4:2.183-201
Pragmatics
Prasmatiqs
Association
International
PRAGMATIC CONNECTTVES
AND L2 ACQUISTTION:
THE CASE OF FRENCH AND DUTCH'
Beatrice Lamiroy
0. Introduction
The purposeof this paper is to addressthe questionof connectivesfrom a contrastive
pointof view. By connectivesI understanda particularclassof linguisticelements,viz.
adverbsand conjunctionsthat function as linkers betweenclauses,sentencesor larger
textfragments.The languagesto be dealt with are French and Dutch.
The present researchgrew from the following empirical observation:Dutch
speakingstudentsof French encountervery seriousdifficulties in correctly handling
Frenchconnectives.2
Given that connectivesalso exist in their native language,the
obviousquestionto raise is why this is so: why do thosestudentshave so much trouble
to usepoLrtant, toul de mAme, alors, etc. in an appropriate way ? The equivalence
problem,which is a central issuein contrastivelinguistics,appears to be particularly
seriousin the domain of connectives.
Furthermore,when we dealwith syntacticand semanticproblemsof French,we
canusuallyrefer to some rule system,to a chapterof the grammar or to a dictionary.
Whenwe deal with connectives,however,we are often left with a statementsuch as
"thisis not quite the way a French native speakerwould sayit". This amountsto saying
thatconnectives
are not only badly usedby students,but that they are also often poorly
taughtby teachers.
In order to gain insightinto this problem,and eventuallyto solveit, I would like
to explorefwo hypotheses,the secondbeing probably a little more controversial- I
hoperefreshinglycontroversial- than the first. The first idea I will provide evidencefor,
is that adequateuse of connectivesis intrinsicallya difficult matter, even for native
speakers.This may be so becauseconnectivesinvolve a variety of processes,both
cognitiveskills and linguistic competence.To the former belong abilities such as
capacity,logicalreasoning,concentration,etc.The latter in turn involvesboth
deductive
grammaticaland pragmaticcompetence,two componentsof languagewhich are not
1 I *ish ro thank an anonymous reader of this journal for his/her very valuable comments and
suggestions.
2 Most of the data are taken from M.A. dissertations written by Belgian Dutch speaking, i.e.
Flemish students. Although those students are non-native speakers of French, they usually have had an
averageof 10 years of training in written and spoken French.
184
Beatrice Lantiroy
necessarilystructured in a similar way.
Lrt me now turn to the second hypothesis.I will adventure the following
question: lexicon studieshave establishedthat languagesvary as to their resourcesto
expressdifferent things.As is well known,Eskimosreportedlyhavemany differentways
to talk about snow, while Arabic has many different words to refer to a camel: for
example,a pregnant female camel is calleddifferentlynot only from a male camel,but
also from a non pregnant female camel. lndeed,she is worth the price of two camels
when she is sold at the market. So, my hypothesiswill be : why would the same not
hold for other domains of language,in particular that of pragmatic connectives? In
other words, could it be the casethat evenif two languages,suchas French and Dutch,
use connectives,they use them in different ways,and more particularly,to a different
extent? In fact, there seemsto be some evidencefor this, as I will attempt to show.
Although the motivationto undertakethe presentresearchon connectivesstems
from a teaching problem, I would like to point out that other fields of applied
linguisticsbadly need a better understandingand a more thorough knowledge of
connectives,for example that of Natural LanguageProcessing.Computers, in more
ways than one, can be consideredas non native speakersof a language,and are
probably the worst foreign languagestudents in the world. In many current NLP
projects, the analysis,synthesisor translationof connectivesis a major, and to my
knowledge unsolved,problem. Therefore,whatever progresscan be achievedin our
understandingof connectives,it will also be of use in other domains than foreign
languageteaching.
Another introductoryobservationruns alongsimilar lines.Although it is usually
the casethat a scientifictheoryleadsto applications,
the road sometimesmay alsogo
the other way round. Indeed, a significantneed in one of the "applied" fields of a
scientificdomain may suddenlyrevealan area that still lackssolid descriptiveresearch,
let alone theory construction.In my view, connectivesmight be such a case.Although
theoreticallinguisticshas witnesseda stronginterestin pragmaticallyoriented matters
for the last decades,there still are many unexploredand unresolvedquestionswith
respectto connectives,especiallyfrom a contrastivepoint of view.
The paper is organisedas follows. I will, first, sketch a number of general
properties of connectives.In the secondpart, I will illustratethe problem that foreign
languagestudentshavewith Frenchconnectivesby presentinga sampleof errors.Most
of them are taken from textswritten by my own students,but somecan be found in two
other papers (Kielhofer & Poli I99I, Nolke 1989).Interestingly,the authors of these
papers,who are respectivelyGerman and Danish,report about their studentsof French
havingsimilar problemswhen dealingwith pragmaticconnectives.
This of courseleads
to the idea that one of the tactors at stake here might be a more general difference
betweenGermanic languagesand French.In the third part, I will turn to the languageinternal hypothesis and finally, in the fourth, to the typological or contrastive
hypothesis.
hagmaticconnectives
185
1. General properties of connectives
lrt me recall, for a start, that connectivesbelong to the basicsof languagestructure,
as was pointed out by Harris (1968). According to Harris, language structure is
fundamentally
basedon relationshipsthat relate minimal units to constructs,so that the
resultingmodel includes:
-
a set of kernel sentences
a set of transformations
a set of connectives
one set of rules as to which transformations can be carried out in successionon the same kernel
another set of rules as how connectives can be strunq in succession.
Though connectives are essential components of language, we lack rystematic3
knowledgeabout them. A first reason why this is so may be their extremely
heterogeneous
characterwith respectto parts of speech.The parts of speechapproach,
whichis commonpracticein linguisticanalysis,cannotyield satisfactoryresultsbecause
aboutany lexicalcategorycan take on the role of connective,as shown in (1):
(1)
a.
b.
d.
e.
