seapolifacético [...] ¿no? 5

10 Ulrich Detges

a very hard challenge, because you are obliged to show manyfacets [...], no ?

In example (6) the utterance introduced by yo creo (line 4) isnot directed against any possible contrary opinion, since S1formulates a rather trivial point of view (line 4) whichnobody would possibly call into question. Here, yo creo is asimple rhetorical procedure with the function to emphasizethe following utterance. In contexts such as (6), where the SP no longer express a

real functional contrast, they are often pronounced with aneutral intonation, i.e. without any contrastive accent. Thispossibility is excluded, however, for the inverted variantcreo yo, which obligatorily marks a real contrast and whichtherefore is always pronounced with a strong accent on theSP. Due to its communicative utility, yo creo with the SP inanteposition has evolved, in Standard Spanish, into aformulaic expression which is part of the speakers’linguistic routines. This is a direct consequence of theconstruction’s rhetorical “over-use” in cases like (6).Frequent usage of this type has deeply modified the value ofthe construction yo creo, and in particular the value of the SP,which, within this specific construction, has lost itsoriginal contrastive effect. At the same time, it hasconsiderably increased in frequency. Processes of this kindare discussed by Vennemann (1974: 368) under the label of“pragmatic unmarking”. Dahl (1998) refers to them asinstances of “rhetorical devaluation”, a notion whichunderpins the inflationary character of this type ofprocesses. In a restricted context class of the type yo creo, yo pienso, yo

digo etc.,14 in which the speakers topicalize themselves, theSpanish SP show a tendency to appear much more frequent thananywhere else; in the same contexts however, they have partlylost their formerly obligatory contrastive value. The verysame phenomenon can be observed in other pro-drop languages:Mair (1992: 373) reports on a quasi stereotyped use of egocredo ‘I believe’ in Vulgar Latin. Jeo crei ‘I believe’, je cuit ‘Ithink’ are formulas often used in OF, as well as io credo ‘Ibelieve’, io penso ‘I think’ in contemporary modern Italian. Iwould like to call these techniques formulas of self-topicalization.Hence, in a restricted class of contexts –verbs of opinion incontexts of self-topicalization – a frequency increase as14 The formulaic character of these expressions is particularly

obvious in the case of yo digo / digo yo, where the SP is obligatoryif the whole expression signifies ‘I think’ and not ‘I say’.

L’obligatorisation des pronoms sujets 11

well as a loss of the original contrastive effect of the SPcan already be observed in pro-drop languages, even if in allother contexts no change takes place at all. As we shall seein later sections of this paper, this small-scale changefollows the very same internal logic as the process whicheventually leads to a full obligatorization of the SP.

3.2. Subject pronouns and expressions of self-topicalizationas techniques of turn-taking

The techniques of self-topicalization play an important rolein contexts of turn-taking. Normally, in the course of aconversation, no more than one speaker speaks at the sametime. This is so because the allocation of the right to speakis regulated by a small set of elementary mechanisms (seeSacks et al. 1978). Of these, self-selection is the mostimportant one. If the actual speaker hesitates or wishes toend his contribution, this mechanism provides that the rightto speak is passed on to the participant who starts speakingfirst (“First starter goes”, Sacks et. al. 1978: 31). Turn-allocation and turn-taking are often realized under enormoustime pressure. Therefore, speakers need efficient routines tocope with such contexts. Among these, the expressions ofself-topicalization of the type yo creo as well as stressed SPalone (Martín Rojo/Meeuwis 1993: 109, Bentivoglio 1987: 38-9)play an important role (see (6)). This can be seenparticularly in contexts where the right to speak iscontested. In example (5), S1 tries to interrupt S2, so that,at a given moment, S1 and S2 speak simultaneously. By using asimple first person pronoun or a formula of self-topicalization, the speaker marks a contrast with thecontribution of the preceding speaker: I, CONTRARY TO YOU, THINKTHAT.... The communicative utility of this procedure is thesame as described in the last section: a contrast announcesunexpected information; the more the incipient speaker’scontribution appears to contain unexpected information, themore his attempt to take over the right to speak seemsjustified.(7) SP in contexts of turn-taking (Corpus Oral 1992)

S2 : [...] son muy duras 1[...] they are very hard

12 Ulrich Detges

S1 : Pero yo esto que tú <simultáneo> dices 2But I, that what you <simultaneously> you say

S2 : muy dolorosas </simultáneo> 3

Very painful </simultaneously>S1 : Yo es que estoy en contra de toda estrategia, porque

[...]. 4I, it is that I am against all kind of strategic behaviour,

because [...]

Quite often, the use of the SP in situations of turn-takingis a mere rhetorical procedure. If the SP ceases to mark areal contrast with respect to the preceding speaker’scontribution, the above described rhetorical technique turnsinto an efficient routine, used to assure the speaker theright to speak. Contexts of turn-taking are thereforeparticularly suited to drain the original contrastive forcefrom SP.This observations also holds for Modern French. Unlike pro-

drop languages like Italian or Spanish, which only have oneoptional SP, Modern French uses a combination of an(obligatory) clitic pronoun (je) and an optional tonic SP(moi), of which only the latter marks the effects described inthis section. Discussing the use of moi, je in modern French,Honnigfort (1993: 229) makes the following observations:

Moi, je, or je ... moi refers to the speaker who starts to speak, who‘enters’ in a conversation, who personally takes a stand. Thisis particularly evident in constructions with verba dicendi andverbs of opinion, which express a personal statement or adesire: moi je trouve que, moi je veux bien, moi je (ne) sais pas, moi je crois, moi jepense, moi j’aime, moi j’ai l’impression etc. (Translation mine, U.D.).

From these examples, it can be seen that in Modern French,tonic moi is typically used, just like the stressed SP inSpanish or Italian, in contexts of self-topicalization and ofturn-taking. Interestingly, in the constructions mentioned byHonnigfort, moi no longer obligatorily marks a referentialcontrast.

3.3. Expressions of self-topicalization, communicative“weight” and anacolutha

L’obligatorisation des pronoms sujets 13

Until now, we have treated formulaic expressions of the typeyo creo (see (5) and (6)) in the same fashion as simple tonicSP (example (7)). Nevertheless, there is a disparity betweenboth types of constructions, which is above all a differencein communicative “weight”. Formulaic expressions of the typeyo creo are “heavier” procedures than the simple SP. Such asituation of competition between “heavy” and relatively“lightweight” procedures has nothing surprising, since thecontexts of self-topicalization and turn-taking create aconstant need for fresh contrastive constructions, while atthe same time systematically contributing to an erosion oftheir contrastive effect. As a consequence, languages usuallypossess extensive arsenals of “heavy” constructions designedfor self-topicalization. In Modern Spanish, we find, apartfrom yo creo, among others yo para mí ‘I, for myself’ and yo a mí‘I, to me’. Both forms are anacolutha, i.e. syntactically ill-formed constructions characteristic of spoken language (seeKoch & Oesterreicher 1990: 84-6). In addition to such “heavy”constructions, Modern Spanish even possesses “extra-heavy”techniques such as yo para mí creo ‘I for myself think’ in (8d),which is a combination of two “heavy” procedures. (8) “Heavy” and “extra-heavy” constructions of self-

topicalizationa. Yo a mí no se me pasó por la imaginación. (Corpus Oral 1992) I, to me this has never entered my mind.

b. Yo para mí le he dado mucha importancia [....]. (Corpus Oral1992)

I, for myself, I´ve attached a lot of importance to it.

c. Yo para mí creo el... el nombre genérico es ‘ataúd’ [...].(Corpus Oral 1992)

I for myself I believe that the generic term is ‘coffin’ [...].

