Professional Documents
Culture Documents
Hansen & Visconti - The Diachrony of Negation-John Benjamins Publishing Company (2014)
Hansen & Visconti - The Diachrony of Negation-John Benjamins Publishing Company (2014)
Editors
Werner Abraham Elly van Gelderen
University of Vienna / Arizona State University
University of Munich
Editorial Board
Bernard Comrie Christian Lehmann
Max Planck Institute, Leipzig University of Erfurt
and University of California, Santa Barbara
Marianne Mithun
William Croft University of California, Santa Barbara
University of New Mexico
Heiko Narrog
Östen Dahl Tohuku University
University of Stockholm
Johanna L. Wood
Gerrit J. Dimmendaal University of Aarhus
University of Cologne
Debra Ziegeler
Ekkehard König University of Paris III
Free University of Berlin
Volume 160
The Diachrony of Negation
Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti
The Diachrony of Negation
Edited by
Jacqueline Visconti
University of Genoa
Despite intensive research, negation remains elusive in many ways. Its expression
across languages, its underlying cognitive mechanisms, its development across
time, and related phenomena, such as negative polarity and negative concord,
leave many unresolved issues of both a definitional and a substantive nature. The
importance of works on negation for recent developments of linguistic theory is
reflected in an ever increasing interest in the topic. As well as monographs and
collections addressing negation from a synchronic point of view, such as Horn
(2010), Penka (2011), and de Swart (2010), the last few years have seen the publi-
cation of a number of volumes focussing on the diachrony of negation from dif-
ferent theoretical angles, viz. Jäger (2008), Larrivée & Ingham (2011), Roberts &
Roussou (2003), and Willis et al. (2013). In addition, this brief list leaves out a
significant number of workshops and lone-standing papers published in journals
and edited volumes with broader overarching topics, such as van Gelderen (2009)
on linguistic cycles as an important model of language change, an idea further
explored in van Gelderen (2011).
The idea of a cycle, and in particular the so-called Jespersen Cycle, is in
various ways central to much current work on the diachrony of negation, incl.
the studies mentioned above and several of the chapters in the present vol-
ume. The Jespersen Cycle (or “Jespersen’s Cycle”, cf. Dahl 1979: 88) is short-
hand for the well-known phenomenon that, across a number of languages, the
historical evolution of standard clause negation1 seems to proceed in a cyclical
. By “standard clause negation” we refer to the most common kind of negation marking
found in a given language, a form of marking which is used productively in declarative
main clauses, among other environments, and which does not involve quantifiers (cf. Payne
1985: 198).
Maj-Britt Mosegaard Hansen & Jacqueline Visconti
f ashion, along largely similar lines. Jespersen (1917: 4) described this evolution
as follows:
The original negative adverb is first weakened, then found insufficient and
therefore strengthened, generally through some additional word, and this in turn
may be felt as a negative proper and may then in the course of time be subject to
the same development as the original word.
French, which is the object of the contributions in chs. 7–9 of this volume, is often
adduced as an illustration, as in Table 1 below:2
Table 1. The evolution of French clause negation (sample sentence: ‘I do not say…’)
Stage 0. [Classical Latin] non dico The negator is preverbal
. There are a few things to note concerning Table 1: First, as column 1 specifies, Stages 0
and 6 of course do not represent French as such, nor is there any necessity for French to
ever develop to the hypothesized Stage 5 (let alone Stage 6). While certain dialects (notably
Québécois) have all but eliminated preverbal ne in conversational speech (Sankoff & Vincent
1977, Fonseca-Greber 2007), the marker continues to be normatively used in writing in those
dialects, and it is used with greater frequency in the spontaneous speech of native speakers
from France (Ashby 2001). Indeed, it has been suggested (Fonseca-Greber 2007) that, in Swiss
French at least, preverbal ne may be in the process of acquiring a new pragmatic function. If
that is so, and if the function in questions spreads to other dialects, Stage 5 may never become
instantiated. In other words, Stages 0, 5, and 6 are included merely for the purpose of showing
what a complete Jespersen Cycle might look like. Secondly, Stages 2 and 4, i.e. the stages that
are characterized by variability in the realization of negation, are not recognized as separate
stages by all researchers, but are included in Stages 3 and 5, respectively, thus yielding a three-
stage development for French: I: ne > II: ne…pas > III: pas.
The diachrony of negation
the “weakening” of the original, preverbal, negative marker. Against this hypoth-
esis, Kiparsky and Condoravdi (2006: 4) observe that cross-linguistic evidence for
it is not strong, and indeed, other scholars, starting with Meillet (1912: 140), have
suggested that the addition of a new postverbal marker may more plausibly be
triggered by pragmatic factors.
Secondly, typological studies have suggested that there is not just one, but sev-
eral, Jespersen Cycles (van der Auwera 2009), as languages may add not just one,
but two, and occasionally even three postverbal markers. In other words, rather
than Stage 4 being the logical next step after Stage 3, alternative developments are
possible.
Thirdly, a related, but broader, typological question concerns the degree to
which European languages, where the Jespersen Cycle has so far been most amply
attested, are representative of the languages of the world in respect of the evolution
of negative marking. The answer to this latter question is of importance not least
to the issue of reconstruction, for while the Jespersen Cycle appears in principle to
be an eminently useful basis for reconstruction of earlier stages of languages with
few or no textual records of any significant time-depth, it is not clear that it is cur-
rently sufficiently understood and sufficiently well-attested throughout the world
to serve this purpose.
Fourthly, even those languages where the empirical reality of the basic pat-
tern posited by Jespersen has never been in doubt show substantial variation in
the speed with which they progress through the cycle, and the degree to which its
different stages are clearly delimited in time. Thus, languages may remain stable at
a given stage of the Cycle (often Stage 3, where negative marking is bipartite and
embraces the verb) for very long periods of time. In addition, languages may not
just go through stages of simple variation such as Stages 2 and 4, where bipartite
negation alternates with either pre- or postverbal negation, but stages of complex
variation may also be instantiated, where preverbal, embracing, and postverbal
negation are found simultaneously. Because, as mentioned above, the typologi-
cal representativity of the Jespersen Cycle has yet to be firmly established, such
languages raise the issue of whether the evolution of negation is unidirectional or
whether it might be possible for postverbal negative markers in embracing con-
structions to be lost.
The bulk of research on the history of negation having traditionally focused
on European and Mediterranean languages (including, most recently, Willis et al.
2013 and Larrivée & Ingham 2011), a first important element of relative novelty
offered by this volume is the variety of languages studied (although a number of
them are also adduced – albeit in lesser depth – in van Gelderen 2011: Chapter 8).
Thus, the first five chapters offer a mix of large-scale typological surveys and
Maj-Britt Mosegaard Hansen & Jacqueline Visconti
bipartite clausal negation, nor for preverbal negation. Moreover, against frequent
claims in the literature, the survey reveals that the strategy of expressing clausal
negation only through inherently negative pronouns or adverbs is not only found in
Europe, but also in the Americas, and that it is also found in verb-initial languages.
Quantifier negation raises the further issue of the possible existence of a Quan-
tifier Cycle (Ladusaw 1993) bearing some resemblance, and possibly related, to the
Jespersen Cycle. Such a cycle, if real, and its parallelisms with the Jespersen Cycle,
could be illustrated as in Table 2 below, again using French as the test language.
There are two central questions here: The first is whether system-level (or
“macro-parametric”) change is involved, such that the development of all quanti-
fiers used in negative or negative-polarity environments in a given language is
triggered in the same way and follows a largely identical diachronic trajectory, or
whether the evolution of such quantifiers may be determined more by individual,
more or less idiosyncratic, (or “micro-parametric”, cf. Déprez 2011) factors per-
taining, for instance, to their source meaning or the particular part of speech from
which they originate, than by the existence of a Quantifier Cycle as such.
The second question is whether the diachronic development from polarity
neutral item > NPI > n-word3/negative indefinite, which in and of itself is uncon-
troversially attested in many cases, is (at least partly) unidirectional or whether
indefinites are rather subject to what Jäger (2010) calls a “random walk”, such that
items may move freely back and forth between polarity types and/or between
. N-words (Laka Mugarza 1990: 107) are quantifiers which share properties of both nega-
tive indefinites and negative polarity items (NPIs). Thus, n-words can express negation on
their own, without an accompanying marker of clausal negation, but co-occurrence of two or
more such markers will by default be interpreted as expressing a single negation (e.g. French
Je [ne] vois plus personne “I don’t see anybody anymore/I see no-one anymore’).
Maj-Britt Mosegaard Hansen & Jacqueline Visconti
In sum, our volume shows the importance both of large-scale typological stud-
ies and of fine-grained studies of individual negative markers, their different dia-
chronic sources and trajectories, as well as differences in the pace of change.
One peculiarity worth mentioning, however, is the use of the term “double
(clausal) negation” in Chapters 2 and 3 to refer to what others term bipartite, or
embracing, negation (e.g. French Je ne le vois pas “I don’t see it”). It is worth not-
ing that the term “double negation”, besides referring to the use of two negation
morphemes to express one semantic negation, as in the chapters mentioned and,
for instance, WALS (Dryer & Haspelmath 2013), is frequently used elsewhere to
designate the relation of mutual cancelation between two negation morphemes
resulting in a positive sentence (e.g. Zeijlstra 2004: 58–50; 261–263). Double nega-
tion in this latter sense is opposed to “negative concord”, i.e. the phenomenon
where two (or multiple) negatives appear without canceling each other out. Nega-
tive concord is known, alternatively, as “multiple negation”, “negative doubling”
or “negative spread”.
The first two chapters of the volume take a broad typological perspective on
the expression and evolution of negation, focusing on a range of non-European
languages.
In Chapter 2, “On the relation between double clausal negation and nega-
tive concord”, Lauren Van Alsenoy and Johan van der Auwera study the relation
between double clausal negation (as exemplified by French ne … pas ‘not’) and
negative concord (as exemplified by French ne … personne ‘nobody/not…any-
body’). Using a database of 179 languages from Asia, Africa and the Americas,
the authors test the proposals by Zeijlstra (2004) and de Swart (2010) that nega-
tive concord may be a necessary condition for double clausal negation, and they
conclude that although French shows that the two phenomena can be related, they
coincide only rarely. The subsequent discussion focuses on Ewe and Karok, lan-
guages in which the two phenomena interact, but in a non-French way, and on
seven languages with negative concord, but no double negation, thus sketching a
battery of parameters of variation.
Chapter 3, by Frens Vossen and Johan van der Auwera, discusses “The
Jespersen cycles seen from Austronesian”. The authors set out to collect and
evaluate evidence for what is known as the ‘Jespersen cycle’ in the Austrone-
sian language family, their database containing information on the verbal main
clause negation of 409 Austronesian languages. It is shown, mostly on the basis
of the synchronic comparison of related languages, that the Jespersen cycle
manifests itself in that family too, most clearly so in the languages of Vanuatu,
but that some languages take the cycle further by using triple, or even qua-
druple, negation.
Maj-Britt Mosegaard Hansen & Jacqueline Visconti
The following chapters take a closer look at aspects of the evolution of nega-
tion in three less-studied, also non-European, languages.