ADVERBS: pourtant, donc
[however,therefore]
CONJUNCTIONS: mais, de sorte que
[but, so that]
NOUN (PHRASE)S: somme toute
[it. sum total, all in all]
VERB (PHRASE)S: cela dit, tout compte fait
[it. this said,however;lit. all bill made, after all]
PREPOSITION(AL PHRASE)S: par contre, en revanche
[it. by against,however;lit. in revenge,however]
IDIOMATIC EXPRESSIONS:tant qu'd faire, comme quoi
[it. so much as to do, while you are at it; lit. like what, from
where it showsthat]
Casessuchas thoseexamplifiedin (lf) are hardly translatable:they are often fossilized
remnantsof earlier stagesof the language,whoseoriginal meaninghas become totally
opaqueeven to native speakersof French.The examplesin (1) show that there is no
3 French
connectives have been more thoroughly investigated than the Dutch equivalents. For
French,the most common connectives have been listed and classified (Berendonner 1981; Bieber 1988;
Blumenthal 1980; Ducrot et al. 1980; Roulet et al. 1985; Ruquet & Bodin 1988) and there exist several
careful and detailed studies of a number of them (e.g. Anscombre & Ducrot 1977; G{ilich & Kotschi
1983;Zenone 1982,1983). In the Dutch domain, the ANS (Geerts et al. 1984: 401ff and 64ltt.) provides
significantinformation on adverbial connectives and on conjunctions, but detailed studies are almost
exclusivelyconcerned with connectives of contrast (e.g. Foolen 1993; Spooren 1989; Van Belle & De
Vrooij 1992) and cause (Van Belle 1989).
186
Beatrice Lantiroy
"word class"of connectives,so that if one looks for them in standardgrammars,they
are scatteredin different parts of the grammar,some of them being totally absent.
A secondrelevantobservationis that grqmmar hastraditionallybeen conceived
of in a narrow sense,as the study of word combinationsthat are to be accountedfor
in terms of rules and principleswithin the sentence.However, from a formal point of
view, one of the defining propertiesof connectivesis that they mostly function outside
the syntacticdomain of the sentence,thus falling outsidethe domain of syntacticrules.
This is not only suggestedby their positional properties - they usually occur at the
beginning of the sentence-, but also by the fact that they are not sensitiveto basic
syntacticparameterssuchasvalencyor embedding.As illustratedin (2a-b),connectives
do not belong to the actantialpattern of the sentence.In (2a),au fottd has a connective
function and is not an argument of the verb avoir raisort,whereas in (2b), au fond
fulfills the role of a locativecomplement,argumentof the verb mettre.Connectivesdo
not have either a valencyof their own. Whereasthe locativecomplementof (2b) can
take a complement of its own, e.g.du tiroir in (2c), the latter possibilitydoes not exist
when we are dealingwith the connective,as is shownby (2d):
(2)
a.
b.
c.
d.*
Au fond, Max a raison
[(lit. At the bottom) In fact, Max is right]
Max a mis la lettre au fond
[Max put the letter at the bottom]
Max a mis la lettre au fond du tiroir
[Max put the letter at the bottom of the drawer]
Au fond de I'histoire,Max a raison
[At the bottom of the story,Max is right]
The examplesin (3) showthat, as hasbeenpointedout by Rubattel(1982)and others,
many connectivesare not subject to elementary linguistic operations such as
embedding,i.e. they are not allowedin an embeddedclause:
(3)
a.
b.*
Max ne se sentaitpas bien. Aussi n'est-ilpas venu d la r6union
[Max did not feel well. So he did not come to the meeting]
Pierre m'a dit que Max ne se sentaitpas bien et qu'aussiil n'est
pas venu ir la r6union
[Peter told me that Max did not feel well and that so he did not
come to the meeting]
On the other hand, connectivesdo not seemto exhibit a particularsyntacticbehaviour
of their own, hence the difficulty of assigninga specific syntactic function to this
composite class.So, if they can be analyzedneither in terms of lexical categoriesnor
in terms of syntacticcategories,what is there to be saidabout their pragmaticfunction?
Connectiveshave at least a twofold discursivefunction, an argumentativeand
an interactivefunction (cf. Berendonner1981;Mainguenau1987;Roulet et al. 1985).
They either link differenttext segmentsor utteranceswith one anotheror they simply
hagmaticconnectives
I87
connectspeakerand hearer.To put it still differently,they may establishrelationships
betweenpredicativestructuresas well as betweenpeople,viz. the participantsof the
linguisticexchange.aWh_ereas
(4a) containsun u.gu-entative connectiv", in 1+U;it,
.
merefunction is to signal the relationshipwith the speechpartner:
(4)
a.
b.
Max a menti h son patron. par cons6quentil a 6,t6,vir6,
[Ma" has lied to his boss.Therefore he has been fired]
Voyons, c'est 6vident que Max a menti
[(lit. Let's see) Come on, it is obviousthat Max has lied]
In what follows, I will only deal with argumentativeconnectives"s
But even from the
argumentative
point of view, connectivesmay contributeto what Lundquist (1987)calls
"the argumentativeprogram" by perfbrming quite diversifiedspeechacts.As has
also
beenstressedby other linguists(Auchlin & Zenone 1980;Duirot et al. 1980;Morel
1984;Rubattel 1982),connectivesmay refer to the mere propositionalcontent of the
utterance,as in (5a), to the speaker'sinvolvementwith respectto his utterance,as in
(5b),or to its implicit import, as is the casein (5c) where the implicit part of the
utteranceis somethinglike "...(mais)je n'en boirai pas,parce que ...",
(5)
a.
b.
une pierre tombe parce qu'elleest pesante(Stendhal)
[A stone falls becauseit is heary]
Il doit €tre parti puisquesa voiture n'est plus ld
[He must be gone sincehis car is no longer there]
J'ai envie d'un whisky mais c'est mauvaispour la sant6
[I t-eellike having a whisky but it is bad for my health]
Moreover,argumentativeconnectivesmay differ according to the illocutiona act
ry
performedby the sentence.Roulet et al. (1985: 145) thus unrouer"d that ainsi and,
aussi,
which both function as connectivesof consequence,
are not equally compatible
with two ditferent speech acts.Theyare equivalent in (6a), which only coniains a
statement,but in (6b), which combinesa statementand an order, ainsi js no longer
possible:
(6)
a.
b.