The particular frequency of anacolutha in contexts of self-topicalization and turn-taking is due to the fact that SP arenot the only linguistic elements to realize the abovedescribed rhetorical operations. In principle, all sorts ofpronominal constructions lend themselves to this purpose,under the condition that they can express a referentialcontrast to the person of the speaker. Thus, besidesexpressions headed by grammatical subjects (e.g. yo creo ‘Ibelieve’), Spanish also possesses formulaic constructions

14 Ulrich Detges

introduced by stressed object pronouns, e.g. a mí me parece… ‘tome, it seems…’.15 As already pointed out, self-selection inturn-taking is often done under considerable time-pressure.Therefore, the rate of “false starts” – and hence thefrequency of anacolutha – is particularly high in this typeof contexts. We shall come back to this observation insection 8.

3.4. Turn-taking and degrees of topicality

The contexts of turn-taking provide a simple explanation ofthe fact that formulas of self-topicalization with the SP inanteposition (yo creo) are much more frequent thanconstructions with inverted subjects (creo yo). With itsoriginally contrastive value, the SP is the element which ofthe entire construction’s communicative impact. In contextsof turn-taking, e.g. in (4) or (5), this impact is exploitedat a maximum if the SP appears in initial position. For thesame reasons, the inverted subject pronoun in creo yo, unlikeits preposed variant, has lost nothing of its contrastiveforce.Turn-taking also explains the quantitative distribution of

subject pronouns by grammatical person, which ischaracteristic of the usage of SP in pro-drop languages (seetable (4)). In order to legitimize his attempt to take overthe right to speak, the speaker presents a referent which islikely to capture the attention of his interlocutor(s). Inany situation of direct face-to-face communication, thespeaker in the first place and his interlocutor in the secondplace are the referents with the highest communicativerelevance. First person subject pronouns therefore areparticularly suited for the rhetorical techniques sketched insection 3.1. and 3.2. Reference to the addressee by means ofa tonic SP is particularly frequent in the contexts where anexplicit transfer of the right to speak takes place (of the15 As is predicted by the theory outlined in section 3.1., the

originally tonic element (a) mí has actually lost its obligatorycontrastive effect in Modern Spanish. In Detges (in print), itis argued that the same discursive mechanisms which caused theobligatorization of the SP in Old French are responsible for theprogressive grammaticalization of the prepositional directobject marking in Iberian-Romance (veo a Juan ‘I see Juan’). Inthese languages the rhetorical techniques described above werepredominantly realized by means of tonic object pronouns andreinforced by the preposition a(d) (e.g. ad mihi).

L’obligatorisation des pronoms sujets 15

type ‘What do YOU think?’, Honnigfort 1993: 240-2). Thesecontexts are the symmetrical corollaries of turn-taking,although they are much less frequent than these. Apart fromspeaker and addressee, every referent can in principle bepresented as a topic sufficiently relevant to guarantee thespeaker the right to speak – the question of exactly whatkind of referent is relevant enough to assure this functionlargely depends on the individual context. In (9) – a dis-cussion between experts dealing with ecological questions –it is La utilización de flúor ‘the use of fluorine’, i.e. aninanimate and abstract referent, which serves this purpose.(9) Turn-taking by means of an abstract referent (Corpus oral 92)

S2 : La utilización de flúor, según nuestros datos, doctorFernando Martín, también... 1

‘The use of fluorine, according to our data [or: facts], Doctor Fernando Martín, also...

<vacilación> en exceso, provoca molestias gástricas[...] 2

<hesitation> in excess, provokes stomach trouble [...]

The more a referent is animated, definite, individualized andhuman, the more it is endowed with potential relevance andthe more it lends itself to this kind of operations. Thedegree of potential relevance of the various referent classesis reflected in the topicality hierarchy of the correspondingclasses of linguistic expressions (see, e.g. Givón 1976: 152,Lazard 1994: 191-204): (10) Topicality hierarchy

Personal pronouns (1st > 2nd > 3rd pers.) > proper nouns >common nouns: animate definite > common nouns: animateindefinite > common nouns: inanimate definite > commonnouns: inanimate indefinite etc.

The higher the position of a given expression is in thishierarchy, the more it is likely to appear as a topic inturn-taking context. The contexts of turn-taking transformthe grades of the abstract topicality hierarchy into realfrequency. As we have seen, the pronouns of the 1st personoccupy the highest position, followed by the pronouns of the2nd person. The inferior position of the pronouns of the 3rd

person can be explained by two factors: not only do they

16 Ulrich Detges

refer to entities endowed with a minor degree of topicality,but moreover, in this function they are found in competitionwith proper nouns and common nouns. Therefore their effectivefrequency in the contexts turn-taking will always be inferiorto the SP of the two other categories. For this reason, tonic3rd person SP are relatively less likely to become non-contrastive elements.It is plausible to assume that if, in a given language, the

functions of (self-) topicalization and turn-taking arerealized by means of originally stressed SP,16 then there is ahigh possibility that in these contexts, the SP will losetheir contrastive value and that they will undergo, at thesame time, a considerable rise in frequency. This hypothesisexplains the high frequency of the SP in pro-drop languagessuch as Italian as well as their distribution by grammaticalperson. As we have seen in (4), the process which eventuallyleads to the full obligatorization of the subject pronouns asin Brazilian Portuguese, seems to be an extension ofsomething which already goes on in pro-drop languages.

Self-topicalization und turn-taking are characteristic ofthe spoken language, to which we have no immediate access inOF. Therefore, our hypothesis cannot be tested for thislanguage. Nevertheless, in written texts of OF we can observephenomena which are closely connected with the rhetoricalmechanisms described in the above sections. These phenomenacould explain why the SP became frequent not only in contextsof self-topicalization and turn-taking, but were generalizedto almost all contexts.

4.“Strong” speech acts and rhetorical devaluation ofthe SP in OF

In the Song of Roland we find a manifest rhetoric of the firstperson. Its main function is to underpin the speaker’sauthorship or his communicative responsibility for the stateof affairs represented in the proposition. This is documentedin (11).(11) Pur ço le juz jo a pendre e a murir. (ChRol 3831)

‘For this, I condemn him to hang and to die.’

16 This reserve is not trivial, since the discursive techniquesin question can. in principle, also be realized by means oftonic object pronouns. See above, note 15.