In Chapter 4, “The development of Standard Negation in Quechua”, Edith
Pineda-Bernuy shows that, across the several languages and varieties of this lan-
guage family, three different patterns of standard negation can be discerned: sin-
gle, preverbal negation with mana, bipartite, embracing negation with mana(-m)
… -chu and single, postverbal negation with -chu. These three patterns exist along-
side a large and varied number of non-standard patterns, all of which are carefully
described. The author subsequently discusses three possible paths of development,
arguing that the most plausible one takes the form of a three-stage evolution:
mana > mana… -chu > -chu. While this appears compatible with the definition of
the Jespersen Cycle, Pineda-Bernuy emphasizes that phonetic erosion of mana is
not a factor.
Chapter 5, “Taiwanese Southern Min V2 negation: a historical perspective”,
by Hui-Ling Yang, provides a historical account of some issues which have not
previously been fully addressed in connection with the Taiwanese Southern
Min V bo XP construction. One issue is an ambiguity over two interpretations,
while another involves the presence or absence of post-verbal negation. Assum-
ing that the diachrony of bo is parallel to that of Mandarin mei, and using inner
aspect as her theoretical framework, Yang analyzes Southern Min negative bo
as originating in the V and as having been reanalyzed as an aspect marker in
V bo XP. Other than the diachrony of Chinese negation, previous treatments
of one particular type of V2 negation are also reviewed. Contra Huang (2009),
who analyzes the semantics of bo in V bo DP as lexically determined by the
abstractness of the nominal phrase following bo, Chapter 5 argues that bo in this
construction can occupy two positions: while the aspectual head bo yields an
episodic interpretation in the nominal argument, bo in a higher head gives rise
to genericity.
Rounding off this section of the book, Vermondo Brugnatelli gives an over-
view of “Berber negation in diachrony” in Chapter 6. The morphosyntax of nega-
tion in Berber is rich and complex, and appears to be the outcome of multiple
processes that have taken place over different time-periods from prehistory to
the present day. The most noteworthy issue is the tendency towards a redundant
marking of negation, not only by means of discontinuous morphemes (circum
fixes) but also through the use of special “negative verb stems” – a feature that is
attested in nearly all of the Berber-speaking area, regardless of the type of negative
affixes in use. The author attempts to single out the main processes that have led
to the present stage, taking into account the etymologies of prefixal and suffixal
negative morphemes, the origin of negative stems and the role of the Jespersen’s
cycle in the evolution of Berber negation.
The diachrony of negation
The three final chapters provide new perspectives on the evolution of negative
expressions in French, one of the most widely studied of languages in this respect,
further demonstrating the richness and complexity of the diachronic development
of negation.
In Chapter 7, Maj-Britt Mosegaard Hansen explores “The grammaticaliza-
tion of negative indefinites: the case of the temporal/aspectual n-words plus and
mais in Medieval French”. The author argues that while the contemporary French
n-words have undergone a process of paradigmaticization, in the sense that the
language has evolved from making available a relatively open set of quantifying
expressions, many of which could be used independently of polarity, to offering
speakers a small closed set of items which are syntactically confined to contexts of
negative polarity, individual members of this functional paradigm appear to have
followed different paths of evolution. Thus, while the evolution of standard clause
negation in French is well-established as an instantiation of the Jespersen’s Cycle,
there is no parallel unidirectional macro-development of quantifying negators, i.e.
no Quantifier Cycle, in that language. The paper is concluded by a discussion of
the status of functional paradigms and the role of paradigmatic pressure as a factor
in language change.
Further contributing to the elucidation of how indefinites evolve, Richard
Ingham and Amel Kallel adduce “Evidence from a correspondence corpus for dia-
chronic change in French indefinites 1450–1715” in Chapter 8. Data from personal
letters is interpreted using Haspelmath’s (1997) semantic map of indefinites, and
it is shown that the re-categorisation of quelque (‘some’) as an ordinary positive
indefinite is associated with a major change in the use of the hitherto all-purpose
indefinite aucun (‘some’ > ‘any’/’no(ne)’), leading the latter to become an n-word in
Modern French. Like Hansen’s paper, Ingham & Kallel’s analysis contributes to an
account of the evolution of syntactic negation that emphasizes micro-parametric
factors involving the properties of individual lexical items, rather than a macro-
parametric change affecting the structural representation of negation.
Pierre Larrivée’s investigation of “The continuity of the vernacular. The evolu-
tion of negative doubling in French” in Chapter 9 completes the volume. Larrivée’s
aim is to assess Labov’s claim that the vernacular is the most stable variety of a
language, using geographically distant varieties of French as a test case. He tracks
the co-occurrence of a clausal negator with an n-word (e.g. ‘I didn’t do nothing’
with a single-negation interpretation, i.e. ‘I didn’t do anything’) over a period of
six centuries. The quantitative study of negative doubling in Quebec and France
historical and contemporary vernacular sources demonstrates overall stability for
France from the 14th century to the present. That negative doubling is ten times
more frequent in contemporary vernacular Quebec French than in contemporary
vernacular France French may be due to difficulty in accessing the vernacular
Maj-Britt Mosegaard Hansen & Jacqueline Visconti
References
Ashby, William J. 2001. Un nouveau regard sur la chute du ne en français parlé tourangeau:
s’agit-il d’un changement en cours? Journal of French Language Studies 11: 1–22. DOI:
10.1017/S0959269501000114
Dahl, Östen. 1979. Typology of sentence negation. Linguistics 17: 79–106. DOI: 10.1515/ling.1979
.17.1-2.79
Déprez, Viviane. 2011. Atoms of negation: An outside-in micro-parametric approach to nega-
tive concord. In Larrivée & Ingham (eds), 221–272.
Dryer, Matthew. 1998. Universals of negative position. In Studies in Syntactic Typology [Typo-
logical Studies in Language 17], Michael Hammond, Edith A. Moravcsik & Jessica Worth
(eds), 93–124. Amsterdam: John Benjamins.
Dryer, Matthew S. & Haspelmath, Martin (eds). 2013. The World Atlas of Language Structures
Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. 〈http://wals.info〉
Fonseca-Greber, Bonnibeth Beale. 2007. The emergence of emphatic ne in conversational Swiss
French. Journal of French Language Studies 17: 249–275. DOI: 10.1017/S0959269507002992
van Gelderen, Elly (ed.) 2009. Cyclical Change [Linguistik Aktuell/Linguistics Today 146].
Amsterdam: John Benjamins. DOI: 10.1075/la.146
van Gelderen, Elly. 2011. The Linguistic Cycle. Language Change and the Language Faculty.
Oxford: OUP.
Hansen, Maj-Britt Mosegaard. 2011. Viviane Déprez: “Atoms of negation. An outside-in micro-
parametric approach to negative concord.” Discussion. In Larrivée & Ingham (eds),
273–283.
Harris, Martin. 1978. The Evolution of French Syntax: A Comparative Approach. London:
Longman.
Haspelmath, Martin. 1997. Indefinite Pronouns. Oxford: Clarendon.
Horn, Laurence R. 1989. A Natural History of Negation. Chicago IL: University of Chicago Press.
Horn, Laurence R. (ed). 2010. The Expression of Negation. Berlin: Mouton de Gruyter. DOI:
10.1515/9783110219302
Huang, Hui-Yu. 2009. The Reinvestigation of the So-Called Taiwanese Postverbal Negation
[V u/bo 'have/not-have' NP]. MA-dissertation, National Tsing Hua University, Taiwan.
Jäger, Agnes. 2008. History of German Negation [Linguistik Aktuell/Linguistics Today 118].
Amsterdam: John Benjamins. DOI: 10.1075/la.118
Jäger, Agnes. 2010. Anything is nothing is something. On the diachrony of polarity types of
indefinites. Natural Language and Linguistic Theory 28: 787–822. DOI: 10.1007/s11049
-010-9113-1
Jespersen, Otto. 1917. Negation in English and Other Languages. Copenhagen: Høst & Søn.
Kiparsky, Paul & Condoravdi, Cleo. 2006. In Proceedings of the 2nd International Conference of
Modern Greek Dialects and Linguistic Theory, Mark Janse, Brian D. Joseph & Angela Ralli
(eds). Mytilene: Doukas. 〈http://www.stanford.edu/~kiparsky/Papers/lesvosnegation.pdf〉
The diachrony of negation
Ladusaw, William A. 1993. Negation, indefinites, and the Jespersen cycle. In Proceedings of the
19th Annual Meeting of the Berkeley Linguistics Society. General Session and Parasession
on Semantic Typology and Semantic Universals, Joshua S. Guenter, Barbara A. Kaiser &
Cheryl C. Zoll (eds), 437–446. Berkeley, CA: Berkeley Linguistics Society.
Laka Mugarza, Miren Itziar. 1990. Negation in Syntax: On the Nature of Functional Categories
and Projections. Ph.D. dissertation, MIT.
Larrivée, Pierre & Ingham, Richard P. (eds). 2011. The Evolution of Negation. Beyond the J espersen
Cycle. Berlin: Mouton De Gruyter. DOI: 10.1515/9783110238617
Meillet, Antoine. 1912. L’evolution des formes grammaticales. Scientia 12(6): 384–400.
Miestamo, Matti. 2005. Standard Negation. The Negation of Declarative Verbal Main Clauses in a
Typological Perspective. Berlin: Mouton de Gruyter.
Payne, John. 1985. Negation. In Language Typology and Syntactic Description, Vol. 1, Timothy
Shopen (ed.), 197–242. Cambridge: CUP.
Penka, Doris. 2011. Negative Indefinites. Oxford: OUP.
Roberts, Ian & Roussou, Anna. 2003. Syntactic Change: A Minimalist Approach to Grammatical-
ization. Cambridge: CUP. DOI: 10.1017/CBO9780511486326
Sankoff, Gillian & Vincent, Diane. 1977. L’emploi productif du ne dans le français parlé à
Montréal. Langue française 34: 81–109.
de Swart, Henriëtte. 2010. Expression and Interpretation of Negation. An OT Typology. Dordrecht:
Springer. DOI: 10.1007/978-90-481-3162-4
van der Auwera, Johan. 2009. The Jespersen cycles. In van Gelderen (ed.), 35–71.
Vennemann, Theo. 1974. Topics, subjects and word order: From SXV to SVX via TVX. In Inter-
national Conference on Historical Linguistics I, John M. Anderson & Charles Jones (eds),
339–376. Amsterdam: North-Holland.
Willis, David, Lucas, Christopher & Breithbart, Anne (eds). 2013. The History of Negation in the
Languages of Europe and the Mediterranean, Vol. I. Oxford: OUP.
Zanuttini, Raffaella. 1997. Clause Structure and Negation. Oxford: OUP.
Zeijlstra, Hedde. 2004. Sentential Negation and Negative Concord. PhD-thesis, University of
Amsterdam.
On the relation between double clausal
negation and negative concord*
Negative concord and double clausal negation are separate phenomena, which
show some striking similarities, the main similarity being the fact that negation is
formally expressed more than once, but semantically only once. These similarities
have led scholars to hypothesize on whether and how these two are related. In
this chapter some issues in the diachrony of both processes are discussed. On the
basis of a survey of 103 non-European languages it is concluded that NC cannot
be considered to be a necessary condition for DN nor for preverbal negation,
thus refuting claims that have been proposed in the literature. The peculiar
constellation found in French, in which a new negator grew out of NC concord,
was not found in any other language of the sample, not even Ewe and Karok,
the two other languages that have both NC and DN. It is further suggested that
DN may be more frequent than NC and it is argued, against other claims in the
literature, that the strategy of expressing clausal negation with only negative
pronouns or adverbs is not only typical for Europe but also for the Americas and
that it is also found in verb-initial languages.