Ils ont obtenu leur visa,ainsi/aussiils ont pu traverserla frontidre
[They have obtained their visa, (lit. so, also) so they have been
able to crossthe border]
Vous devezobtenir un visa, aussi/*ainsicontactezl,ambassade
* In
the latter case, Nglke (19g9) talks of secondary connectives, because they are no longer
linkers between utterances properly speaking, but they do give the hearer instructions as to the correct
interpretationof the utterance.
5 For a
particular type of interactive connectives, viz. that of verbal connectives, see l-amiroy
& Swiggers(1992).
188
Beatrice Lamiroy
[You have to obtain a visa, so get in touch with the embassy]
Worth noting too is that one and the same connectivemay have different properties
accordingto whether its function is argumentativeor interactive.Sentence(7a) and (7c)
are equivalent,but not (7b) and (7d). In (7a), commequoi introducesa consequence
and functionsas an argumentativeconnective:the relationbetweenthe two clausescan
therefore be reversedinto a causalrelation,as shownin (7c). In (7b), however,we are
dealing with an interactive use of the same connectivewhich allows the speaker to
expresshis conclusion:
(7)
a.
b.
c.
d.*
Il a quitt6 la r6gion,comme quoi il est impossibleque tu I'aies
rencontr6
[He has left the region, (lit. like what) therefore it is impossible
that you have met him]
Il s'esttromp6, comme quoi personnen'est infaillible
[He has made a mistake,(lit. like what) from where it showsthat
nobody is pertect]
Comme il a quitt6 la r6gion, il est impossibleque tu I'aies
rencont16
[Since he has left the region, it is impossiblethat you have met
himl
Comme il s'esttromp6,personnen'est infaillible
[Sincehe has made a mistake,nobody is perfect]
A final reasonwhy connectivesare a thorny problem for linguisticdescriptionis that
they hardly can be analyzedwithout taking into account the larger background or
context againstwhich information is processed.That the role of the context cannot be
dismissed seems undoubtful, since it supplies the assumptionsthat bear on the
assessment
of the adequateinterpetationof the text. Accordingto Sperber & Wilson's
(1986) relevancetheory, hearersinterpret utteranceson the assumptionthat they can
recover information relevant to them. Following Blakemore's (1989) analysis,
connectivescan be viewed as constraints6on the hearer'schoice of context,or to put
the same differently,their function is to guide the interpretationprocessby specifying
the relevant properties of the context: connectivesare constraintson relevance.Not
only do different connectivescontribute to the overall coherenceof the discoursein
different ways,but the very presenceof a connectiveis significant:the speaker,by using
one, directs the hearer towards the relevant interpretation by making certain
assumptions immediately accessible.In the opposite case, viz. the absence of a
connective,the relevant connectionis lessreadily or not at all accessible:the speaker
is lessdirectivewith respectto his speechpartner. For example,both sentencesof (8)
conveya contrastbetweenthe two propositions.However,as Blakemore (1989) points
6 Thi. view is in fact very close to the analysis proposed by Moeschler (1985) and Roulet et al.
(1985) in which connectivesare consideredas "instructions'for the correct interpretation of the text.
hagmaticconnectives
189
out, in (8b) the idea of contrastis likely, becausethere is a natural antinomy between
tall an sltort, but the speakerdoesnot necessarilyhave this contrastin mind, whereas
in (8a) it is totally undoubtful, sinceit is explicitlystated:
( 8)
a.
b.
Jane is tall but Mary is short
Jane is tall. Mary is short
The latter point is, as we will see,not without interestfor the typologicalpart of the
presentresearch.But before turning to this aspect,let us take a look at some of the
currenterrors made bv non-nativestudentsof French .
2. Sampleof errors
Threetypesof errors occur in French argumentativetexts.A first type consistsin not
usinga connectivewhere we would expect one. The secondtype correspondsto the
incorrectuse of a particular connective.The third type is the excessiveuse of a few
connectives
such as alors ot quand m4me which, for some reason,are the student's
favoriteconnectives,not only of Dutch speakingstudentsbut also of German students,
asKielh6fer& Poli (1991)report.
The first type of "error" is illustratedby the examples(9 - 11)"The textsin (9)
and (10) are due to a Belgian student,example (11) is taken from Kielhofer & Poli
(1991).' Although the absence of connectivescannot always be penalised as a
"mistake",the total lack of linkers between successivepropositionsbrings about a
feelingof not having to do with a piece of French text. In certain cases,however,the
coherence
of the text really requiresthe presenceof a connective,thus ainsi in (9) and
tttertmoilBin (10) seemquite indispensable:
(9)
Cette analysedu passifanglaissera pr6sent6edans la synthdsequi suit.
[Cependant,toutefois] Notre r6sum6 tient compte du sujet de notre
m6moire. [Ainsi] Nous avons essay6de r6duire les 6l6ments qui se
rapportent uniquement e I'anglais et de mettre en 6vidence [au
contraire, en revancel ce qui a trait au passifen g6n6ral.
[This analysisof the English passivewill be presentedin hte following
synthesis.(However) our synthesistakes into accountthe subject of our
dissertation.(Thus) we havetried to reducethoseelementswhich regard
English only and to stress(instead)what holds for passivesin general]
(10)
La significationdesauxiliaresdu passifdiffdre d'une langued I'autre. [En
effetl en anglaischacundes auxiliairesdonne une nuancesp6cifiqueb la
7 The connectives
between brackets in (9) and (10) of course did not figur in the original text
but correspond to the corrected version. As f o r ( 1 1 ) , ( 1 l a ) i s p r o d u c e db y t h e G e r m a n s t u d e n t , ( 1 l b )
is the corrected version.