L’obligatorisation des pronoms sujets 17

Example (11) represents a “strong” declarative speech act.Its illocutionary force depends crucially on the person ofthe speaker, referred to by the SP jo, who guarantees itsvalidity. In the Song of Roland, this type of first personrhetoric is systematically put to profit in all sorts ofspeech acts. We find it in directives (see ex. (12)), indeclaratives (ex. (11) and (13)), and especially incommissive speech acts, where the stressed SP serves to lendillocutionary strength to promises ((14) and (15)), menaces(see (16)) and to spectacular announcements ((17)). Moreover,SP are typically employed in commissive expressions of thetype jo vos durrai ‘I will give you’, which have a formulaiccharacter (see ex. (14)). Another context which favours theuse of a first person SP are verba dicendi (see (18)), where theSP highlights the speakers responsibility for what is said(‘I TELL YOU….)’ . Apart from these relatively clear-cutcontext classes, we find SP in all kinds of “strong” speechacts, which would be difficult to systematize ((19), see alsoBuridant 2000: 436). Of course, SP also appear frequently inindirect speech acts. In (12)-(14), where I refer to the useof SP in direct speech acts, I only quote the respectiveperformative verb.(12) Directive speech acts

a. comander ‘to command’ (ChRol 273, 328, 2659, 2673, 2815,3015)

b. mander ‘to command’ (ChRol 2770), c. defendre ‘to defend’ (ChRol 2438) d. pardoner ‘to pardon’ (ChRol 2007)e. jugier ‘to judge’ (ChRol 3831)

(13) Declarative speech actsJo vos en faz le dun! (ChRol 3059)‘I give it to you as a present !’

(14) Comissive speech acts (I): Promises, performative verbsa. plevir ‘to promis’ (ChRol 968, 1058, 1069, 1072, 1704), b. otrier ‘to give’ (ChRol 3202), c. estre garant ‘to guarantee’ (ChRol 1521)

(15) Comissive speech acts (II): formulas to express a promiseJo vos durrai ‘I will give you’ (ChRol 75, 3205, 3398)

18 Ulrich Detges

(16) Comissive speech acts (III): Menaces (see also ChRol 867)Jo t’en muvrai un si grant contraire Ki durerat a trestut

tun edage. (ChRol 290-1)‘I´ll take revenge on you for this, [revenge] of a kind which will

last all your life.’

(17) Comissive speech acts (IV): Spectacular announcements (see also ChRol 246, 269, 799, 892, 985, 2180, 2684)Jo irai, par vostre dun. (ChRol 246)‘I shall go, with your permission.’

(18) Verba dicendi: ChRol (582, 1716, 1718, 2913, 2919, 1708, 1716,2919)

(19) “Strong” speech acts of all kinds a. Jo ne vous aim nïent. (ChRol 306) ‘I don´t like you at all’

b. Trestuz les altres ne pris jo mie un guant. (ChRol 3189) ‘All the others have not [even] the value of a gauntlet for me.’

c. Fel seie se jol ceil! (ChRol 3757) ‘Call me a scoundrel if I ever conceal it!’

d. Tut seie fel se jo mie l’otrei! (ChRol 3897) ‘Call me a complete scoundrel if I ever give my accord to this !’

In all these examples, the use of the SP follows an identicalmotivation: the stressed first person pronoun indicates thatthe speaker assumes a particularly high degree ofresponsibility for the speech act in question. All otherthings being equal, speech acts marked by SP appear moreserious than speech acts with null-subjects. Therefore,“strong” speech acts with stressed SP will, under normalconditions, be more efficient than speech acts with null-subjects. This is a powerful stimulus to employ the SP moreoften than necessary, including in situations where this isno longer justified by the real relevance of the speech actin question. Such a rhetorical “over-use” will haveconsequences for the SP. On the one hand, their frequencywill increase dramatically, while on the other, theircontrastive value will progressively weaken. The more speech

L’obligatorisation des pronoms sujets 19

acts will be marked as particularly “strong”, the lessimportance will be attached to every single one of thesespeech acts. This mechanism has the power to transform non-obligatory, contrastive SP of a given language into extremelyfrequent elements, which at the same time will be deprived oftheir original contrastive force. By its inherent logic, this process is an instance of the

mechanism of rhetorical devaluation (see above, section 3.1).It repeats, albeit on a much larger scale, the type of changewhose results we have seen in Modern Spanish (see section 3).Unlike change restricted to self-topicalization and turn-taking, however, it will entail a spread of unstressed SP inmuch more general contexts. This hypothesis explains thecurious empirical finding, noted in section 2, that the pro-gressive spread of the SP – as documented in contemporaryBrazilian Portuguese – seems to be a radicalization of theirnormal usage in pro-drop languages such as Italian (see table(4)). According to Haspelmath (1999: 1057), grammaticalization

processes are motivated by a desire of the speakers toexpress themselves in an “extravagant” fashion. Our briefdiscussion shows that this view misses an essential point.The mechanism of rhetorical devaluation, outlined in thepreceding paragraphs is a typical “invisible-hand process”(Keller 1994), brought about by an extremely predictable andorderly collective behaviour which is all but extravagant. Itis motivated by the speakers’ whish to participate in thecommunicative benefits of a generous over-use of the SP. Likeevery “invisible-hand process”, the mechanism of rhetoricaldevaluation produces results which are highly independent ofthe individual speaker’s motivations – in the case discussedhere, it will eventually lead to the obligatorization of theSP in the grammar of the respective language. The hypothesisthat this process is triggered by a fundamental and arguablyuniversal communicative motive – namely the desire to markthe seriousness of speech acts – is capable to explain itspolygenetic character, in particular its independence oflanguage-specific grammatical conditions.

5.Rhetorical devaluation at work

The mechanism of rhetorical devaluation outlined in thepreceding paragraphs is a mere theoretical model which should

20 Ulrich Detges

not confounded with historical reality itself. Is it possibleto observe this mechanism at work somewhere in our writtenrecords of OF? In a famous series of articles on the personalpronouns in Old French, Lucien Foulet (1935/36: 299) cites,as one of the first texts where the increased use of the SPin OF is on its way, the Conquest of Constantinople. This text inprose was written at the beginning of the 13th century, andits style undoubtedly reflects in many ways the spokenlanguage of the epoch (Foulet, ibid.). According to Foulet,an almost modern use of the SP can be observed in this text.17

A passage examined by him in some detail – a speech thatConon de Béthune adresses to the emperor of Constantinople –is given in (20):(20) Speech of Conon de Béthune addressed to the emperor of

Constantinople (§§ 213–4)Sire, nos somes venu a toi de par les barons de l’ost et de

par le duc de Venise. Et 1Sire, we´ve come to you on behalf of the barons of the army and on

behalf of the duke of Venice. And

saches tu que il te reprovent le grant servise que il t’ont fait, con la gent sevent et 2

know that they represent to you the great service that they have rendered to you, as everybody knows

cum il est apparisant. Vos lor avez juré, vos et vostre peres, la convenance a tenir que 3

and as it is apparent. You have sworn to them, you and your father, to keep the promise that

vos lor avez convent ; et vos chartes en ont. Vos ne lor avezmie si bien tenu com vos 4

you have sworn to them, and they have your charters on this. You did not keep as well to it as you

deussiez. Semont vos en ø ont maintes foiz, et nos vos en semonons, voiant toz vos 5

should. They have repeatedly urged you, and we urge you on behalf of them, in front of all

barons, de par als, que vos lor taignoiz la convenance qui est entre vos et als. 6

your barons that you may keep to the convention which exists between you and them.