1. Introduction
Negative concord and double clausal negation are separate phenomena. In double
negation, there are two clausal negation markers (e.g. ne and pas in (1)), but in
negative concord at least one negation is marked on a pronoun or an adverb of
time, place, or manner (e.g. rien in (2)).
(1) French
Je ne le vois pas.
i neg him see neg
‘I do not see him.’
* Thanks are due to the Science Foundation Flanders (predoctoral project 4750, first author)
and to the Japanese Society for the Promotion of Science (visitor fellowship second author).
We are also grateful to Nada Gbegble (Winneba, Ghana) for her help with the Ewe data, and
also to Volker Gast (Jena). This paper is partially based on Van Alsenoy (2011).
Lauren Van Alsenoy & Johan van der Auwera
(2) French
Je n’ ai rien vu.
i neg have nothing seen
‘I have seen nothing.’
Yet the two phenomena show striking similarities, the main similarity being the
fact that negation is formally expressed more than once, but semantically only
once. These similarities, discussed in Section 2, have made people hypothesize on
whether and how these two are related. Thus de Swart (2010) expressed the hunch
that negative concord could be a necessary condition for double negation and, ear-
lier, Zeijlstra (2004) made a proposal that went in a similar direction. These claims
are tested on a sample of 103 languages. After refuting these two claims, we have
a closer look at Ewe and Karok, the only two languages of the sample that exhibit
both negative concord and double clausal negation. We elucidate the non-French
ways in which the two phenomena can be intertwined. In a final section we briefly
discuss the seven languages that exhibit negative concord only, thereby shedding a
non-French light on this phenomenon.
‘DN’ typically arises from what is known as the ‘Jespersen Cycle’. The term is due to
Dahl (1979), referring to an important account offered by Otto Jespersen in 1917.1
In this account DN is a stage in a scenario through which a simplex negation is
replaced by another simplex negation. The middle stage has both the old and the
new negation. The cycle is represented in (4).
. This account is arguably neither the first nor the best, a good contender being Meillet
(1912). On Meillet (1912) and a proposal for a wide definition, see van der Auwera (2009,
2010).
On the relation between double clausal negation and negative concord
For a language such as French, each of the three stages is documented. In (1)
ne is the old NEG1 and pas is the new NEG2. In colloquial spoken French ne is
dropped, resulting in (5).
(5) French
Je le vois pas.
I him see neg
‘I do not see him.’
In earlier stages, only ne, the successor to Latin non, appeared and there are relic
uses even today (see Muller 1991: 227–245).
(6) French (example from Le Monde 1982; Muller 1991: 230)
La réalité, c’ est que la France ne peut
the reality it is that the France neg can
pratiquer une politique fondamentalement différente
practise a policy fundamentally different
des autres pays à économie comparable
of.the other countries at economy comparable
‘In reality, France cannot have a policy that is fundamentally different from
countries with a similar economy.’
With negative concord, henceforth also ‘NC’, one single negative meaning is also
expressed twice, and in the typical case one negation is expressed on a pronoun or
adverb. In (2) the negation is expressed both through the particle ne – the same as
the one in (1) and (6) – and through rien ‘nothing’. A further parallel is that spoken
colloquial French simplifies the doubling pattern of (2) into a pattern with just one
exponent of negation.
(7) French
J’ ai rien vu.
I have nothing seen
‘I have seen nothing.’
So in both (1) and (2) it is the negative particle ne that is deleted. Note, however,
that though the double exponence pattern of both DN and NC go to simple expo-
nence patterns, it is only the DN pattern which historically derives from a single
exponence pattern. The DN pattern with just ne is not paralleled by a NC pattern
with just ne (in (8)).
(8) French
*Je n’ ai vu.
i neg have seen
*‘I have seen nothing.’ (OK for an incomplete sentence ‘I have not seen …’)
Lauren Van Alsenoy & Johan van der Auwera
There are more parallels. For both DN and NC doubling some linguists have
argued that though doubling needs two exponents, it does not follow that both
are themselves ‘truly negative’. The view that in present-day French ne is not nega-
tive (any more) is called ‘the standard view’ by de Swart (2010: 177).2 Voices of
authority often referred to in this context are those of Damourette and Pichon
(1911–1930, 1: 131, 138) and Tesnière (1959: 111). The former are responsible for
the terms ‘discordantiel’ for ne and ‘forclusif ’ for pas. What ne does, in the view of
Damourette and Pichon, is to express discord (‘une discordance’) or, in the clearer
terms of Tesnière (1959: 111):
Le discordantiel ne forme pas à lui seul la négation. Il la prépare seulement. Et
c’est le forclusif qui la réalise.3
The claim that the second element of the doubling is not really negative either is
also found. Present-day French personne, for instance, still has a non-negative use,
as in (9), and when ne … pas was long in place, pas too had non-negative uses, as
in (10).
(9) Present-day French
Elle le fait mieux que personne.
she it does better than anyone
‘She does it better than anyone else.’
(10) 18th century pas (Muller 1991: 25)
C’est la plus jolie fille qu’ y a pas
this is the more pretty girl that there has ??
dans le canton.
in the canton
‘This is the prettiest girl there is in the canton.’
So both DN and NC doubling have seen both of their exponents accused of not
really being negative – and this is a similarity.
. De Swart (2010: 177) traces the standard view back to Bréal (1897: 221–222), but the latter
only says (1897: 222) that pas, point, rien etc. became negative through contamination (conta-
gion) by the negative particle ne, not that ne lost its negative force.
. Note that the text is not that clear. On the one hand, Tesnière says that ne is not negative by
itself, which is not the same as saying that ne is negative at all. On the other hand, he says that
ne prepares for the negation, and a preparation is not the same thing as that which it prepares
for, viz. the realization of the negation. Also, whereas the idea that ne is not really negative is
arguably standard, there is no standard view on what ne is instead. For Tesnière (1959), it is
something that prepares for the negation, for de Swart (2010: 177) ne is a scope marker, and
for Breitbarth (2008) it is a negative polarity item.
On the relation between double clausal negation and negative concord
For DN, the recognition that the Jespersen scenario of weakening and strength-
ening can have the doubling stage followed by strengthening and the addition of
a third negative marker instead of weakening and the deletion of the first nega-
tive marker took some time. Possibly the first linguist to discuss triple exponence
in the context of the Jespersen cycle was the Austronesianist Robert Early, when
he discussed the Vanuatu language Lewo (Early 1994a: 200). (12) is one of his
examples.
(12) Lewo (Early, 1994b: 411)
naga pe Ø-pa re poli
3sg r.neg 3ss-r.go neg neg
‘He hasn’t gone.’
Here too, though, the similarity is limited. With NC, multiple exponency is less
restrained than with DN. Negation can easily spread over four, five, six or seven
words. (13) is a West Flemish example with seven markers.
(13) West Flemish (Vandeweghe 2009: 13)
Hij en heeft sedertdien nooit nievers met
he neg has since never nowhere with
niemand niet vele geen leute niet meer g’had.
nobody not much no fun not more had
‘Since then he’s never had much fun with anyone anywhere.’
For clausal negation, however, marking with three negations is fairly rare already,
marking it with four negations is exceedingly rare and nothing more than four has
been attested (van der Auwera et al. 2013; Vossen & van der Auwera this volume).
Still, for both phenomena multiple exponence is not restricted to doubling.
A final point of similarity is that at least in a language such as French, DN can
be analyzed as a kind of NC or, better, as a later development, and both can be
seen as developments of negative polarity constructions. Personne and rien¸ for
instance, originally ‘person’, and ‘thing’ had a general use, but in negative polarity
contexts they implied ‘even one person’ and ‘even one thing’. In a negative con-
text the emphasis carried by the ‘even’ sense bleached and ne contaminated them
to ‘nobody’ and ‘nothing’. Pas has a similar origin, ‘not even one step’, originally
Lauren Van Alsenoy & Johan van der Auwera
in the context of movement verbs, the ‘step’ meaning also bleached and got con-
taminated, but in addition, the nominal negative sense gave way to a the sense of
clausal negation (‘not’).4 This evolution is sketched in (14).5
(14) negative polarity context NC DN
personne ‘person’ ‘even one person’ ‘nobody’
rien ‘thing’ ‘even one thing’ ‘nothing’
pas ‘step’ ‘even one step’ ‘no step’ ‘not’
. Not every indefinite often considered to be in the same paradigm as personne and rien
underwent the same evolution, however. See Hansen (2011).
. Note that we are fully aware of the fact that rien and especially personne retain negative
polarity uses. We are thus committed to accepting two meanings or uses for these words. But
this conclusion is forced upon us anyway: in elliptic answers, both rien and personne occur
without ne and must thus be negative all by themselves. See (a).
The debate on whether words like rien and personne can really be negative is a version of the
debate on the true nature (negative or not) of ne and pas, briefly sketched in the above.
On the relation between double clausal negation and negative concord
Let us look at this quote in some detail. First, a condition can be necessary or suffi-
cient. De Swart (2010) uses the term ‘crucial condition’. We take this to mean ‘nec-
essary condition’, implying that DN ‘along the lines of French’ will only happen if
the language also has NC. Second, the NC has to be of the type called ‘strict’, which
means that it is obligatory. The other type is non-strict, and in this case, irrelevant
then, NC depends on the position of the negative indefinite vis-a-vis the verb.
French, the language de Swart has in mind here, is analyzed as exhibiting strict
NC.6 Third, the ‘discontinuous negation’ of the quote is, strictly speaking, only one
type of DN, i.e. the one where the two negations are not next to one another, but
it seems that for de Swart all double negation is discontinuous with the negations
embracing the verb.7 Fourth, the link between DN and strict NC would only con-
cern DN ‘along the lines of French’. This expression is vague: just how similar does
the DN of a language have to be to qualify as DN ‘along the lines of French’? And
finally, de Swart realizes that she does not have sufficient typological support for
substantiating the claim. Indeed, de Swart’s data are strongly Eurocentric and so
the decision to leave the issue for future work is the right one.8
In (15) we reformulate de Swart’s conjecture in a simple format.
(15) Strict NC is a necessary condition for DN ‘along the lines of French’
Note that although de Swart does not set out to support the universal with cross-
linguistic evidence, her book contains statements on the frequency of both NC
. The phrase left out in the quote says that a language must be a so-called ‘Type III lan-
guage’. This means, in her framework, that a language will have so-called ‘non-strict NC’. Yet
the quoted passage continues with requiring the language to have ‘strict NC’ and since stan-
dard French has strict NC, we take it that the deleted phrase has a typing error.
. De Swart (2010) is by no means the only linguist to interpret the Jespersen cycle as a
scenario taking the language from an obligatorily preverbal negation via an obligatorily dis-
continuous negation to an obligatorily postverbal one. The idea seems to have originated in
Vennemann (1974). Interestingly, neither Jespersen (1917) nor Meillet (1912) – see Note 2 –
took word order facts as an ingredient of the Jespersen cycle. See also van der Auwera &
Neuckermans (2004: 459–460). It is true that DN very often involves a combination of an
older preverbal and a newer postverbal component – see also Vossen & van der Auwera (this
volume), but not always. Latin, for example, had a minimizer oenum ‘one thing’ and an older
negator ne, but instead of embracing the verb with a postverbal minimizer contaminating into
a negator, the two elements underwent univerbation, yielding the new negator non. Or take
Navajo, discussed further down: Navajo has DN with preverbal and postverbal parts, but the
new negator is not the postverbal, but the preverbal one.
. Bernini and Ramat (1996) is an earlier typological work, which explicitly focusses on
European languages and advances the hypothesis that European negation is special enough to
constitute a feature of ‘Standard Average European’.