790
Beatrice Laniroy
phrase.Ils pr6sententdes caract6ristiques
diff6renteset ils ont [dEslors,
doncl leur propre domaine d'emploi. [Ndanmoins,cependant] ils font
partie du m€me domaines6mantiquepuisqu'ilsserventtous d former le
passifexprimantune action.[Par contre] en franEais,I'auxilairedu passif
sert d la fois ir exprimer l'6tat et I'action.
[The meaningof the passiveauxiliariesdiffers from one languageto the
other. (Indeed) in English every auxiliary adds a specialnuance to the
sentence.They have different propertiesand (therefore) their own use.
(However) they belongto the samesemanticdomain sincethey all allow
to form passivesthat expressan action.In French (however),the passive
auxiliary is used to expressa state as well as an action]
( 11 )
a.
b.
Le franEaisa beaucoup de grammaire. C'est une langue tr0s
difficile. Toutes les rdgles ont des exceptions.C'est pourquoi il
faut l'6tudierbeaucoupet commencertdt.
Le franEaisest une langue trds difficile [car] il possdde une
grammaire trds complexe. [En effet] chaque rdgle a des
exceptions.Il faut [donc] l'6tudier d fond pour l'apprendre, [par
cons6quentlil faut commencert6t
(French is a very difficult language (since) it has a complex
grammar.(Indeed) everyrule has its exceptions.(Therefore) one
has to study it thoroughlyin order to learn it, (so) it is necessary
to start early.
The secondtype of error occurswhen studentsplainly choosethe wrong connective.
Notice that the sameconnectivesthat causemost difficultiesto Dutch speakingstudents
also seem to be the most problematic ones for German and Danish students, as
reported in the two above mentioned papers. Cases in point are pourtant (yet),
d'ailleurslpar ailleurs (moreover, besides),en effetlenfait (indeed, in fact) and parce
quelpuisquelcar(because,since, as). The examplesin (12-13) illustrate the case of
pourtant. Although pourtant expressescontrast, it is not a mere synonym of mais or
cependant. As Moeschler & de Spengler(1982) argue,pourtant denies an implicit
conclusionthat can be drawn from the precedingclause,as in (12):
(12)
a.
b.
Il est gentil. Pourtant tout le monde le d6teste
[He is friendly. Yet everybodyhates him]
Bien qu'il soit gentil, tout le monde le d6teste
[Although he is friendly, everybodyhates him]
Such an implicit conclusion cannot be drawn from the first clause in (13a), from
Kielhiifer &. Poli (1991). Compare in this respect (13a) and (13c):
(13)
a.
L'anglaisest plus simple. Pourtant le frangaisest plus beau
[English is more simple. Yet French is nicer]
Pragmatic connectives
b.*
c.
d.
191
Bien que l'anglaissoit plus simple,le franEaisest plus beau
[Although English is more simple,French is nicer]
L'anglais est plus simple que le franEais.Pourtant les 6tudiants
ont autant de mal i apprendreI'anglaisque le franEais
[Englishis more simple than French. Yet studentshave as much
trouble in learningEnglishas French]
Bien que I'anglaissoit plus simple que le franEais,les 6tudiants
ont autant de mal d apprendrel'anglaisque le franEais
Example(i4), due to a Belgianstudent,illustratesa confusionbetweendifferent causal
Whereasthe use of car andpuisquehingeson the attitude of the speaker
connectives.
towardshis utterance,the use of parce que only refers to the propositional content
(Ducrotet al. 1980;.8The former involve a double speechact, the latter involvesonly
one.This is what makes (laa) ungrammatical;the second sentenceis equivalent to
(14b),in which car or puisquewould be totallyimpossible:
(14) a.
b.
Dans certainscas l'6chec est d0 au fait que les 6tudiants ne
maitrisent pas la matidre. Dans d'autres cas ils 6chouent car
[parce que] ils ne peuventpas tenir t€te au stressdes examens
[In certain casesthe failure is due to the fact that the studentsdo
not master the subject.[n other casesthey fail since (because)
they cannot stand the stressof the exams]
C'est parce qu'ils ne peuvent pas tenir t€te au stress qu'ils
6chouent
[It is becausethey cannot stand the stressthat they fail]
As both Nolke (1989) and Kielhofer & Poli (1991) report, d'ailleurs and en effet are
otherstumblingblocks tor foreign studentsof French. In example (15), taken from a
MA dissertationon connectivesby a Belgian student,d'ailleursgot confusedwith en
effet.Theformer brings in a new argumentwhich correctsor reinforcesthe preceding
one,whereasthe latter only confirmswhat is saidbefore.In (16), taken from Kielhofer
& Poli (1991),en effet and en fait are mixed up:
(15)
Un connecteurqui suit les deux seracomptabilis6dansles tableauxdans
la position m6diane.D'ailleurs [En effet] les phrasesreli6espar les deux
points sont plus fortement li6esque les phrasess6par6espar un point ou
un point-virgule.
[A connectivefollowing a colon will be countedin the figurs as being in
the intermediateposition.Moreover (indeed)sentencesthat are united
by a colon are linked more tightly than sentencesthat are separatedby
8 In fuct, parce que functions as the all-purpose connective of cause and can
replace car and
puisquein certain contexts.
I92
Beatrice Laniroy
a period or a semicolon]
(16)
Le franEaispossddeun grand patrimoine: en fait [en effet], c'est la
langue de la litt6rature, de la philosophie
[French has an important heritage:in fact (indeed) it is the languageof
the literature and of the philosophy]
The final and third type of error coversthe excessiveuse of two or three particular
connectives, viz. alors and quand mAme, less often tout de mAme. Although more
researchwould be needed to substantiatethis, the pervasiveuse of these particular
connectivesis maybe to be explained by interference with Dutch dan and toch
(German dann and doch), two common adverbial particles in Germanic languages.