17 A contrary interpretation is put forward by v. Wartburg(31970: 66-7), who claims that every single use of a SP in thepassage cited above is triggered by some grammatical rule ofOF..

L’obligatorisation des pronoms sujets 21

In this passage there is indeed a great number of SP,sometimes even in places where they would be excluded in MF(e.g. in line 2, saches tu). But is their value in this passagereally the same as in MF? Instead of answering this questionimmediately, let us look at another passage taken from thesame work, namely the narrative passage which immediatelyprecedes the discourse in (20).(21) Narrative preceding Conon’s speech (§§ 211–3)

[...] Ensi monterent li message sor lor chevax, les espeesçaintes, et ø chevaucherent 1

Thus, the messengers mounted their horses, armed with their swords, andtogether they rode

ensemble trosque al palais de Blaquerne. Et sachiez que il [?

i] alerent en grant peril et 2up to the palace of Blaquerne. And know that they [?] went [?there] ingreat peril and

en grant aventure selonc la traïson as Grex. Ensi que ødescendirent a la porte, et 3

in great adventure, because of the perfidy of the Greeks. So theydescended at the gate, and

ø entrerent el palais, et ø troverent l’empereor [...].4entered the palace and found the emperor.

Unlike the passage cited in (20), which represents a(fictitious) direct speech, (21) is a narrative in a third-person perspective. Note that the use of the SP which we findhere corresponds entirely to the conventions of OF: in casesof topic continuity, the text uses null subjects. There is,however, an exception to this rule, namely the occurrence ofil in line 2, to which we will come back later. Why then do wefind so many SP in the discourse passage (20), whereas thereare almost none in the narrative passage (21)? The answer tothis question has to do with the communicative function ofthe respective passages: the direct speech in (20) is centredaround a bold challenge (line 5 ‘we urge you […] in front ofall your barons […]’), i.e. a “strong” speech act. The entirepassage is characterized by an extremely impolite, evenoffensive tone, which can – among other things – be seen fromthe SP in the imperative et saches tu ‘and know’ (line 2).18 As18 The SP are, as already noted by Vance (1997: 294-308),

particularly rare in the second person plural when being used asa form of polite address. This is a strong indication of a close

22 Ulrich Detges

this example shows, the frequent use of the SP in thispassage is part of an overall rhetorical strategy. The topicof the discourse passage in (20), which concerns questions ofreciprocal obligations and responsibilities, provides afurther motivation for the use of the SP (see above, section4). Thus, the use of the SP in (20) is a major stylisticdevice which gives the text an extremely direct and forcefulcharacter. At the same time, however, the stylistic value ofevery individual occurrence gets lost in the sheer mass ofthe SP used. The passage given in (20) is therefore a goodexample for a rhetoric “over-use” of the SP.An explanation is still needed for the use of il in line 2

of (21). Note that at this exact place, several manuscriptsof the Conquest do not have an il ‘they’ but the place adverb i‘there’. However, the group of manuscripts containing i‘there’ is characterized by a systematic lectio facilior.Therefore, we may suppose that they are less faithful thanthe manuscripts containing an il (Faral 21961: XLVII). Why thenthis il ? As the sentence in question contains a summaryevaluation of the protagonists’ behaviour, the function ofthe SP could be to underline their responsible role in therelated events. Interestingly, the SP is not the only deviceof emphasis used here – the formulaic expression et sachiez (l.2) is a stylistic procedure often used in the Conquest whenthe reader’s attention is drawn to a fact considered asnoteworthy by the narrator (see Schon 1960: 161).

6.Inanimate referents and impersonal constructions

If the hypothesis developed in section 3.4. is correct, thenwe should expect SP referring to inanimate entities to becomeobligatory much later than SP with animate referents. This isindeed what we find in contemporary Brazilian Portuguese,where third person SP are markedly less frequent whenreferring to inanimate referents (see Thomas 1969: 94, Barme2001: 207). In Modern French, the possibility to not realizethe SP has been conserved especially in impersonalconstructions. This is not, as one could assume, aninnovation of popular French, but an instance of persistence of

correlation between the use of the SP and politeness phenomenain OF.

L’obligatorisation des pronoms sujets 23

an archaic grammatical feature (Hunnius 1983 [1975]: 355-6).19

Even though SP in impersonal constructions are attested in OFat a relatively early date, their non-realization has nevergot out of use in the spoken language. In (22) we find threeexamples of different epochs of French. Example (22b) istaken from the Journal d’Héroard, a text which can be consideredas an authentic document of the spoken language of the early17th century.(22) a. 16th century French

Jamais ø ne faille debvoir, jamais ø ne faille prester (RabelaisIII, 5, cf. Mair 1992).[It] should never be necessary to owe, and never should [it] be necessary tolend.

b. 17th century spoken French ø fau coupé cela (Hérorad, Journal, cf. Ernst 1993 : 70)[It] is necessary to cut this.

c. Contemporary spoken Frenchø faut te faire une raison (Queneau, Zazie, 13, 87, 88, cf.Mair 1992: 363)[It] is necessary for you to accept the inevitable.

19 For the concept of persistence and its relation togrammaticalization see Hopper (1991: 28-9).

24 Ulrich Detges

7.Subject pronouns and pronominal repriseconstructions

On the basis of the hypothesis laid out in section 3, newlight can be cast on other notorious problems of the historyof French. In Modern French, tonic first and second person SPmust obligatorily be construed with a pronominal reprise.Thus, moi je parle ‘me, I speak’ is grammatical, whereas *moiparle ‘me speak’ is excluded. In OF, however, the pronominalreprise is a rare and extremely marked construction, and itis only used in very specific conditions (Marchello-Nizia1998: 331, Priestley 1950: 148). Originally, it is a means ofcontrastive topicalization, i.e. it expresses a strong noteof emphasis (Waltereit 1996: 73). In early Old French thisconstruction takes up only nominal subjects (‘THE MAN, HECOMES’); specifically, it is never used to repeat subjectpronouns (of the type ‘I, I COME’, Moignet 1973: 130, Härmä1990: 171). Only in the course of the 13th century do repriseswith SP start to become more and more frequent. As we haveseen, the 13th century is the epoch where, corresponding toour hypothesis, rhetorical devaluation makes the SP losetheir originally contrastive effect. Therefore, a new “heavy”procedure of self-topicalization (see section 3.3.) wasneeded, and it was the pronominal reprise construction whicheventually took up this function. The earliest examples ofreprises of SP discussed in the literature (Foulet 1935/36:307, Härmä 1990: 171), show this construction in contextswhich we already know from the foregoing sections: in (23a)the reprise construction accompanies the verb cuidier ‘tothink’, in (23b, c), it is used in contexts of “strong”speech acts.20 (23) Pronominal reprise of subject pronouns (13th century)

a. Cuidier ‘penser, to think’Et jou je cuit si bien feriés / que vos au lonc n’i perdriésja. (Cour. Renart, 1617, cf. Foulet 1935/36: 307)And I, I think that if you did well, / you would not go wrong for along time now.