Lauren Van Alsenoy & Johan van der Auwera
and DN that are at least favorable to (15). DN, de Swart (2010: 10) claims, is a
rare phenomenon, whereas NC – at least the general type, both strict and non-
strict – is claimed to be widespread (de Swart 2010: 21; see also Israel 2011: 43).
Abstracting from the fact that (15) is about strict NC and not about NC as such,
these frequency claims are in accord with the frequency implied in the universal
of (15): some NC languages will have DN and some not, hence the number of NC
languages will be higher than the number of DN languages.
Interestingly, a few years before, Zeijlstra (2004) studied both DN and NC,
again mostly on European languages, he too ventured a typological claim and
considered DN to be definitionally discontinuous and embracing, i.e. with one
preverbal and one postverbal negation. His claim (2004: 146–147) is the one in
(16) – ‘PreVN’ stands for ‘Preverbal negation’.
(16) NC is a necessary condition for PreVN9
A few remarks are in order. First, for clarity’s sake we added the word ‘neces-
sary’ to this formulation. We think that it captures what Zeijlstra had in mind.
Second, his NC does not have to be strict. Third, the clausal negation does not
have to be double, all that is required is that there is a preverbal negation. This
is always the case with DN, according to Zeijlstra, but a preverbal condition can,
of course, occur without a postverbal one. So the hypothesis does not really con-
cern the relation between DN and NC, but because of the similarity to de Swart’s
claim, we discuss it anyway. Finally, Zeijlstra does not wait for ‘future work’ and
already provides evidence in support of his claim. If (16) holds, it will allow for
three types of languages: (i) languages with PreVN and NC, (ii) languages with-
out PreVN but with NC, and (iii) languages without PreVN and without NC. One
language type is predicted not to exist: a language with PreVN and without NC.
Zeijlstra’s data set consists of 21 languages and for some also different dialects.
In this dataset only English is registered with two dialects, i.e. Standard English
and NC English. As the name ‘NC English’ suggests, this is the variant allowing
constructions such as (17).
(17) I don’t see nobody.
In our version of Zeijlstra’s analysis, all of the languages support his claim, as
shown in Table 1, a simplified and slightly adapted version of Zeijlstra (2004: 147).
Each of the three admitted language types is attested and there are no attestations
. This universal is part of a more wide ranging and more specific universal, involving the
distinction between so-called ‘strict’ and ‘non-strict’ NC and the one between so-called ‘true’
and non-‘true’ negative imperatives. These details do not matter here.
On the relation between double clausal negation and negative concord
Table 1. Languages studied by Zeijlstra (2004) and supporting the claim in (16)
PreVN NC
4. N
egative concord as a prerequisite for double negation
or preverbal negation
In this section we will test de Swart’s claim on data coming from the Austronesian
and Austro-Asiatic families and from the families from Africa and the Ameri-
cas (see Appendix A). These languages are taken from a representative 179-lan-
guage sample based on the one used by Miestamo (2005). Though there are still
gaps in the sample and the language families do not cover the whole world,11 we
think that we have sufficient information to put her hypothesis to the test. In total
we have found relatively reliable information for 103 languages. Table 2 classifies
. In the original version of the table, there is actually one counterexample, viz. Standard
English. In Zeijlstra’s analysis, English not is considered preverbal and the verb that matters is
the main verb, not the auxiliary. Yet he does not consider the fact that Standard English goes
against his generalization to be too problematic. What saves Standard English, in his view
(2004: 145), is that it allows NC-like behavior with negative polarity items such as anybody
in (a).
A more satisfactory analysis, it seems to us, is to consider Standard English to have postverbal
negation and to consider the relevant verb to be the auxiliary. In any case, even under his
analysis it seems that the generalization in (16) is backed up by a good number of languages.
. Whole world coverage is the aim of Van Alsenoy (in prep.).
Lauren Van Alsenoy & Johan van der Auwera
them according to the four subtypes defined in the universal in (15), the three
possible ones and the impossible one. For DN we consulted WALS (Dryer &
Haspelmath 2011). In contrast to WALS, however, Quechua and Navajo were clas-
sified as double negation languages (for Quechua, see Cole 1982: 86, for Navajo,
see van Gelderen 2008). We counted both obligatory and optional DN. For NC
there is often not enough information to say whether it is strict or not. For this
reason, we did not tabulate strict NC, but NC tout court. This is still relevant: the
hypothesized impossible language type will have DN but no strict NC. If we find
languages with DN and no NC tout court, these will also count as languages with
DN and without strict NC. It is also true that we often lack information on the
uses of the relevant items in negative polarity contexts and elliptic answers. In such
cases our ‘null hypothesis’ was to take them as NC items. It is more likely therefore
that we took too many items – rather than not enough – to be NC items. To give
a very rough indication of where the languages are found, we group the languages
in three ‘macro-areas: ‘Af ’ for Africa, ‘Am’ for the Americas, and ‘Au’ for Austro-
nesian and Austro-Asiatic.
What we can conclude from Table 2 is first and foremost that the hypothesized
impossible language type does exist. There are 16 languages that have DN but no
NC. Only 2 languages exhibit DN as well as strict negative concord, namely the
On the relation between double clausal negation and negative concord
Kwa language Ewe, spoken in Ghana, and the nearly extinct isolate Karok, spoken
in the United States. Based on these results, we consider the claim or – to be fair –
the hunch that NC is a necessary condition for DN to be wrong. Second, we can
also comment on de Swart’s frequency claims. Is it the case the DN is rare and NC
frequent? The answer is negative: in our sample there are 18 DN languages (16 + 2)
vs. only 9 NC languages (7 + 2).
Note that our sample of Austronesian and Austro-Asiatic languages did not
contain a single DN language. That does not mean, of course, that DN is not
attested there. At least for the Austronesian languages Vossen and van der Auwera
(this volume) show that DN is not rare at all. For 24 of their DN languages (see
Appendix B) we checked whether they also exhibited NC. None of them did. So
they all instantiate the ‘impossible’ language type.
We can also test the claim that Zeijlstra made about the relation between
PreVN and NC. The number of languages for which we had sufficient information
is 103, as in Table 2.
The conclusion is clear: more than half of the languages studied are counter-
examples to Zeijlstra’s claim that NC is a necessary condition for PreVN. So we
should consider Zeijlstra’s universal to be wrong.
Lauren Van Alsenoy & Johan van der Auwera
Note also that Table 3 shows preverbal negation to be more common than post-
verbal negation (71 vs. 32 languages). This relates to a principle called ‘Neg(ative)
First’ by Horn (1989: 292), associated with Jespersen (1917: 5) and cross-linguistically
confirmed by Dryer (1998) and Van Olmen (2010).
So far we have established that NC is not a necessary condition for either
DN or PreVN. Note that NC is not a sufficient condition for DN or PreVN either.
Table 2 shows that there are 7 NC languages without DN and Table 3 shows that
there are 3 NC languages without PreVN. We can also check whether there is a
statistically significant relation between NC and either PrevN or DN. Consider
first Table 4, with figures extracted from Table 2.
Table 4. Correlation between the presence of negative concord and double negation
properties number of languages properties number of languages
It says that of the 9 languages with NC 2 have DN and 7 don’t, whereas in the
94 languages without NC 16 have DN and 78 don’t. The proportion of languages
with and without DN is roughly the same, independently of whether or not the
language has NC. That a language would have DN does not seem to depend on
whether it has NC, an impression confirmed by a Fisher’s Exact Test. There is also
no statistically relevant correlation between the data sets shown in Table 5: know-
ing that a language has DN has no predictive value for answering whether the
language has NC not.
Table 5. Correlation between the presence of double negation and negative concord
properties number of languages properties number of languages
Tables 6 and 7 show data sets for the Zeijlstra claim involving NC and PrevN,
with the figures rearranged from Table 3. From glancing at the proportions and
Table 6. Correlation between the presence of negative concord and preverbal negation
properties number of languages properties number of languages
checking it with a Fisher’s Exact, it is clear that NC is not a predictor for PrevN and
neither is PrevN a predictor for NC.
Table 7. Correlation between the presence of preverbal negation and negative concord
properties number of languages properties number of languages
At the typological level therefore, we must conclude that NC, on the one hand,
and DN and PreVN, on the other hand, are independent phenomena. That still
doesn’t imply that there can’t be an interesting relation on the language-particular
level. For French, for example, there clearly is an interesting relation between NC
and DN, viz. a diachronic one, for the latter is a development from NC. In the next
sections, we will have a closer look at the sample’s 9 NC languages. We will start
with the two languages that have both DN and NC.
In French the relation between DN and NC is first and foremost a semantic one.
Put in diachronic terms: pas got contaminated by the negative polarity context
(‘even a step’), then by the negative context (‘not even a step’), and the meaning
further changed from nominal to clausal negation (‘not’). Contamination, we will
see, is important for Ewe, but what happened in Ewe, is also rather different. In
Karok, however, the relation is first and foremost a formal one: one of the expo-
nents of clausal negation merged with the pronoun. We will start with Ewe.
For ‘somebody’ Ewe uses the generic noun ame together with a determiner aɖe
‘one, some’, thus yielding the phrase ame aɖe, but when the pronoun is in the
scope of negation, it is not ame aɖe that appears, but ame aɖeke, in which -ke is an
emphatic marker, comparable to ever, as used in the English free relatives what-
ever, whoever, whenever, etc. A positive, free choice use is illustrated in (21).
Lauren Van Alsenoy & Johan van der Auwera
Another parallel with French, is that more than one negative quantifier can occur
with only one semantic negation.
(23) Ewe (Nada Ggegble, p.c.)
Ame aɖeke me-kpɔ ame aɖeke o
person no neg-see person no neg
‘Nobody saw anybody.’
But different from French, in full clauses the ‘nobody’ construction does not com-
bine with just one of the exponents of DN (ne in standard French French, pas in
Québec French, see (39) below), but with both. Ewe thus truly exhibits a triple
exponence. Of course, what we are discussing here is not a pronoun, but a deter-
miner: ame and aɖeke are still taken as separate words. For ‘nothing’, however, the
corresponding generic noun nú ‘thing’ has merged with aɖeke¸ yielding naɖéké/
nánéke ‘nothing’ (Ameka 1991: 46).
The presence of scalar focus markers on negative indefinites is addressed and
described by Haspelmath (1997: 157–164), who notes that there are two possible
pathways. First, the constructions may develop from non-specific free relative
clauses or from concessive conditional clauses, as schematically represented by
the pseudo-English example in (24).
On the relation between double clausal negation and negative concord
Second, the scalar focus particles may combine directly with the generic or inter-
rogative, depending on what the derivational base of the indefinite is. Again a
pseudo-English example can be used for illustration as in (25):
(25) Pseudo-English
I never saw even-person.
. This emphatic element -xay is, remarkably, also the element that can replace the second
negator, as in (a).
This suggests that the -xay marker competes or may have competed with the postverbal –
(h)ara because they might have shared or still share, at least in some contexts, an emphatic
meaning.
. Not surprisingly, negative absorption doesn’t occur often in Austronesian or Austro-
Asiatic languages, since these are often isolating languages. Apart from that, negative indefi-
nites in these languages are often expressed by means of a positive indefinite and a negative
existential verb. In contrast to affixes and particles, negative verbs are unlikely to be absorbed.