Another reason,which in fact parallelsthe contrastiveexplanation,might be the fact
that alors and quand mdme also are typical of Belgian French. Those adverbs are not
uncommon in standardFrench, but seemto belong more readily than to spoken than
to written language:in order to be used in written discourse,alors and qttand mAme
have to meet specificsemanticand pragmaticconditionsthat are far more restrictive
than the conditons of use observedin the essaysof Belgian (German) students.The
examplein (17a)was producedby a Belgianstudent,(17b)is due to a German student
(Kielhofer & Poli 1991):alorsis superfluousin both cases.As for quand m4me in (18ain (18a),whereas
b), both examplesby Belgianstudents,the connectiveis unnecessary
it should be replaced in (18b) by mais (or toutefoisor cependant):
(17)
a.
b.
(18)
a.
b.
La rigueurm6thodologiqueestgrande:les trois constructionssont
trait6es s6par6mentet leurs attestationssont alors group6esen
fonction des temps et des modes
[The methodologicalrigour is impressive:the three constructions
are treated separatelyand their attentions are then grouped
accordingto their tenseand mode]
A c6t6 de l'utilit6 6conomiqueil y a alors le prestige social du
tranEais
[Besidesde economicutility there is then the social prestige of
Frenchl
Avant d'aborder les r6sultatsde nos observations,Eavaut quand
m6me la peine de faire une r6flexion pr6liminaire
[Before interpreting the results of our findings, it is still worth
while to make a preliminary observation]
Nous ne doutons pas que cette explication se justifie. Quand
m6me des statisquesd6montrent que le pourcentage d'agents
exprim6sne peut 6tre n6glig6
[We do not doubt that this explanation is right. Still statistics
prove that the percentageof expressedagentsis not unimportant]
Pragmaticconnectives
193
fu I have suggestedat the beginning of this paper, one of the reasonswhy foreign
studentsclf French encounter these difficultiesmay be the fact that connectivesare
intrinsically
a difficult issue,evenfor nativespeakers,so that the problem only increases
whenit has to be masteredby foreign languagestudents.
3. Connectivesand native speakersof French
That connectivesare per se a difficult matter is in fact stronglysuggestedby a survey
conducteda tew years ago at the Universitt Catholiquede Louvain, a French speaking
Belgianuniversity (Klein & Pierret 1990).eAbout 2000 first year students,native
speakersof French, from different departments,answereda languagetest bearing on
4 major topics, one of which was connectives.The other topics were: morphosyntax
(irregular verbs, choice of indicative vs subjunctive, etc.), grammatical analysis
(identificationof subjectand object,etc.) and lexicon.Thescoreof correct answersfor
was, surprisingly,only 50Va.
connectives
The purposeof the investigationwas to determinethe averagelevel of linguistic
competencerequired to deal with the scientifictype of languagethat students are
confrontedwith during their studies.In order to identify different types of language
deficit,separatelevels of linguisticcompetencewere distinguished,in particular the
sentencelevel and the discourselevel. The test consistedof 90 questions to be
answeredin 2 hours. Although it was a multiple choice test, "none of the indicated
solutionsis correct" was alwaysincluded as a possibleanswer.Hence studentscould
(andsometimeshad to) come up with a solution of their own.
The overall mean score of the test was 59 % of correct answers.When the
differentcomponentsof the test are taken into account, the figurs break down as
follows:84 % of correct answersfor grammaticalanalysis,66 % for morphosyntax,50
Vo for connectivesand 40 Vo for the lexicon. One of the interesting results of the
investigationis the following: when the correct answer did not appear among the
availablesolutions,so that the studenthad to produce it himself,the medium score of
the test fell down to only 18 % of correct answers.The latter data suggeststhat
arbitrarychoiceamong possibleanswersplays a much more important role than one
wouldexpect,on the one hand, and that it is far more difficult to produce the right
answerthan to chooseit, on the other.
With respectto connectives,the degreeof difficulty of the test varied from text
to text.The shortesttext fragment containedonly one connective,the longestfive. An
exampleis givenin (19):
9 As *ill appear from what follows, the experiment with
French native speakers only illustrates
one of the errors mentioned in the previous section, viz. choice of the wrong connective. In order to
establishwhether native speakers also have difficulties with the other types of error (especially type 1,
if the interference hypothesis for type 3 is correct), one should have to look at data stemming from
spontaneouslywritten texts by French native speakers. I do not know ofany research that has dealt with
the topic, but the issue is surely worthy of further research.
194
Beatrice Laniroy
(b)
G)
1. d'ailleurs
0
0
0
2. sauf si
0
3 . b i e n que
0
0
0
4. encore que
5 . q u o i que
0
0
6. car
0
X
7. c'est-)r-dire
0
0
0
8. contrairement d
0
9. au mOmedegr6 que
0
0
10. grAced
0
0
1 1 "e n v er t u de
0
0
12. comme
0
0
1 3 .a u cuner 6pons ene c o n v i e n t
la r6ponseexacteest pour
( 1 e a ). . "
[66.8%]
( 1 e b ). . .
[a7.zVo]
( 1 e c )...
[3L2o/o]
[C-score : I5.6Vo)
(c)
0
0
0
0
0
0
X
0
0
0
0
0
With respectto connectives,the main resultsof the test can be summarizedas follows"
First of all, the averagescore can be consideredas quite low, since only half of the
answers were correct. Furthermore, this figure refers to correct answers for the
connectivestaken individually.Indeed,when the combinationof connectivesthroughout
the text fragment is taken into account,the averagescoregoesdown to 21Vo of correct
answers.As shownfor example(19), the C(ombination)-scoreis much lower than each
of the individualscores:15.6% for the combination,against66.8 7o for (19a),47.2%o
for (19b) and 31.2 Vo for (19c).These resultsprovide evidencefor the hypothesisthat
I mentioned at the beginning of the paper, vtz. that connectivesare intrinsically a
difficult matter, even fbr native speakersof French. The reasonswhy this is so are
probably manifold. Firstly, connectivesare far more typical of written rather than of
spoken language.Secondly,they belong to a particular type of written discourse,viz.
argumentativetexts,which first year universitystudentsare not necessarily
very familiar
with. But there may be a further, lessobviousreason,which is in fact suggestedby a
more caretul analysisof the resultsof the test, as will be shown below.