b. “Strong” speech act

20 However, a real contrastive effect between je ‘I’ and tu ‘you’must be added on top of this in (23c).

L’obligatorisation des pronoms sujets 25

Ysengrins dist : ‘Venés vos ent / et je vos encor endonrai.’ / Renars respont : ‘jou je n’irai.’ (Cour. Renart :596-8, cf. Foulet 1935/36: 307)Ysengrim says: ‘Come with me / and I shall give you more of it.’ /Reinard responds: ‘I, I won´t go.’

c. “Strong” speech actMais je, se je vouloie, je savroie miauz dire que tu le tuen[...]. (Merlin 64/44, cf. Härmä 1990: 171)But I, if I wanted, I would be able to recite mine better than you[are to recite] yours.

Interestingly, in Modern French the pronominal reprise,originally a “heavy” construction, has lost its contrastiveforce in contexts of turn-taking and self-topicalization: inthe spoken language, moi je pense ‘I, I think’ can be used as anon-contrastive construction. Moreover in Modern French, thereprise of the tonic pronoun by a clitic SP has becomeobligatory much later in the third person than in the firstand second person. In written Standard French, which in manyrespects reflects the 17th century usage, it is still possibleto say lui pense instead of lui il pense (see also Koch 1993: 182-3).Our hypothesis is indirectly supported by a look at

Brazilian Portuguese, where the use of pronominal repriseconstructions – extremely rare in peninsular Portuguese – hasincreased simultaneously with the obligatorization of the SP.It seems that in certain varieties of Brazilian Portuguese,the pronominal reprise is on its way to become an unmarkedsentence type, which no longer exclusively serves tohighlight left-dislocated elements (Barme 2001: 238).

8.The expansion of the oblique forms of the SP

Another phenomenon which deeply affected the history of theFrench SP is the expansion of the tonic oblique forms at theexpense of the subject case. For this phenomenon, severaltraditional explanations exist. The most prominent of thesemaintains that the use of the oblique forms of the SP goesback to constructions of the type entre moi et toi ‘me and you’(Foulet 1935/36: 415-6, Gamillscheg 1957: 131, Moignet 1965:77). The argumentation pointed out in this paper suggests anexplanation of a different kind. As pointed out in section3.3., Modern Spanish has “heavy” turn-taking constructions of

26 Ulrich Detges

the type yo para mí ‘I, for myself…’ or yo a mí ‘I, to me…’ (seeex. (8a), repeated below for convenience). Often, suchconstructions are anacolutha, i.e. syntactically ill-formed“false starts”, which nevertheless attain a considerablefrequency in turn-taking contexts (see 3.3.). MoiDAT jeNOM, whichreplaced jouNOM jeNOM in OF, is the symmetrical mirror image ofyoNOM a míDAT. Thus, it is conceivable that in a non-standardizedlanguage as OF, anacolutha of this kind can – as a result oftheir increased frequency – become conventionalconstructions. An argument in support of this hypothesis isthe fact that similar changes took place in the dialects ofNorthern Italy and in spoken English. Once again, thispolygenetic language change, would be difficult to explain bylanguage-specific grammatical factors. The examples (24c, d)show that the change in question is, in principle,independent of a previous grammaticalization of thepronominal reprise.(8a) Anacoluth

Yo a mí no se me pasó por la imaginación. (Corpus Oral 1992)‘ISUBJ to meOBJ this has never crossed my mind.’

(24) Polygenetic replacement of subject forms by oblique forms a. Medieval Boulognese, me < lt. mihiOBJ, a < lt. egoSUBJ Me a deg (cf. Tagliavini, 1973 : 322)‘I, I say.’

b. Medieval Genovese, mi < lt. mihiOBJ, e < lt. egoSUBJ

Mi e me fijo (Arch. glott. 15, 19, cf. Rohlfs 1949: 157).‘I, I imagine.’

c. Venetian, mi < lt. mihiOBJMi son (Rohlfs 1949 : 157)‘I am.’

d. Modern EnglishHow poor we are, you and me! (H. Stretton, cf. Foulet

1935/36 : 74)

9.“Strong” speech acts and parasitical topics.The obligatorization of the SP and the grammaticalization of SVO word order

It is widely accepted that in the course of the Middle Ages,the canonical OF word order T(opic) V(erb) X – with a sentence

L’obligatorisation des pronoms sujets 27

initial topic and the verb in second position, followed byall other constituents – was progressively ousted by SVO, inwhich the subject invariably precedes the verb. Vance (1997:324-6) considers that the obligatorization of the SP is aconsequence of the emergence of the SVO pattern, whileRoberts (1993: 207) views both processes as results of acomplicated parameter-resetting. In this section, I wouldlike to argue that it was the rhetorical devaluation of theSP in OF which paved the way for the SVO order. An earlyexample of the new structure is given in (25), taken from Erecand Enide, a late 12th century text. The topic of the utteranceis the direct object (DO) cest chevalier. This element appears infirst position where it is followed by the subject je, whichalso appears preverbally. Whereas to 21st century speakers ofFrench, this structure looks absolutely normal andunspectacular, it is extremely rare and marked in OF. In(26), we see how the same propositional content would havebeen expressed within the conical sentence structure of OF:the topic appears in first position, the position of thesubject is in the post-verbal slot, where it is preferablyomitted if it is a pronoun (Foulet 1982: 313, Herman 1954:375). (25) Innovative structure: (T) – S – V

Cest chevalierT/DO jeT/S ne l’aimV pas. (Erec 602)This knightT/DO IT/S don´t likeV him.

(26) Canonical OF structure: T – V – XCest chevalierT/DO n’aimV (je)X/S pasThis knightT/DO don´t likeV (I)X/S

In the original text of Erec and Enide, however, the utteranceappears in the form as given in (25) and not as (26). Whythen are two preverbal constituents, cest chevalier ‘this knight’and je ‘I’ needed in the text? The answer is that (26) is a“strong” speech act. In order to give additionalillocutionary force to this speech act, the speaker uses a SP– a grammatically random, “parasitical” second topic,realized for stylistic reasons only. The real topicconstituent of (25) is the direct object cest chevalier, which,in order to assure coherence with the preceding context, mustbe realized outside the sentence borders, since the canonicaltopic-position is occupied by the “parasitical” je. Thus, (25)represents a “double-topic” structure T1–T2V with T2 realized

28 Ulrich Detges

as subject (S). Modern SV first appears as T–SV in contextssuch as (25). The increased frequency of this structure, inwhich a preverbal subject is no longer necessarily the(principal) topic of the utterance (as is always the case incanonical TVX in OF), marks the emergence of SVO word order(see Vance 1997: 264-74, Kaiser 1998: 191). At the end of the12th century however, this construction was still a highlyrare structure, designed to express a special effect. It wasto become an unmarked sentence type only to the extent thatthe “parasitical” je, as a consequence of its rhetoricaldevaluation, lost its contrastive function and became anextremely frequent element. This line of reasoning alsoexplains the observation made by Marchello-Nizia (2000) thatthe macro-syntactic evolution of the French phrase-typepattern is characterized by “topic decumulation” (décumul duthème), i.e. by a progressive increase of the preverbal slotsreserved for topical elements.