An example from Austro-Asiatic is Pnar, in which negative indefinites are always expressed
by mean of a negative existential verb plus a generic element, an indefinite NP, the indefinite
numeral ‘one’ or a relative clause (Koshy 2009: 46).
Like Old High German n(i)-/nih-/ne(h)-, the Karok negator pu- is probably the
older negator. The postverbal -ara is the newer one and it was probably emphatic.
One recognizes -ara in the emphatic forms vura ‘just’ (Bright 1957: 15) and hara
‘and all’ (Bright 1957: 150), whereby it seems likely that the second part used to
mean something like ‘at all’, adding emphatic force. Interestingly, negative absorp-
tion is not obligatory. As one can see in (29), elements can occur in between the
indefinite and the first negator, in this case two emphatic markers.
(29) Karok (Bright 1957: 140)
pú-xay vúra fâˑt mah-ára
neg-emp emp what v-neg
‘He didn’t see anything.’
Other languages that resort to the use of negative existential verbal constructions for the
expression of indefinites without indefinites through negative absorption are the African
languages Iraqw (Mous 1992: 100,121, 211), Kinyarwanda (Kimenyi 1980: 131), the American
languages Wayampi (Jensen 1994: 346), Lillooet (Davis 2005: 20), Kutenai (Canestrelli
1927: 39), the Austronesian language Chamorro (Cooreman 1987: 45).
Lauren Van Alsenoy & Johan van der Auwera
Yet another variant, which was not found in our data consists of the absorption
of a negative scalar focus particle meaning ‘not even’. This kind of absorption is
different from the absorption of a clausal negator, since a negative scalar focus
particle obviously does not take part in clausal negation. It is found in Russian
and Spanish where the negative indefinites contain the scalar focus particle ni, e.g.
Russian nikt ‘nobody’ and Spanish ninguno ‘no’. Of course, even in the case of these
negative scalar focus markers, we are dealing with a contraction of the clausal
negator and a non-negative scalar focus particle, preceding the contraction to
negative indefinites. According to Haspelmath (1997: 222), indefinites of this type
are more closely related to indefinites that are formed with positive scalar focus
markings, as described for Ewe, than to indefinites that arise by negative absorp-
tion. The important point, he notes, is that ‘although the negative focus particles
[as in Russian or Spanish ni] are negative in some sense, they are quite indepen-
dent of (and sometimes formally unrelated to) the verbal negator which expresses
sentence negation’. Still we decided to treat them together with other cases of nega-
tive absorption. Though it is true that the negative scalar focus marker and clausal
negator are unrelated in Finnish for example, they are clearly related in the case
of Russian and Spanish, the clausal negators being ne and no respectively. In addi-
tion, it doesn’t seem to be a coincidence that both Russian and Spanish exhibited
or still exhibit non-strict negative concord, the absence of a clausal negator on
the verb being a side-effect of negative absorption. This will be discussed in more
detail below.
The second parameter is the strength of the absorption. In German, Russian
and Spanish, the negative element(s) and the positive pronoun are inseparable, but
. Haspelmath (1997: 207) writes the preverbal negator ma and the indefinite ‘one’ as two
separate words, but Lucas (2009: 205–206) considers maħaddiš to be a negative quantifier.
On the relation between double clausal negation and negative concord
in Karok and in Egyptian Arabic, they are separable. For Karok this is shown in
(29), for Egyptian Arabic in (32).
(32) Egyptian Arabic (Lucas 2009: 207)
ma-šaf-nī-š ḥadd
neg-see.prf.3msg-me-neg anyone
‘No one saw me.’
In (32) ma and -š embrace the verb, not the pronoun, and note that what (31) and
(32) have in common is that the negation comes early in the sentence, which is an
illustration of the Neg First principle, appealed to before when it came to explain-
ing why most languages prefer preverbal to postverbal negation.
An interesting variation is Navajo. Navajo has double negation. The postver-
bal part seems to be the oldest part (van Gelderen 2008: 220), its verb is clause-
final and the preverbal part has a variable position. But when there is an indefinite,
the preverbal part is attracted to the indefinite.15 This attraction resembles the one
in Karok in that other elements are still allowed in between the preverbal negative
and the indefinite (see the emphasizer xay in (29)), but the preferred pattern has
the indefinite immediately following the negator. Unlike in Karok, the attraction
has not reached the morphological level – the Navajo preverbal negator still has
word status – and does not therefore constitute absorption.
(33) Navajo (Hale & Platero 2000: 73)
doo háída níyáa-da
neg anyone 3.go-neg
‘No one has arrived.’
(34) Navajo (Hale & Platero 2000: 75)
i-zhé’é béégashii doo nayiisnii′-da
my-father cow neg 3.buy-neg
‘My father did not buy a cow.’
The third parameter of variation in negative absorption concerns the relation with
NC. In Egyptian Arabic there is no NC,16 for the indefinite attracts either both
parts of the DN or none – both in Example (31) and none in Example (32). In
Karok and German, we see NC and the absorption is the result of the fact that
. “What is essential here is that the negative particle doo must precede the indefinite,
despite the fact that the unmarked position for this element […] is just before the verb”
(Hale & Platero 2000: 75).
. Possibly the language has some kind of negative concord, depending on the analysis,
but only with the determiner wala ‘not even’, ‘no’ (for an extensive discussion, see Lucas
2008: 209–214).
Lauren Van Alsenoy & Johan van der Auwera
What the language then does to make the semantics clearer is to add a clausal
negator, resulting in NC. This, we assume, happened in Old High German. It prob-
ably also happened in Russian, which had non-strict NC, which was (slightly) pre-
ferred in Old Russian (Haspelmath 1997: 211), as an intermediate stage, and also
in Italian, Spanish and Portuguese, but they truly stopped at non-strict NC, the
‘first stage of restoration’ (Haspelmath 1997: 212–213).
In Russian and German, the order got restored by adding the old, preverbal
negator again, but a new negator can also restore the order. This can be illustrated
with Quebec French, and here we furthermore see that this process is indepen-
dent of whether or not the negative indefinite shows absorption. Québec French
personne had lost NC with ne, like colloquial French (see Examples (2) and (7)),
resulting in the undesirable constellation just described, but different from collo-
quial French, it now allows clause scope pas, thus reinstating NC.
The constellation with only a negative pronoun and no clausal negator is, of course,
not only found in the progressive colloquial French illustrated in (7) and older
German, but also in the other Germanic languages English and Dutch. H aspelmath
(1997: 202) claims that it is typical for Europe and rare elsewhere. In his sample, it
is only represented by European languages. This might have something to do with
the Eurocentricity of the sample. Haspelmath (2011) notes that it is common in
On the relation between double clausal negation and negative concord
. Strictly speaking, this variety is not part of the sample. The latter contains Sochiapan
Chinantec.
Lauren Van Alsenoy & Johan van der Auwera
In order to say ‘Adolfo didn’t say anything’, however, one would expect a sentence
of the type ‘not-saw … something’, but instead Chinantec negativizes the pronoun
and puts it in first position.
(43) Tepetotutla Chinantec (Westley 1991: 17)
caL-ʔeM kaMhuáʔM zïMdoMH
neg-what said Adolfo
‘Adolfo didn’t say anything.’
What could account for this pattern is the fact that, as in many other languages the
indefinite is based on an interrogative. In Tepetotutla Chinantec, the interrogatives
are always in front position (Westley 1991: 97), despite the verb-initial nature of
the language. The details of this kind of patterning are left for further research, but
in any case, the Tepetotutla Chinantec facts are not covered and are predicted not
to exist if you let the absorption pattern depend on the basic constituent order of
a language.
There are seven languages in our sample that have NC but no DN, viz. Kanuri,
Kunama, Somali, Chiapas Zoque, Huave, Páez and Chamorro. In three of them,
the NC system seems to be similar to that of Ewe, except that Ewe is a DN language
and the three languages are not. In another three languages, we see contact inter-
ference as an important factor.
The Kanuri, Kunama and Somali negative indefinites can be compared to
the Ewe indefinites. They also have some sort of emphatic marker on their nega-
tive indefinites. This is not to say that the emphatic markers can fulfill the exact
same functions in these languages. The Kanuri marker ma, which is used to
derive negative indefinites (sentence (44)), can be used to express focus, as in
sentence (45).
(44) Kanuri (Cyffer 2009: 86)
ndúmá ráksə lezə̂-nyí
who-neg could go-neg
‘Nobody could go.’
(45) Kanuri (Cyffer 2009: 86)
Bíntu-má íshin
Bintu-emp will come
‘It is Binyu who will come.’
But ba can also function as an intensifier meaning something like ‘(not) at all’, ‘not
ever’. In sentence (47), one can see how a negative concord construction results.18
(47) Somali (Saeed 1999: 186)
waxbá má aan sɪ́n
thing-neg neg I gave
‘I didn’t give him anything.’
Páez seems to exhibit NC and there is absorption, but the origin of the negative
determiner juxpa as used in Example (49) is unclear.
(49) Páez (Jung 2008: 196)
kɪ̃x juxpa pija-me: wala:na ũs-taʔ
what neg.det grow.up-neg learning decl-3pl
‘They grow up without learning anything.’
Apart from these 4 languages, there are three others where the negative indefi-
nites are clearly at least partly borrowed from Spanish. These languages are Zoque,
. Somali can also derive negative indefinites by adding a negative nominal suffix. We do
not understand this process.
Huave and Chamorro. Both in Huave and in Zoque, the negative marker ni is
borrowed:
(50) Zoque (Faarlund 2012: 61)
te’ yomo’isñe ni-’isuna’ajk ji-’myujsje ñünji
the wife’s neg-some neg-know name
‘Nobody knew the wife’s name.’
(51) Huave (Kim 2008: 246)
como aj kuchu ñunch ngo mundiak, ngo mangoch ñi-kwej
as the little boy neg speak neg answer neg-what
‘Since the little boy didn’t talk, he didn’t answer anything.’
In the example from Zoque, ni attaches to the indefinite pronoun i ‘some’ (Faarlund
2012: 61). In Huave, the base kwej ‘thing’ is used to express ‘nothing’ (Kim
2008: 217), and the interrogative jang ‘who’ to express ‘nobody’ (Kim 2008: 238).
In Zoque, negative concord seems to be strict; both preverbal and postverbal
examples of negative indefinites co-occurring with clausal negation can be found
(Faarlund 2012: 61). In Huave, only postverbal negative indefinites were found, so
nothing can be said on the nature of negative concord.
The Austronesian language Chamorro uses the same negative indefinite
marker ni- to derive negative indefinites. Like Huave and Zoque, it borrows the
indefiniteness marker, and, in most cases, it retains the original base.
(52) Chamorro (Chung 2009: 103)
Ti hubisita ni háyiyi ha’
neg visit not anyone emp
‘I didn’t visit anyone.’
When there is a pronominally used ‘one’, however, Chamorro seems to use the
Spanish equivalent.
(53) Chamorro (Cooreman 1987: 45)
taya’ ni unu tumungo’ kao guaha si rai
neg.ex no one know q have/be unm king
haga ña
daughter 3sg.pos
‘There was no one at all who knew the king had a daughter.’
As to the presence of strict versus non-strict negative concord, it seems that Cham-
orro resembles Spanish.
(54) Chamorro (Chung 2009: 103)
Ni unu siña tasokni nu esti na chinätsaga
neg one can blame obl this l hardship
‘We can blame no one for this hardship.’
On the relation between double clausal negation and negative concord
It differs from Spanish in that it is only in focus construction that the indefinite can
be the sole marker of negation. Non-focussed negative arguments and adjuncts
always have to follow the negator, the clausal negator ti in the case of negative
nonsubjects and the negative existential verb taya in the case of subjects (Chung
2009: 103–104).