A secondsurprisingconclusionto be drawn from the figurs is that there does
not seemto be a necessaryrelationshipbetweenthe number of connectivesto be found
in the text fragment and the scoreof correctanswers.For (19),where three connectives
are missing,15.6Vo of.the answerswere correct (for the combinationof connectives),
but in another case,with only two connectivesto fill in, the scorewas only 3.9 Vo.The
resultsof coursedepend on the degreeof difficultyof the text but also,surprisingly,on
the place where the connectivesare located in the text. In the above mentioned case
with two connectives,both were situatedat the beginningof the text, so that students
supposedlydid not concentrateupon the secondpart of the text which was. however,,
hagmatic
connectives
195
necessary
to find the correct answers.
Another striking result of the test is that studentssometimesoffer a correct
answerfrom a logical point of view, but without paying attention to the syntaxof the
remainingsentence.Some students,for example,proposedcomme in (19) instead of
car.The secondclause introducesan explanation,somethingthat can be expressed
indeedin French by comme. However, the comme clausehas to precede the main
clause,so comme could not possiblybe a correct answerhere.
A final result,which runs along similar lines,is that studentsseemto operate on
a short-termbasis when dealing with argumentativetexts. This is suggestedby the
significantdifferencebetween the C-scoreand the individual scores,as illustrated in
(19).Studentsoften propose a connectivewhich is compatiblewith the chunk of text
that immediatelyprecedesor follows,but that doesnot make any sensewhen the text
is takenas a whole. In other words,they tend to atomizethe text, being unawareof its
overalllogicalcohesion.They fail to producethe correct connectivesbicause they fail
to perceivethe general coherenceof the text. This aspectof the problem, which no
doubt brings in cognitive capacitiessuch as concentration,memory, etc. and which
wouldrequire much more research,may shed some new light on th" question why
connectives
are so often badly masteredby students,be they native speakersor not.
4. Connectivesin French and Dutch
In order to substantiatethe hypothesisthat French and Dutch may differ as to the
organization
of argumentativetexts, in particular with respect to connectives,I will
providetwo kinds of evidence.First, adverbialsand conjunctionsare traditionally
analyzed
as a closedclassof function words. However, the degreeof "closeness"may
varyfrom languageto language.Thus the classof French connectivesappearsto be far
moreproductivethan that of its Dutch counterparts.Following a functional principle "la fonctioncr6e I'organe"-, the fact that there are more connectivesin Fiench also
stronglysuggests
that they are used more.
The secondkind of evidencestems from a sample study of translationsfrom
Frenchtexts into Dutch. These have shown that it is possiblefor two languagesto
arguein differentways,either by usingexplicit connectivesor by a more impticii hnO
of argumentation,
e.g. by ordering the argumentsin a particular way, or by choosing
particularlexical items other than connectives.The former solution seems to be
preferredby French,the latter by Dutch.
In what follows,I will illustratethe two phenomenaby focusingon two particular
classes
of connectives,
those of contrast(in which I subsumeconceision)10
ind those
of consequence.
10 Connectives
of contrast and concession share a characteristic, as Moeschler & de Spengler
(1982)argue:they are part of an argumentative movement that puts forth an argument in favour of a
certain conclusion p and, at the same time, another, stronger argument in favour of the opposite
conclusionnon-p.
196
Beatrice Lamiroy
The trvo phenomenaof coursepoint towardsa typologicaldistinctionbetween
the two languages,which may also be responsablefor some of the "errors" made by
Dutch speakingstudentswhen writing French. When a languageusesthe ordering of
argumentsin a text insteadof usingconnectives,this may lead to a lack of connectives
used in the L2 of speakersof thoselanguages.
Thus the typologicaldifferencebetween
Dutch and French points to an explanationfor error type 1 in a straightforwardway.
If the excessiveuse of particular connectivessuch as alors or quand m4me is to be
accountedfor, as I suggestedbefore,by interferencewith the Germanic particles(and
maybe by the local variant of French spoken in Belgium), the questionremains as to
whether error type 2 (wrong connective)can also be explained by the typological
hypothesis.I will come back to this point later.
4.1. That Frenchll connectives,belonging to the two categoriesof adverbials and
conjunctions(coordinationand subordination),are more productivethan their Dutch
counterpartsr2is illustrated in (20) and (21). The examplesin (20) expresscontrast
:
, those in (21) consequence
(20)
a.
b.
(2r)
Ihowever]
cependant
n'emp€che
n6anmoins
pourtant
quand m6me
tout de m€me
toutefois
[whereas]
alors que
pendant que
tandisque
[so that]
d celau point que
d (un) tel point que
d un point tel que
d tel titre que
d telle preuveque
toch, echter
toch, desondanks
toch, (des(al)niettemin
toch, nochtans
toch (des)niettegenstaande
toch
toch, evenwel
terwijl
terwijl
terwijl
zo(zeer) dat
zo(zeer) dat
zo(zeer) dat
zo(zeer) dat
zo(zeer) dat
11 Thir list is based on Ruquet &
Quoi-Boclin (1988), a textbook with the revealing title
"Comment raisonner i la franEaise' (How to reason in the French way), and on theoretical work done
by Gross and his team, in particular Gross (1986) and Piot (1978, 1988).
12 For Dutch, the data where gathered in various grammars (ANS
1984, Smedts & Van Belle
1993), dictionaries (Van Dale 1990, 1992) and from judgments by native speakers of Dutch.