The hypothesis of a rhetorical devaluation of the SPsketched in this section provides an explanation only for theemergence of the (T)SV structure (25) and its progressivefrequency increase. It does not, however, account for thelater disappearance of TVS with subject inversion, as given in(26). In classic OF, a “strong” speech could, in principlealso be realized in the form TVS, as shown in (26). Example(27) documents this case:(27) Et encoreT vos diV geS autre chose (Graal, 6, 3)

‘Et moi je vous dis encore une chose...’‘And I, I tell you even one more thing...’

According to Vance (1997: 350, 1989: 282), the word orderpattern TSV as in (25) was not the only innovative structurewhich emerged during the Middle French period. Anotherstructure, namely TVSP as in (27)), an instance of TVX with Xrealized as an inverted pronominal subject (SP), was alsomodestly successful, albeit to a much lesser extent than TSV,which. This temporary frequency increase of TVSP is all themore surprising since, at the same time, the pattern TVSNwith nominal Subject became more and more rare. The rise infrequency of both TSPV and TVSP was a direct consequence ofrhetorical devaluation, which increased the number ofoccurrences of the SP independent of their position. Why thendid this process favour constructions with the SP inpreverbal position as in (25) more than inverted SP as in(27)? Interestingly, the frequency increase of the SP inBrazilian Portuguese only promotes the structure (T)SV, while

L’obligatorisation des pronoms sujets 29

SP in post-verbal position are almost excluded today. Barme(2001: 233) hypothesizes that after the loss of theircontrastive effect, the SP of Brazilian Portuguese are nolonger suited for inversion, which is reserved to rhematic(and hence accentuated and contrastive) elements. Thishypothesis is partly confirmed by the behaviour of theSpanish SP in expressions of self-topicalization of the typeyo creo – here, the variant creo yo with subject inversion is amarked construction, which expresses a veritable contrastwith another referent (see section 3.4.). Prévost (2001)shows that such an effect is systematically observable forinverted SP in Middle French:

Subjects in post-verbal position, both nominal and pronominal,not only almost always have to do with utterance continuity – oftenexplicit – with the preceding context; at the same time they arealso something like a form of rupture [...]. Introduction and re-introduction of referents, narrative transitions and transitionsbetween utterance-exchanges, unexpected linkages, syntactic andlogic-pragmatic breaks: the continuity is always at the sametime discontinuity [...]. Thus, [subject-inversion, U.D.] at thesame time entails a ‘highlighting effect’, which, depending onthe individual case, emphasizes either the subject as such orthe utterance as a whole. (Prévost 2001: 309, translation mine,U.D.).

In preverbal position, however, the “parasitic” SP of MiddleFrench gradually became a mere dummy-element which onlymarked mere topic continuity, a function which was hithertofulfilled by ø alone (see (1a)). Marking the grammaticalperson of the verb, which, since the loss of the verbalinflection in MF, is the principal raison d’être of the clitic SPin Modern French (see section 2), was originally nothing morethan a dispensable side-effect of their rhetorical over-useand the consequent devaluation.

30 Ulrich Detges

10. Bibliography

Reference texts

Alex = La Vie de Saint Alexis. An eleventh century Old French poem in 125stanzas. Ed. and transl. by Joseph E. Price. – IndianaUniversity : http://www.geocities.com/ Athens/ Acropolis/ 8716/alexis.html.

ChRol = Das Altfranzösische Rolandslied : Nach der Oxforder Handschrift. ed. by A.Hilka. 8., rev. Ed., by Max Pfister 1997. – Tübingen : Niemeyer(Sammlung romanischer Übungstexte 314).

Conquête de Constantinople = Villehardouin. La conquete de Constantinople. 2 vol.Ed. and transl. by Edmond Faral. – Paris 21961 : Belles Lettres.

Corpus Oral 1992 = Francisco A. Marcos Marín et al. : Corpus de Referenciade la Lengua Española Contemporánea : Corpus Oral Peninsular. – Madrid (s.a.) : Universidad Autónoma, http://www. lllf.uam.es/ ~fmarcos/informes/ corpus/ corpulee.html.

Erec = Chrétien de Troyes. Erec und Enide. Transl. and Preface by IngridKasten. – München 1979 : Fink (= Klassische Texte desromanischen Mittelalters 17).

Graal = La queste del Saint Graal : roman du XIIIe siècle. Ed. by AlbertPauphilet. – Paris 21984 : Champion (= Les classiques français dumoyen âge).

Scientific works

Barme, Stefan (2001) : Der Subjektausdruck beim Verb in phonisch-nähesprachlichen Varietäten des europäischen Portugiesisch und Brasilianischen.– Frankfurt a. Main etc. : Lang (= Publikationen desFachbereichs Angewandte Sprach- und Kulturwissenschaft derJohannes-Gutenberg-Universität Mainz in Germersheim : Reihe A,Abhandlungen und Sammelbände 28).

Barrenechea, Ana María, A. Alonso (1977) : Los pronombrespersonales sujetos en el español hablado en Buenos Aires. – In :

L’obligatorisation des pronoms sujets 31

J. M. Lope Blanch (ed.) : Estudios sobre el español hablado en las principalesciudades de América, 333-350. Mexico : UNAM.

Bentivoglio, Paola (1987) : Los sujetos pronominales de primera persona en elhabla de Caracas. Caracas : Universidad Central de Venezuela.

Berschin, Helmut, J. Felixberger, H. Goebl (1978) : FranzösischeSprachgeschichte. – München : Hueber.

Bossong, Georg (1979) : Prolegomena zu einer syntaktischenTypologie der romanischen Sprachen. – Dans : M. Höfler, H.Vernay, L. Wolf (eds.) : Festschrift Kurt Baldinger zum 60. Geb., Vol. 1.,54-68. Tübingen : Niemeyer.

Buridant, Claude (2000) : Grammaire nouvelle de l’ancien français. –http : //www.editions-sedes.com.

Dahl, Östen (1998) : Grammaticalization and the life-cycle ofconstructions, Keynote lecture at the 17th ScandinavianConference of Linguistics, August 20-22, 1998. (Exposé presentedau Linguistisches Kolloquium du Seminar für Sprachwissenschaft in Tübingen,3rd May 1999).