7. Conclusion
Despite similarities between NC and DN, the two phenomena seem to coincide
only rarely. We surveyed 103 non-European languages and we concluded that
NC cannot be considered to be a necessary condition for DN nor for preverbal
negation, thus refuting two claims that have been proposed in the literature. The
peculiar constellation that we see in French, in which a new negator grew out of
NC concord, was not found in any other language of our sample. The two other
languages that have both NC and DN are Ewe and Karok. In Ewe we see a DN
system contaminating an indefinite pronoun into a negative pronoun, thus giving
rise to triple exponence. In Karok we see a DN system creating a negative pronoun
out of an indefinite by absorbing one of the clausal negators into the pronoun. We
analyzed absorption in some detail, distinguishing between three parameters of
variation, and we also discussed NC as it occurs in seven additional languages, i.e.
languages with NC but without DN. We further expressed our hunch that DN may
be more frequent than NC and we argued, against other claims in the literature,
that the strategy of expressing clausal negation with only negative pronouns or
adverbs is not only typical for Europe but also for the Americas and that it is also
found in verb-initial languages.
Abbreviations
References
Ameka, Felix K. 1991. Ewe: Its Grammatical Constructions and Illocutionary Devices. Ph.D.
dissertation, Australian National University.
Bender, Lionel. 1996. Kunama. München: Lincom.
Bernini, Giuliano & Ramat, Paolo. 1996. Negative Sentences in the Languages of Europe. A Typo-
logical Approach. Berlin: Mouton de Gruyter. DOI: 10.1515/9783110819748
Bittner, Maria. 1995. Quantification in Eskimo. In Quantification in Natural Languages, Emmon
W. Bach, Eloise Jelinek, Angelika Kratzer & Barbara H. Partee (eds), 59–80. Dordrecht:
Kluwer.
Bréal, Michel. 1897. Essai de sémantique. Paris: Hachette.
Breitbarth, Anne. 2008. A hybrid approach to Jespersen’s cycle in West Germanic. Journal of
Comparative Germanic Linguistics 12: 81–114. DOI: 10.1007/s10828-009-9027-7
Bright, William. 1957. The Karok Language. Berkeley CA: University of California Press.
Canestrelli, Philippo. 1927. Grammar of the Kutenai language. International Journal of American
Linguistics 4: 1–84.
Chung, Sandra. 2009. Six arguments for wh-movement in Chamorro. In Hypothesis A/
Hypothesis B: Linguistic Explorations in Honor of David M. Perlmutter, Donna B. Gerdts,
John C. Moore & Maria Polinsky (eds), 91–110. Cambridge MA: The MIT Press.
Cole, Peter. 1982. Imbabura Quechua. Amsterdam: North-Holland.
Cooreman, Ann M. 1987. Transitivity and Discourse Continuity in Chamorro Narratives. Berlin:
Mouton De Gruyter. DOI: 10.1515/9783110851014
Curnow, Timoth Jowan. 1997. A Grammar of Awa Pit (Cuaiquer): An Indigenous Language of
South-western Columbia. Ph.D. dissertation, Australian National University.
Cyffer, Norbert. 2009. Negative Patterns in Kanuri. In Negative Patterns in West African languages
and Beyond [Typological Studies in Language 87], Norbert Cyffer, Erwin E bermann &
Georg Ziegelmeyer (eds), 71–92. Amsterdam: John Benjamins.
Damourette, Jacques & Pichon, Édouard. 1911–1930. Des mots à la pensée: Essai de grammaire
de la langue française, 8 Vols. Paris d’Artrey.
Dahl, Östen. 1979. Typology of sentence negation. Linguistics 17: 79–106. DOI: 10.1515/ling.1979
.17.1-2.79
Davis, Henry. 2005. On the syntax and semantics of negation in Salish. International Journal of
American Linguistics 71: 1–55.
de Swart, Henriëtte 2010. Expression and interpretation of negation. An OT typology. Dordrecht:
Springer. DOI: 10.1007/978-90-481-3162-4
Dryer, Matthew. 1998. Universals of negative position. In Studies in Syntactic Typology [Typo-
logical Studies in Language 17], Michael Hammond, Edith Moravcsik & Jessica Wirth
(eds), 93–124. Amsterdam: John Benjamins.
Dryer, Matthew S. & Haspelmath, Martin (eds). 2011. The World Atlas of Language Structures
Online. München: Max Planck Digital Library.〈http://wals.info/〉 (4 May 2012).
Early, Robert. 1994a. Lewo. In Typological Studies in Negation [Typological Studies in Language
29], Peter Kahrel & René van den Berg (eds), 65–92. Amsterdam: John Benjamins.
Early, Robert. 1994b. A Grammar of Lewo, Vanuatu. Ph.D. dissertation, Australian National
University.
Faarlund, Jan Terje. 2012. A Grammar of Chiapas Zoque. Oxford: Oxford University Press.
On the relation between double clausal negation and negative concord
Galucio, Ana Vilacy. 2001. The Morphosyntax of Mekens (Tupi). Ph.D. dissertation, University
of Chicago.
Hale, Ken & Platero, Paul. 2000. Negative polarity expressions in Navajo. In The Athabaskan
Languages: Perspectives on a Native American Language Family, Theodore B. Fernald &
Paul R. Platero (eds), 73–91. Oxford: OUP.
Hansen, Maj-Britt Mosegaard. 2012. On the evolution of temporal n-words in Medieval French.
Language Sciences 34: 76–91. DOI: 10.1016/j.langsci.2011.06.003
Haspelmath, Martin. 1997. Indefinite Pronouns. Oxford: Clarendon Press.
Haspelmath, Martin. 2011. Indefinite pronouns. In The World Atlas of Language Structures
Online, Matthew S, Dryer & Martin Haspelmath, Munich: Max Planck Digital Library,
Chapter 46. 〈http://wals.info/chapter/46〉 (7 March 2013).
Henningsson, Lars-Ake & Holmer, Arthur. 2008. The distribution of quantifiers in Seediq. Lund
University Working Papers 53: 23–41.
Horn, Laurence 1989. A Natural History of Negation. Chicago IL: University of Chicago
Press.
Israel, Michael. 2011. The Grammar of Polarity. Cambridge: CUP. DOI: 10.1017/CBO97805
11975288
Jäger, Agnes 2008. History of German Negation [Linguistik Aktuell/Linguistics Today 118].
Amsterdam: John Benjamins. DOI: 10.1075/la.118
Jensen, Allen A. 1994. Wayampi. In Typological Studies in Negation, Peter Kahrel & René van
den Berg (eds), 343–364. Amsterdam: John Benjamins.
Jespersen, Otto. 1917. Negation in English and Other Languages. København: A. F. Høst & Søn.
Jung, Sandra. 2008. Gramática del páez o nasa yuwe. Descripción de una lengua indígena de
Colombia. München: Lincom.
Kari, Ethelbert Emmanuel. 2004. A Reference Grammar of Degema. Köln: Rüdiger Köppe.
Kim, Yuni. 2008. Topics in Phonology and Morphology of San Francisco del Mar Huave. Ph.D.
dissertation, University of Berkeley, California.
Kimenyi, Alexandre. 1980. A Relational Grammar of Kinyarwanda. Berkeley: University of
California Press.
Koshy, Anish. 2009. Indefinite pronouns in Pnar. Mon-Khmer Studies 38: 41–56.
Lucas, Christopher. 2009. The Development of Negation in Arabic and Afro-Asiatic. Ph.D.
dissertation, University of Cambridge.
Mous, Martinus Petrus Gerardus Maria. 1992. A Grammar of Iraqw. Rijksuniversiteit Leiden.
Man, Edward Horace. 1889. A Dictionary of the Central Nicobarese Language (English-Nicobarese
and Nicobarese-English). London: s.n.
Meillet, Antoine. 1912. L’évolution des formes grammaticales. Scientia 12: 384–400. Reprinted
in Meillet, Antoine. 1926. Linguistique historique et linguistique générale, 130–148. Paris:
H. Champion.
Miestamo, Matti. 2005. Standard Negation: The Negation of Declarative Verbal Main Clauses in a
Typological Perspective. Berlin : Mouton de Gruyter.
Muller, Claude. 1991. La négation en français. Syntaxe, sémantique et éléments de comparaison
avec les autres langues romanes. Genève: Droz.
Saeed, John I. 1999. Somali [London Oriental and African Language Library 10]. Amsterdam:
John Benjamins. DOI: 10.1075/loall.10
Tesnière, Lucien. 1959. Éléments de syntaxe structure. Paris: Klincksieck.
Lauren Van Alsenoy & Johan van der Auwera
Appendix
A. The sample19
The languages between square brackets are the ones for which insufficient data have been
gathered.
. For reasons of space only the references cited in the text are in the reference list. For
more information on the sources from the appendix, the authors can be contacted.
On the relation between double clausal negation and negative concord
Africa
Khoisan Central Khoisan Khoekhoe Hagman 1977
Böhm 1985
Northern Khoisan Ju’huan Snyman 1970
Dickens 2005
Niger-Congo Adamawa-Ubangian Gbeya Bossangoa Samarin 1966
Northern Atlantic Diola-Fogny Sapir 1965
David J.Sapir, p.c.