Pragmaticconnectives
I9l
d tell enseigneque
zo(zeer) dat
de faEon/manidre(d ce) que
zo(danig) dat
que zo(danig) dat, op zo'n manier
de telle faEon/manidre/sorte
Cat
d'une telle faEon/maniOre
te l l e ' q u e
zo(danig)dat, op zo'n manier
dat
de/en sorte que
zodat
si bien que
zodat
tant et si bien que
zodat
tant (et tant) que
zodanigdat
tantilyaque
zodanigdat
tellementque
zodanigdat
tellementbien que
zodanigdat
With respectto the French connectiveslisted in (20), it shouldbe noted that although
all of them can be translatedby Dutch toclt, thrsdoes not amount to sayingthat they
areall semanticallyor stylisticallyequivalent.As shownin (13),pourtant and cependant
arenot alwaysinterchangeable.
And whereastouteforsrather belongsto formal speech,
quandmAmeis, as mentioned before, rather colloquial. A similar remark holds for
severalitems listed in (21), e.g.qu point que (informal) vs d telle enseigneque (formal)"
Althoughthe typologicalhypothesismore readilyaccountsfor error type 1 than for the
othertypes,it seemsfair to hypothesizethat the more connectivesthere are (and thus
the more variegated)in a language,the greater the difficulty for L2 speakersof that
languageto master the sometimes subtle semantic and/or pragmatic differences
betweenthem and hence, to chooseadequately among them. Thus the typological
betweenFrench and Dutch may alsoaccount,to a certain extent at least,for
difference
errortype 2.
Two further remarks are in order here. First, in addition to adverbs and
conjunctions,
French also has a series of idiomatic expressionsfor which Dutch
equivalents
are not alwayseasyto find:
(22)
a.
b.
dan nog
encore faut-il que
[it. still need it that]
[still]
toujours est-il que
dit neemt niet weg dat
[it. always is it that]
flit.thisdoesnot take awaythat]
desalniettemin
il n'empdcheque
[it. it does not prevail that]
il n'en reste/demeurepas
dit neemt niet weg dat
molns que
[it.it staysnot lessthat]
il n'en est pas moins vrai
que
zo veel is zeker dat
[it. it is not lesstrue that]
198
Beatrice Laniroy
It is alsoworth noting that certainconnectiveswhich are still usedin the Southernpart
of the Dutch speakingcommunity(Belgium),tend to be considereclas outdated in the
Northern part (Holland). Casesin point are nochtarrs(yet), vermits(since) and to a
lesserextent,daar (since).As the norm for standardDutch is determinedby the North
rather than by the South,theseconnectivesmight eventuallydisappear.The latter fact
is of course in contrastwith the high productivityof French connectives,as illustrated
i n (2 0 -21) .
4.2. The second kind of evidencefor the typologicalhypothesisstems from a sample
study of translationsfrom French into Dutch. Although much more researchwill be
needed to substantiatethis, the explorationI have undertakenso far has disclosedtwo
phenomena: on the one hand, a certain amount of French connectives simply
disappear in the Dutch translation, as in (23), and on the other hand, certain
connectivesare replacedby different lexicalor syntacticmeans,as in (24) :
(23)
a.
b.
(24)
a.
b.
En 18f18un savantanglais6crivait: "Les femmes non seulement
ne sont pas la race, elles ne sont pas m€me la moiti6 de la race
mais une sous-espdce
destin6euniquementd la reproduction"(S.
De Beauvoir,Le DeuxidmeSexe)
[In 1888 an English scholar wrote: "Women not only do not
representthe race, they do not even representhalf of it but they
are an under-species
whose unique tunction is to procreate]
In 1888 schrijft een Engels geleerde: "De vrouwen
vertegenwoordigen
het ras niet; ze zijn zelfsniet eensde helft van
het ras, ze zijn een onder-soort, alleen bestemd voor de
voortplanting"
Car penser rdellementh quelqu'un,c'est y penser minute apr0s
minute, sans6tre distrait par rien, ni les soins du m6nage,ni la
mouchequi vole, ni les repas,ni une d6mangeaison.
Mais il y a
toujoursdesmoucheset desd6mangeaisons.
C,estpourquoi la vie
est difficile ) vivre (A. Camus,La Peste)
[since to think really of somebodymeansto think of him minute
after minute, without being distractedby anything, neither the
worries of the housekeepingnor the fly that is flying around or
the mealsnor an itch. But there are alwaysflies and itches.That's
why lit-eis difficultl
Want werkelijk aan iemand denken, ieder minuut aan hem
denken, zonder zich door iets te laten afleiden, noch door
huishoudelijkezorgen,noch door een vlieg of door maaltijden of
door jeuk. Maar er zijn altijd maaltijden en men krijgt wel eens
jeuk. Dat maakt het leven zo moeilijk.
To give an idea of the figures:in S. De Beauvoir'snovel, there are 1168instancesof
hagmaticconnectives
799
connectives
of contrast,in the Dutch translationthere are only 1027.In Camus'snovel,
thereare 340 instancesof connectivesof consequence,
in the Dutch translationthere
are only 289. These figures are all the more revealingsincetranslationsby definition
tendto respectthe original version.Hence, the amount of Dutch connectivesused in
spontaneous
text may be estimatedto be even lower.
5. Conclusion
I havetried in this paper to gain insightinto the questionwhy Dutch speakingstudents
of French have serious difficulties to use French connectives.while thev have
in their native language.
connectives
At least three quite different factors seem to be at stake here. First, from a
purelylinguisticpoint of view, connectivesare complexelementsbecausethey function
at the crossroad of syntax, semantics and pragmatics.Second, the investigation
conductedat the UCL with French native speakerssuggeststhat a good performance
in thisdomainhingesnot only on linguisticknowledge,but also on particular cognitive
skillsthat are equallyimportant for nativespeakersof the language.An important issue
hereseemsto be the capacityof concentratingon the text as a whole and of perceiving
its overallcoherence.And finally, from a typologicalpoint of view, I have adventured
the hypothesisthat French and Dutch do not make equal use of logical connectives.