Detges, Ulrich (2003) : La grammaticalisation des constructions denégation dans une perspective onomasiologique, ou : ladéconstruction d’une illusion d’optique. – in: P. Koch, A. Blank(eds.) : Kognitive romanische Onomasiologie und Semasiologie, 213-233.Tübingen : Niemeyer (= Linguistische Arbeiten 467).

– (in print) : La gramaticalización de los acusativospreposicionales en las lenguas iberorrománicas : Una hipótesispragmática. – To appear in : V. Bellosta von Colbe, G. Knaur(eds.) : Variación sintáctica en español : un reto para las teorías de la sintaxis.Tübingen : Niemeyer (= Linguistische Arbeiten).

Duarte, Maria Eugênia L. (1996) : Do pronome nulo ao pronomepleno : a trajetória do sujeito no português do Brasil. – In :I. Roberts, M. A. Kato (eds.) : Português Brasileiro. Uma viagemdiacrônica. Homenagem a Fernando Tarallo, 107-128. Campinas : Unicamp.

– (2000) : The loss of the « avoid pronoun » principle inBrazilian Portuguese. – In : M. A. Kato, E. V. Negrão (eds.) :Brazilian Portuguese and the Null Subject Parameter, 17-36. Frankfurt a. M. :Vervuert (= Iberoamericana. Lingüística 4).

Ernst, Gerhard (1985) : Gesprochenes Französisch zu Beginn des 17.Jahrhunderts : direkte Rede in Jean Héroards « Histoire particulière de Louis XIII »(1605-1610). – Tübingen : Niemeyer (= Beihefte zur Zeitschrift fürRomanische Philologie 204).

Faral, Edmond (21961) : Villehardouin. La conquête de Constantinople. – Paris :Belles Lettres.

Foulet, Léon (1935/36) : L’extension de la forme oblique du pronompersonnel en ancien français. – In : Romania 61, 257-315, 401-463, Romania 62, 27-91.

– (1982 [31928]) : Petite syntaxe de l’ancien français. Paris : Champion.Franzén, Torsten (1939) : Etude sur la syntaxe des pronoms personnels sujets en

ancien français. – Uppsala : Almqvist & Wiksell.

32 Ulrich Detges

Gamillscheg, Ernst (1957) : Historische Französische Syntax. – Tübingen :Niemeyer.

Givón, Talmy (1976) : Topic, pronoun and grammatical agreement.– In : Ch. N. Li (ed.) : Subject and Topic, 149-185. New York etal. : Academic Press.

Grice, H. Paul (1975) : Logic and conversation. – In : P. Cole, J.L. Morgan (eds.) : Speech Acts, 41-58. New York : Academic Press (=Syntax and semantics 3).

Härmä, Juhani (1990) : Les constructions disloquées en ancienfrançais : problèmes de définition et de délimitation. – In : G.Kleiber (ed.) : L’anaphore et ses domaines, 159-182. Paris :Klincksieck (= Recherches linguistiques 14).

Harris, Martin (1978): The Evolution of French Syntax. A comparative approach.London: Longman (= Longman linguistic library 22).

Haspelmath, Martin (1999) : Why is grammaticalizationirreversible?. – In : Linguistics 37, 1043-1068.

Herman, József (1954) : Recherches sur l’ordre des mots dans lesplus anciens textes français en Prose. – In : Acta LinguisticaAcademiae Scientiarum Hungaricae IV, 69-94, 351-382.

Hilty, Gerold (1975) : Westfränkische Superstrateinflüsse auf diegalloromanische Syntax. – In : Romanische Forschungen 87, 413-426.

Hochberg, Judith (1986) : Functional compensation for /s/ deletionin Puerto Rican Spanish. – In : Language 62, 609-621.

Honnigfort, Eva (1993) : Der segmentierte Satz. Syntaktische und pragmatischeUntersuchungen zum gesprochenen Französisch der Gegenwart. – Münster :Nodus (= Münstersche Beiträge zur Romanischen Philologie 8).

Hopper, Paul J. (1991) : On Some Principles of Grammaticization.– In : E. Closs Traugott, B. Heine (eds.) : Approaches toGrammaticalization. Vol. 1. Focus on Theoretical and MethodologicalIssues, 17-35. Amsterdam, Philadelphia : Benjamins (= Typologicalstudies in language 19.1).

Hunnius, Klaus (1983 [1975]) : Archaische Züge des français populaire.– In : F. J. Hausmann (ed.) : Die französische Sprache von heute, 345-365. Darmstadt : Wissenschaftliche Buchgesellschaft.

Jensen, Frede (1990) : Old French and Comparative Gallo-Romance Syntax.– Tübingen : Niemeyer (= Beihefte zur Zeitschrift für RomanischePhilologie 232).

Kaiser, Georg (1992) : Die klitischen Personalpronomina im Französischen undPortugiesischen. Eine synchronische und diachronische Analyse. – Frankfurt a.M. : Vervuert (= Iberoromanica 44).

– (1998) : Verb-Zweit-Effekte in der Romania. Eine diachronischeStudie mit besonderer Berücksichtigung des Französischen. –Hamburg : Habil., ms.

Kattinger, Gernot (1970) : Die Verwendung des Personalpronomens als Subjektzum Verbum, dargestellt an « Erec und Enide » von Chrétien de Troyes. Erlangen-Nürnberg : Diss.

Keller, Rudi (21994 [1990]) : Sprachwandel. Von der unsichtbaren Hand in derSprache. – Tübingen, Bâle : Francke (= Uni-Taschenbücher 1567).

Koch, Peter (1993) : Le « chinook » roman face à l’empirie. Y-a-t-il une conjugaison objective en français, en italien et en

L’obligatorisation des pronoms sujets 33

espagnol et une conjugaison subjective predeterminante enfrançais? – Dans : G. Hilty (ed.) : Actes du XXe Congrès International deLinguistique et Philologie Romanes. Zurich, 6-11 April 1992. Vol. 3,Section 4, Typology of the Romance languages, 171-190. Tübingen,Bâle : Francke.

– (1994) : Dépersonnalisation (et repersonnalisation). A propos dela diachronie des verbes impersonnels. – In : L’informationgrammaticale 62, 9-11.

Koch, Peter, W. Oesterreicher (1990) : Gesprochene Sprache in derRomania : Französisch, Italienisch, Spanisch. – Tübingen : Niemeyer (=Romanistische Arbeitshefte 31).

Kuen, Heinrich (1970 [1957]) : Die Gewohnheit der mehrfachenBezeichnung des Subjekts in der Romania und die Gründe ihresAufkommens. – In : H. Kuen (ed.) : Romanistische Aufsätze, 154-184.Erlangen, Nürnberg : Carl Nürnberg (= Erlanger Beiträge zurSprach- und Kunstwissenschaft 35).

Lazard, Gilbert (1994) : L’actance. – Paris : Presses Universitairesde France.

Lehmann, Christian (1995 [1982]) : Thoughts on Grammaticalization.– München : Lincom (= Lincom studies in theoretical linguistics1).