Defoid Yoruba Ashiwaju 1968
Rowlands 1965
Bamgbose 1966
Edoid Degema Kari 2004
Igboid Igbo Green & Igwe 1963
Nupoid Nupe Crowther 1864
Kwa Ewe Ameka 1991
Gur Supyire Carlson 1994
Ijoid Ijo (Kolukuma) Williamson 1965
Nilo-Saharan Bongo-Bagirmi Bagirmi Stevenson 1969
Kresh Kresh Santandrea 1976
Brown 1994
Lendu Ngiti Kutsch Lojenga 1994
Moru-Ma’di Ma’di Blackings & Fabb 2003
Kuliak So Carlin 1993
Heine & Carlin (2010 (draft
dictionary online))
Nilotic Maasai Tucker & Mpaayei 1955
Nubian Nubian (Dongolese) Armbruster 1960
von Massenbach 1963
Surmic Murle Lyth 1971
Kunama Kunama Bender 1996
Maban Masalit Edgar 1989
Saharan Kanuri Hutchison 1976
Cyffer 1998
Songhay Koyraboro Senni Heath 1999
(Continued)
Lauren Van Alsenoy & Johan van der Auwera
(Continued)
On the relation between double clausal negation and negative concord
North America
Eskimo-Aleut Eskimo Greenlandic Bittner 1995
Na-Dene Athapaskan Navajo Hale & Platero 2000
Haida Enrico 2003
Algic Algonquian Cree (Plains) Wolfart 1973
Okimāsis 2004
Wiyot Teeter 1961
Iroquoian Northern Iroquian [Oneida] Abbott 2000
Yuchi Linn 2001
Muskogean Muskogean Koasati Kimball et al. 1991
Kiowa Tamoan Kiowa Watkins 1984
Uto-Aztecan Aztecan Nahuatl MacSwan 1999
(South East Puebla)
Numic Comanche Charney 1993
Tepiman Pima Bajo Estrada Fernández 1996
Wakashan Southern Wakashan Makah Davidson 2002
Salishan Bella Coola Davis & Saunders 1992
Davis & Saunders 1997
Interior Salish Lilloeet Davis & Caldecott 2007
Davis 2005
Chimakuan Chimakuan [Quileute] Andrade 1933
Kutenai Chamberlain 1909
Penutian
Klamath-Modoc Klamath Barker 1964
Maiduan [Maidu]
Sahaptian Nez Perce Deal 2010
Tsimshianic Tsimshian
Wintuan Wintu Pitkin 1984
Hokan
Pomo [Pomo Moshinsky 1974
(Southeastern) ]
Seri Moser & Marlett 2005
Yuman Maricopa Gordon 1986
(Continued)
Lauren Van Alsenoy & Johan van der Auwera
South America
Chibchan Aruak [Ika] Frank 1990
Paya [Pech] Holt 1999
[Rama]
Choco Choco Epena Pedee Harms 1994
Páezan Páez Jung 2008
Barbacoan Awa Pit Curnow 1997
Guahiban [Cuiba] Berg & Kerr 1973
Tucanoan Tucanoan Tuyuca Barnes 1994
Andoke Landaburu 1979
[Betoi] Zamponi 2003
[Yaruro] Mosonyi et al. 2000a
Warao Romero-Figueroa 1997
Yanomam [Sanuma] Borgman 1990
Waorani Peeke 1994
Peeke 1991
(Continued)
On the relation between double clausal negation and negative concord
B. Austronesian languages with DN (based on Vossen & van der Auwera, this volume)
Biak South Halmahera (Eastern
West New Guinea Malayo-Polynesian)
Fehan (Tetun) Central Malayo-Polynesian
Fordata Central Malayo-Polynesian
Hiligaynon Meso-Philippine
Muna Sulawesi
Ambai (Lolovoli) Oceanic
Avava Oceanic
Kwaio Oceanic
Kwamera Oceanic
Lavukaleve Lavukaleve Solomons East-Papuan
Lenakel Oceanic
Lewo Oceanic
Nelemwa Oceanic
Nese (Northwest Malakula) Oceanic
Naman Oceanic
Paamese Oceanic
Raga Oceanic
Rapanui Oceanic
Rotuman Oceanic
Southeast Ambrym Oceanic
Tahitian Oceanic
Teop Oceanic
Tongan Oceanic
Toqabaqita Oceanic
The Jespersen cycles seen from Austronesian*
Most work on the so-called ‘Jespersen Cycle’ has been done on the languages of
Europe, less on languages of Africa. In this paper the authors look at Austronesian
languages, with the aim to determine to what extent Austronesian doubling
is compatible with a Jespersen cycle scenario. The properties of the potential
Austronesian manifestations of Jespersenian double, triple and even quadruple
negation are highlighted, and the role of word order, etyma, emphasis and contact
interference are investigated. However difficult it is to discuss a hypothesis about
the diachrony of negation on data from 410 languages, most of which do not
have a recorded history, still, it is clear that the multiple exponence of negation
in Austronesian is in some cases clearly interpretable in terms of the Jespersen
cycle hypothesis and in many more cases at least compatible with it, the clearest
evidence concerning the partitive negative polysemy, attested in Melanesia.
1. Introduction
* Thanks are due to Brenda Boerger, David Willis and Susanne Holzknecht. The research was
funded by the Research Foundation Flanders. The paper is partially based on Vossen (2011).
The maps in Figures 2 and 3 were made with the WALS tool 〈http://wals.info〉
Frens Vossen & Johan van der Auwera
In many languages clausal negation is expressed by more than one marker. Stan-
dard French is a case in point.
(1) Cléopâtre ne voyait pas le serpent.
Cleopatra neg saw neg the snake
‘Cleopatra didn’t see the snake.’
Whereas the meaning of ‘see’ and ‘snake’ is expressed with only one word, viz.
voyait and serpent, negation has two words¸ viz. ne and pas. This phenomenon
is well-studied, esp. from a language-specific point of view, e.g. that of French,
and we are making advances for the cross-linguistic study, esp. for Europe and
Africa. From both perspectives, textbook references are Jespersen (1917) and Dahl
(1979), with the latter coining the term ‘Jespersen cycle’ for the diachronic phe-
nomenon deemed responsible for generating a double negation. Here is the much-
cited quote from Jespersen (1917: 4):
The history of negative expressions in various languages makes us witness the
following curious fluctuation: the original negative adverb is first weakened, then
found insufficient and therefore strengthened, generally through some additional
word, and this in its turn may be felt as the negative proper and may then in
course of time be subject to the same development as the original word.
Van der Auwera (2009, 2010) shows that Jespersen’s idea has been very influential,
but also controversial, and, at least for French, it is argued that Jespersen’s account
is wrong. It is not the case that French ne was weak and needed reinforcement. As
a negative marker ne was fine, but the point is (i) that one often wants to emphasize
negation, as with du tout (lit.) ‘of all’ in modern French ne … pas du tout ‘not at all’
or with at all in not at all, (ii) that pas, literally ‘step’, once had that function, serving
as a ‘minimizer’ with movement verbs (the movement not extending even one step),
and (iii) that, over time, the emphatic meaning bleached and eventually turned into
an exponent of neutral negation. This kind of account was already expressed by
Meillet (1912). This paper is regularly cited for having introduced the term ‘gram-
maticalization’, but its second claim to fame should have been its clear position on
the development of negation and on its cyclical nature.1 Meillet (1912) is particu-
larly interesting, for he uses a concept much like ‘cycle’, but also a little different:
. For a correct account of French pas, Bréal (1897: 221) was earlier still and his claim that
‘[t]out le monde sait ce qui s’est passé pour les mots pas, point […]’ suggests that the view had
earlier proponents. But Bréal does not say anything about the cyclical nature.
The Jespersen cycles seen from Austronesian
Les langues suivent ainsi une sorte de développement en spirale: elles ajoutent
des mots accessoires pour obtenir une expression intense: ces mots s’affaiblissent,
se dégradent et tombent au niveau de simples outils grammaticaux; on ajoute
de nouveaux mots ou des mots différents en vue de l’expression; l’affaiblissement
recommence et ainsi sans fin. (Meillet 1912: 394 [1926: 139–140])
What both Jespersen’s and Meillet’s hypotheses share, however, is that they posit a
progression from single to double negation, and then back to single. Schematically:
(2) Single NEG1
NEG1 … NEG2
Double
Single NEG2
In French NEG1 is preverbal and NEG2 is postverbal. To the extent one can judge
from the languages of Europe and Africa, this is a strong tendency. This tendency
is so strong that some linguists consider it definitional for doubling (e.g. de Swart
2010). This is not the case for Jespersen (1917) or Meillet (1912): when they dis-
cuss the cycle, they do not mention word order facts at all. And Jespersen is par-
ticularly interesting, for one of his illustrations is Latin. In Latin, doubling does
not involve negative markers embracing the verb. Instead there is a univerbation
of NEG1 ne and the minimizer oenum ‘one thing’, yielding the NEG2 non. Never-
theless, the association of NEG1 and NEG2 with preverbal and postverbal posi-
tions, is crosslinguistically strong. We think that it relates to a principle for which
Jespersen (1917: 5) is credited too. This is the ‘NEG FIRST’ principle, which says
that speakers ‘place the negative first, or at any rate as soon as possible, very often
immediately before the particular word to be negatived [sic] (generally the verb
[…])’ (see also Horn 1989: 293, and for confirmation in large scale typology, Dahl
(1979: 91–92) and Dryer (1998: 102)). Since many second negators have a non-
negative origin and since the first negator will tend to occupy an early position
in the clause, it is no surprise that many second negators will follow the first
negators. And if a second negator is either a repetition of the first negator or if it
results from a right periphery answering particle – two other common sources,
it will also follow the first negator. Furthermore, at least in languages that do not
put their verb late in the sentence, chances are that the first negator comes before
the verb.
Interestingly, nothing will prevent the postverbal NEG2 language from start-
ing another cycle and NEG-FIRST from being relevant again. In that case, the new
negation, symbolized as ‘NEG3’, should prefer an early position and thus precede
NEG2.
Frens Vossen & Johan van der Auwera
Double
NEG1 V NEG2
Single V NEG2
NEG1 (… NEG2)
Double
NEG1 … NEG2
(NEG1 …) NEG2
Single NEG2
For another thing, these stages may be contemporaneous, in the sense that lan-
guages may assign different strategies to different constructions and different soci-
olinguistic settings – a situation convincingly described for French in Martineau
and Mougeon (2003). Thus in certain constructions modern French can still use
the archaic ne strategy (see e.g. Muller 1991: 227–245).
(5) Il n’ a cessé de pleuvoir depuis hier.
it neg1 has stopped of rain since yesterday
‘It has not stopped raining since yesterday.’ (Muller 1991: 230)
Or to illustrate the same point with an Italian dialect: in the village of Cairo Mon-
tenotte in Piemonte (Italy), the first single negation stage has ne¸ the doubling stage
has ne with nent (originally meaning ‘nothing’), and the second single negation
stage just nent, but in fact the three stages are contemporaneous in current speech.
The Jespersen cycles seen from Austronesian
negation is also found, Crowley knew about Lewo, and he expressed the hunch
that it was not that unusual:
Meanwhile, we know that Crowley was correct (see van der Auwera, Vossen & De
Vos 2013) and esp. the Bantu languages provide good illustrations. In fact, triple
negation was discussed as early as in 1958, and even for a European vernacular.
The relevant description was ‘hidden’ in a Dutch dialectological monograph on
the Belgian Brabantic dialect of Aarschot (Pauwels 1958). This book appeared at
a time when the term ‘Jespersen cycle’ had not been coined yet and the general
interest in this type of development had not flared up. In any case, Jan Lodewijk
Pauwels (1959) explicitly says that this dialect allows triple clausal negation, such
as illustrated in (10). 2
(11) Single en
en … niet
Double
Single niet
niet … niet
Double
Given now that the Aarschot dialect allowed en … niet … niet (as en … nie …
nie) the descriptions in Pauwels will have to be adjusted. This is what (12) does. It
shows that the renewed strengthening with the doubling of niet does not have to
wait until the old doubling strategy was simplified to single exponence and that it
can set in already at the doubling stage.
. Note also that NEG2 precedes NEG1. This is generally the case in Dutch subordinate
clauses.
The Jespersen cycles seen from Austronesian
(12) Single en
Double en … niet
way(s) a language has for negation declarative verbal main clauses’. Standard nega-
tion scopes over the entire simple clause. The negation in (13) says that (14) is false.
(13) John doesn’t like dogs.
(14) John likes dogs.
However, we include existential negation when the language negates normal main
clause verbs with an existential construction. This is illustrated with Banoni, a
Meso-Melanesian language spoken on Bougainville (Papua New Guinea). The
existential negation is illustrated in (21a). (21b) illustrates that this verb is also
used for negating present and progressive verbs.
(21) Banoni (Lynch & Ross 2002: 452, 449)
a. borogho ke ghinava
pig 3sg.r not.exist
‘There isn’t any pork.’
b. nna ghinava maa tai
he not.exist neg go
‘He isn’t going.’
For our purpose we count negators irrespective of whether they are verb, auxil-
iaries, adverbs, particles or affixes, which implies, for instance, that (21b) will be
analysed as a case of double negation. Finally, we exclude prohibition, i.e. the nega-
tion of the imperative, such as in (22).
. We also exclude negative concord constructions such as John doesn’t like nobody.
The Jespersen cycles seen from Austronesian
Cases such as Lewo (9) will thus be excluded here. Note though that there are
occasions where a grammar happens to give a non-SN example but where we
think that the strategy used there is identical to the SN strategy.