Frenchseemsto be better equipped in this respect than Dutch, which is in itself
significant
from a functional perspective.Furthermore,translationstudiesshow that a
certainamount of French connectivestend to be dropped by Dutch translators (or
renderedby somethingelse than a connective).Of course,if my hypothesisis correct,
it will be necessaryto searchfor more evidence,and for a better understandingof the
givenevidence,in order to set up a teachingstrategy.At any rate, the questionseems
importantenough to be taken seriously,not only in the field of foreign language
but also,as I mentionedat the beginning,in other domainsof both theoretical
teaching,
andappliedlinguistics.
Re[erences
Al, B.P.F. et al. (1990) Van Dale. Groot Woordenboek Frans-Nederlands.
Anscombre,J.-C. & O. Ducrot (197'7) Deux Mais en frangais. Lingua 43:23-40.
Auchlin, A. & A. T,enone (1980) Cnnversations, actions, actes de langage: Eldments d'un systdme
d'analyse.Cahiers de linguistiquefrangaise I: 6-41.
Berendonner,A. (1981) Eldntents de pragnatique linguistiqr.re.Paris: Minuit.
Bieber, K. (1988) Theoretische Ueberlegungenzur Klassifikation von Konnektoren im Franzdsischen. Bern:
P. hng.
200
Beatrice Lantiroy
Blakemore, D. (1989) Denial and contrast: A relevance theoretical analysis of but" Linguistics and
Philosophy 12: 15-37.
Blumenthal, P. (1980) La Syntaxedu Message.Application au frangais ntodeme. Ti.ibingen: Niemeyer.
Ducrot, O. et al. (1980) Les mots du discours. Paris: Minuit.
Foolen, A. (1993) De betekenis van partikels. Een dokumentatie van de stand van het onderzoek met
bijzondere aandacht voor nraar. Doctoraal Proefschrift, Universiteit Nijmegen.
Geerts, G., W. Haeserijn, J. De Rooij & M. Van Den Toorn (1984) AlegemeneNederlandseSpraakhtnst"
Groningen: Wolters.
Geerts, G. & H. Heestermans (1992) Van Dale. Groot Woordenboekder Nederlandse Taal. Utrecht: Van
Dale lrxicografie.
Gross, M. (1986) Grammaire transfomtationnelle du frangais. La syntaxede l'adverbe. Paris: Cantildne.
Giilich, E. & T. Kotschi (1983) lrs marqueurs de la reformulation paraphrastique. Cahiers de linguistique
frnnEaise 5: 305-351.
Harris, Z. (1968) Mathentatical Stntcruresof Language. New York: Wiley.
Kielhoefer, B. & M. Poli (1991) L'anglais est plus simple, pourtant le franEais est plus beau.
Neusprachliche Mitteilungen aus Wissenschaftund haxis 45.2: 100-110.
Klein, J. & J.-M. Pierret (1990) Une enqudte sur la pratique du frangaisdes 6tudiants en premidre annde
A I'Universit€. Enjeux 2l: l-7.
Lamiroy, B. & P. Swiggers (1992) The status of imperatives as discourse signals. In S. Fleischmann &
L. Waugh (eds.), Discourse-hagnntics and the Verb. I-nndon: Routledge, pp. 120-146.
Lundquist, L. (1987) Programme argumentatif et strat6giesde d6sambiguisationr6fdrentielle. Revue
Romane 22: 163-182.
Mainguenau. D. (1987) Nouvelles en analyse du discours. Paris: Hachette.
Moeschler, J. (1985) Argunrcntation et conversation:Eldmentspour une analysepragmatique du discours.
Paris: Hatier.
Moeschler, J. & N. De Spengler (1982) la concession ou la r6futation interdite, approches
argumentative et conversationnelle. Cahiers de linguistiqueftangaise 4:7-37.
Morel, M.-A. (1984) Caractdres syntaxiquesdistinctifs de deux types de concession en franEais moderne.
In P. Valentin (ed.), L'expressionde la concession,pp.4I-57.
Nolke, H. (1989) Contrastive pragmatic linguistics. In K.M. l-auridsen & O. kuridsen
Linguistics. Aarhus: The Aarhus School of Business, pp. 19-233.
(eds.), Conrrastive
Piot, M. (1978) Etudes transformationnellesde quelques classesde conjonctions de subordination du
franEais. Thdse de doctorat du troisidme cycle. LADL-Paris VIL
Pragmaticconnectives
201
Piot, M. (1988) Conjonctions de subordination et probldmes de classification. In C. Blanche-Benvenisre
et al. (eds.), Grammaire et histoire de ld grammaire. Hommage d Ia mdmoire de Jean Stefanini. Aix-enProvence:Publications de I'Universitd de Provence, pp.335-352.
Roulet, E. et al. (1985) L'aniculation
du discours en frangais contemporain. Bern: P. [:ng.
Rubattel, C. (1982) De la syntaxe des connecteurs pragmatiques. Cahiers de linguistiquefrangaise 4:3761.
Ruquet, M. et J.-L. Bodin (1988) Raisonner d la frangaise. Paris: Cld.
Smedts,W. & W. Van Belle (1993) Taalboek Nederlands. Kapellen: DNB-Pelckmans.
Sperber,D. & D. Wilson (1986) Relevance:Communication and cognition. Oxford: Blackwell.
Spooren, W. (1989) Some aspects of the form and interpretation of global contrastive coherence
relations.Doctoraal Proefschrift, Universiteit Nijmegen.
Van Belle, W. (1989) Want, omdat en aangezien.Leuvense Bijdragen 78: 435-456.
Van Belle, W. & G. De Vrooij (1992) Tegenstellende en toegevende connectoren. Een argumentative
beschrijving.K.U.[ruven: reprint 143.
7-ezone,A. (1982) l,a cons6cution sans contradiction: donc, par consdquent, alors, ainsi, aussi (premidre
partie). Cahiers de linguistique frangaise 4: 107-141.
7.enone,A.(1983) la cons6cution sanscontradicticln: Donc, par consdquent, alors, ainsi, aussi (deuxidme
partie). Cahiers de linguistique frangaise 5: 189-214.