Linder, K. Peter (1987) : Grammatische Untersuchungen zur Charakteristik desRätoromanischen in Graubünden. – Tübingen : Narr (= Tübinger Beiträgezur Linguistik 275).

Mair, Walter (1992) : Expressivität und Sprachwandel : Untersuchungen zur Rolleder Subjektivität in der Entwicklung der romanischen Sprachen. – Frankfurt a.M. et al. : Lang (= Europäische Hochschulschriften 178).

Marchello-Nizia, Christiane (1998) : Dislocations en diachronie :Archéologie d’un phénomène du « français oral ». – In : M.Bilger, K. van den Eynde, F. Gadet, (eds.) : Analyse linguistique et ap-proches à l’oral. Recueil d’études offert en hommage à Claire Blanche-Benveniste, 327-337. Louvain, Paris : Peeters (= Orbis. Supplementa 10).

– (2000) : Le décumul du « thème » dans l’évolution du français.– In : Le français moderne 68, 31-40.

Martín Rojo, Luisa, M. Meeuwis (1993) : Referentes del sujetopronominales y tácitos en la conversación en español : unenfoque pragmático. – In : H. Haverkate, K. Hengeveld, G. Mulder(eds.) : Diálogos hispánicos 12. Aproximaciones pragmalingüísticas al español, 87-118. Amsterdam : Univ.

Meyer-Hermann, Reinhard (1996) : Sobre el uso del sujeto yo en elhabla culta de Costa Rica. – In : Th. Kotschi, W. Oestereicher,K. Zimmermann (eds.) : El español hablado y la cultura oral en España eHispanoamérica, 279-301. Frankfurt a. Main, Madrid : Vervuert,Iberoamericana (= Bibliotheca Ibero-Americana 59).

Moignet, Gérard (1965) : Le pronom personnel français. Etude de psycho-méchanique historique. – Paris : Klincksieck.

– (1973) : Grammaire de l'ancien français : morphologie - syntaxe. – Paris :Klinksieck (= Initiation à la linguistique 2).

34 Ulrich Detges

Oliveira, Marilza de (2000) : The pronominal subject in Italianand Brazilian Portuguese. – Dans : M. A. Kato, E. V. Negrão(éds.) : Brazilian Portuguese and the Null Subject Parameter, 37-53. Frankfurta. M. : Vervuert (= Iberoamericana. Lingüística 4).

Prévost, Sophie (2001) : La postposition du sujet en français aux XVe et XVIesiècles. Anayse sémantico-pragmatique. – Paris : CNRS éditions.

Priestley, L. (1950) : Reprise Constructions in the « Song ofRoland ». – In : Archivum Linguisticum 2, 144-157.

Raible, Wolfgang (1996b) : Kognitive Grundlagen des Sprachwandels.– In : S. Michaelis, P. Thiele (eds.) : Grammatikalisierung in derRomania, 61-80. Bochum : Brockmeyer (= Bochum-Essener Beiträge zurSprachwandelforschung 28).

Roberts, Ian (1993) : Verbs and Diachronic Syntax. A Comparative History ofEnglish and French. – Dordrecht : Kluwer.

Rohlfs, Gerhard (1949) : Historische Grammatik der italienischen Sprache undihrer Mundarten. Vol. 2. Formenlehre und Syntax. – Bern : Francke.

Rosengren, Per (1974) : Presencia y ausencia de los pronombres personalessujetos en español moderno. – Stockholm : Almqvist & Wiksell (=Romanica gothoburgensia 14).

Sacks, Harvey, E. A. Schegloff, G. Jefferson (1978) : A simplestsystematics for the organization of turn taking forconversation. – In : J. Schenkein (ed.) : Studies in the organization ofconversational interaction, 7-55. New York et al. : Academic Press.

Schon, Peter M. (1960) : Studien zum Stil der frühen französischen Prosa : Robertde Clari, Geoffroy de Villehardouin, Henri de Valenciennes. – Frankfurt a. M. :Klostermann (Analecta Romanica 8).

Silva-Corvalán, Carmen (1993) : On the permeability of grammars :evidence from Spanish and English contact. – In : W. J. Ashby,M. Mithun, E. Raposo (eds.) : Linguistic perspectives on the RomanceLanguages. Selected papers from the 21st Linguistic Symposium onRomance Languages (LSRL XXI), Santa Barbara, California, 1991,19-43. Amsterdam, Philadelphia : Benjamins (= Amsterdam studiesin the theory and history of linguistic science 103).

Tagliavini, Carlo (1973) : Einführung in die romanische Philologie.Transl. from Italian by Reinhard Meisterfeld. – München : Beck(= Handbücher für das Studium der Romanistik).

Thomas, Earl W. (1969) : The Syntax of Spoken Brazilian Portuguese.– Nashville : Vanderbilt University Press.

Thun, Harald (1989) : Der Gebrauch des Subjektpronomens der 3.Person im Romanischen aus typologischer Sicht (Diachron undmoderne Synchronie). – In : W. Raible (ed.) : Romanistik,Sprachtypologie und Universalienforschung. Beitr. zum FreiburgerRomanistentag 1987, 191-222. Tübingen : Narr (= TübingerBeiträge zur Linguistik 332).

Thurneysen, Rudolf (1892) : Zur Stellung des Verbums imAltfranzösischen. – In : Zeitschrift für Romanische Philologie XVI, 289-307.

Vance, Barbara S. (1989) : Null Subjects and Syntactic Change in MedievalFrench. – Ithaca, NY, Cornell Univ., Diss.

L’obligatorisation des pronoms sujets 35

– (1997) : Syntactic Change in Medieval French : Verb-Second and Null Subjects.Dordrecht : Kluwer.

Vanelli, Laura (1987) : I pronomi soggetto nei dialetti italianisettentrionali dal Medio Evo a oggi. – In : Medioevo Romanzo XII,173-211.

Vennemann, Theo (1974) : Topics, Subjects, and Word Order : FromSXV to SVX via TVX. – In : J. M. Anderson, C. Jones (eds.) :Historical Linguistics. Syntax, Morphology, Internal and Comparative Reconstruction.Vol. 1, 339-76. Amsterdam, Oxford : North-Holland (= Northholland linguistic series 12a).

Waltereit, Richard (1996) : « Les cours, ils auront déjàcommencé ». Zur Funktion der Linksversetzung des Subjektszwischen Informationsstruktur und Referenzsemantik. – In :Romanistisches Jahrbuch 47, 64-81.

Wanner, Dieter (1993) : L’expression du sujet dans les languesromanes. – In : G. Hilty (ed.) : Actes du XXe Congrès International deLinguistique et Philologie Romanes. Zurich, 6-11 April 1992. Vol. 3,Section 4, Typology of the Romance languages, 448-460. Tübingen,Bâle : Francke.

Wartburg, Walther v. (31970) : Einführung in Problematik und Methodik derSprachwissenschaft. – Tübingen : Niemeyer.