Finally, in one respect we differ from Miestamo (2005). A negation that is
otherwise fully ‘standard’ but that expresses emphasis is not included in his study.
In this paper, however, we do include ‘emphatic standard negation’, because it is a
driving force, if not the driving force, of the Jespersen cycle. This is a tricky matter,
however. What happens in the Jespersen cycle of French is that a negation is made
emphatic by a minimizer. Initially, the minimizer is not itself negative. When ne …
pas still meant ‘not even a step’, we don’t want to say the negation is double. Pas
then lost its ‘step’ sense and became negative, but there was arguably a period
when ne … pas was still emphatic but already negative. For this kind of construc-
tion, we do want to say that the negation is double.
The Austronesian language family is found in the southern part of the Pacific Rim,
and the islands in the Pacific Ocean, with one outlier near the east coast of Africa.
Malagasy, the westernmost language of the family, is spoken on Madagascar and
the Comoro Islands, along the coast of Mozambique. Rapanui, the language of
Easter Island, is the easternmost representative. From north to south, the family
stretches from Taiwan and Hawai’i to New Zealand. Except for Malay and some
smaller languages on the mainland, Austronesian is spoken on a great num-
ber of islands. There are larger islands such as Madagascar, New Guinea and
New Zealand, in the western and southern part of the area, and smaller ones, such
as Easter Island and islands of Melanesia or Micronesia, spreading out over the
southern and the eastern part of the Pacific Ocean. On New Guinea, Austronesian
is only spoken in some coastal areas. The map in Figure 1 shows some subgroups,
with an indication of the numbers of languages, based on the 16th edition of the
Ethnologue: (i) Oceanic, hundreds of languages (n = 506) spoken over a large
eastern area, which together with a much smaller number of languages of South
Halmahera West New Guinea (n = 41) are generally taken to form the Eastern
Malayo-Polynesian languages (n = 547), (ii) the Central Malayo-Polynesian lan-
guages (n = 169), which together with the Eastern Malayo-Polynesian languages –
and 2 further languages – constitute the Central-Eastern Malayo-Polynesian
group (n = 718), (iii) the remaining languages, on the map lumped together under
‘Western Malayo-Polynesian’ (n = 501) and ‘Formosan’ (n = 28). The total number
of Austronesian languages, according to the Ethnologue, is 1257.
Frens Vossen & Johan van der Auwera
JAPAN
CHINA
FORMOSAN
Talwan
0
20
Marianas Hawai
PHIL
IPPIN
Guam Federaled States
Palau Yap
Marshau of Micronesia
ES Islands
SOUTH HALMAHERA /
Kiribati Equator 00
Su
Borneo Nauru
at
Salawasi IRIAN
ra
Co
Rotuma Wallis Samoa FRENCE POLYNESIA
ok
MALAYO- CENTRAL & Futuna Tuamoto
Isla
POLYNESIAN MALAYO- Vanuatu Nitre
also Archipetago
nd
Fiji Tahiti 20
0
POLYNESIAN
s
MADAGASCAR New Tonga Rarotonga
AUSTRALIA
Caledonia
Rapa Easter
Island
NEW 0
ZEALAND 40
* Non-Austronesian languages in areas shown as Austronesian
0 0 0 0 0 0 0 0
100 120 140 160 180 160 140 120
These languages are not equally well described. In general, the languages
with the larger numbers of speakers have the better descriptions, and in Oceanic
the easternmost ones, typically just one language for one island, have at least one
description. For the area known as ‘Melanesia’, comprising (roughly) Fiji, New
Caledonia, Vanuatu, the Solomon Islands and New Guinea, Lynch et al. (2002: 21)
tell us that in 2002 only 10% had received a good description.
There are five survey works to date, on Austronesian languages or on the Oceanic
ones (see Table 1).
These language sets are not representative. Note also that the most wide rang-
ing study, Dryer (2011), tells us about 18 languages with double negation, and
though he looked for languages with more negation markers than two, he did not
find any. In this study we report on more languages and we also found more with
double negation, some with of triple negation, so more than Lewo and Nese, men-
tioned already, and even one with quadruple negation (see Table 2).
Again, we also do not claim any representativeness. We just used all infor-
mation on negation in Austronesian languages that we could find, being well
aware of the fact that many languages are not described and that what we did
Frens Vossen & Johan van der Auwera
find for any one language may not be complete. ‘Surveyed’ in Table 2 means that
we had s ufficient information on negation to analyze it in the terms of this paper.
Languages for which we found a description that did not tell us enough about
negation are not included.
Because neither this paper nor the survey works listed in Table 1 can deal with
a representative sample, it is hard to say whether double negation is common or
not. On the basis of our 409 languages, we are inclined to think that double nega-
tion is not rare.4 We will see though that it mostly occurs in two of its subfamilies
(see below).
The double negation in Table 2 can either be obligatory or optional. The two
types are distinguished in Table 3.
56 25 81
. For whatever it is worth, Miestamo (2005) has a sample, it has 17 Austronesian languages,
and of these 17 there are two with double negation, one with triple negation and one with
quadruple negation; the remaining 13 languages has single negation. This distribution sug-
gests that multiple negation is not rare. But we have to be careful. The language with triple
negation is Rapanui, Miestamo analyzes only one, the single negation, as ‘standard negation’.
The language with 4 negations – 3 in Miestamo’s analysis – is Lewo. It is in the sample largely
because it is well described, with an easily accessible article (Early 1994a), but, paradoxically,
the reason why Lewo was deserving of this article was its unrepresentativeness.
The Jespersen cycles seen from Austronesian
Rapa Nui
Moklen, Thailand
The Chamic family in South Vietnam and Cambodia
Area with nine scattered languages with optional double negation
Area with many languages with double negation
Direction of Easter Island (Rapanui)
In South Vietnam and Cambodia we are dealing with the genetic group of the
Chamic languages. The Ethnologue lists 10 mainland Chamic languages – there is
one more, viz. Achenese, but that is spoken on Sumatra. For 8 of these languages, we
have information on negation: 6 have double negation, viz. formal Eastern Cham,
informal Eastern Cham,5 Chru Roglai, Rade, and Jorai. Outside the Chamic family,
there are eight more languages in ‘non-Oceanic’ with a form of double negation.
They do not form an area or a family, and in none of them are the two negatives
obligatory. One language, Moklen, is spoken in Thailand, the other seven are scat-
tered over Indonesia and the Philippines: Biak, Fehan, Fordata, Paulohi, Kambera,
Muna, and Pendau. Table 4 lists all the languages, using the classification in the
Ehtnologue.
The Oceanic languages are listed in Table 5. We again use the Ethnologue. The
italicized languages Nese, Namakir, and Nemboi are not in the Ethnologue. On
the basis of the information provided in the descriptions, we have classified them
ourselves.
. Formal and informal Eastern Cham are varieties of one language, but the negative strategy
is so different that we count them as two languages.
Frens Vossen & Johan van der Auwera
That double negation is typical for Oceanic has been remarked before (Lynch,
Ross & Crowley 2002: 51–52; quoted with approval by Blust 2009: 466, 471), but
we can be more specific. 63 of the 65 Oceanic languages are spoken in Melanesia.
There are mapped in Figure 3.
Rapanui
301 27 328
Frens Vossen & Johan van der Auwera
56 25 81
Of the two tendencies, viz. (i) the tendency for one of the negative markers
to precede and the other one to follow, and (ii) the tendency for the preverbal one
to be the oldest one, the first one is relatively easy to check. In fact, we have the
relevant word order information for all of the 81 languages. First, for the 56 lan-
guages with obligatory double negation, as many as 54 languages do indeed have
one preverbal and one postverbal negation. Rotuman, a Central Pacific language
from the Island Rotuma, east of Fiji, illustrates this.
In the one language that does not conform to this tendency, both negators are
preverbal. The language is Tungak, a Meso Melanesian language on New Ireland
(Papua New Guinea).
For the 25 languages with optional double negation, the tendency to have one
negation in front of the verb and the other following holds for 19 languages. One
of them is the Chamic language Chru, spoken in South Vietnam.
On the second tendency, the one that has the preverbal negation as the oldest
marker, we have little information – but see Section 10 – yet given the fact that the
NEG FIRST tendency holds maximally, it would not seem too daring to assume that
the second tendency will also hold (and thus assume that these languages are analo-
gous to what we see in Europe). This assumption furthermore allows us to interpret
the 19 languages with optional doubling and with both a preverbal and a postver-
bal negation. Those languages in which the postverbal negation is optional will be
assumed to be between stages 1 and 2 of the Jespersen cycle (the version in (4)), the
languages in which it is the preverbal negation that is optional will be assumed to be
between stage 2 and 3, and the languages in which both are optional will be assumed
to show the contemporaneity of stages as shown in (7). The three types are listed in
Table 7. To symbolize our hypothesis that the preverbal negation is older than the
postverbal one, the first is referred to as ‘NEG1’ and the second as ‘NEG2’.
Table 7. Optional double negation, position relative to the verb (I)
NEG1 V (NEG2) (NEG1) V NEG2 (NEG1) V (NEG2)
11 5 3
The (NEG1) V (NEG2), i.e. the type in which both are optional – except of course
that at least one negation has to be present, is illustrated with Fehan, a dialect of
Tetun, a Central Malayo-Polynesian language of Timor.
The remaining 6 languages with optional doubling do not lend themselves to the
direct Jespersenian interpretation that we have just proposed. In 3 languages both
negatives precede the verb. Then there is a somewhat unclear case of a language in
which both exponents follow the verb, another unclear case in which both expo-
nents either precede or follow the verb, and finally there is also one language that
is said to have optional doubling negation, yet only in riddles, but the author does
not tell what position the negatives take up – this language is the Celebic language
Pendau, spoken in Sulawesi. On the assumption that NEG FIRST principle still
holds, we take the first one to be the oldest.
Table 8. Optional double negation, position relative to the verb (II)
(NEG1) (NEG2) V V (NEG1) (NEG2) (NEG1 NEG2) V Position
(NEG1 NEG2) unknown
3 ?1 ?1 1
The first subtype, (NEG1) (NEG2) V, is illustrated with Lau, a Southeast Solo-
monic language spoken on Malaita, Solomon Islands. It has two negative particles,
langi and si, and they can be combined. The source does not gloss the examples
and only shows the combined negation, not the single one.
The Jespersen cycles seen from Austronesian
For the second subtype, the illustration is not ideal. The language is Kaulong, a
New Britain language, as described by Ross (2002). We only have examples of
doubling with a special negative auxiliary akomen ‘cannot’ and the single expo-
nence illustration of the other marker, the particle som, and this example is not of
standard negation.
(31) Kaulong ( Ross 2002: 401, 401, 404)
a. eiak men ngi-n som
stone.axe delim tooth-3sg neg
‘The axe is not sharp.’
b. ngo ku mara-ngo ko-i akomen
1sg irr eye-1sg see-tr neg.abil
‘I cannot see it.’
c. ngo ku tulu-ngo som ta miuk akomen
1sg irr mind-1sg neg ben 2pc neg.abil
‘I will not forget you three.’
The one language in which the two negatives can both precede and follow the verb
is Fordata, a language of Maluku. It has single preverbal negation, expressed by
wol or wahoel. Then there is a strong negation, with fara ne, preceding or follow-
ing the verb. The trouble is that we don’t know whether fara and ne are negative
on their own.
(33) Fordata (Drabbe, 1926: 47)
a. ajaä wol oengrihi
isg neg 1sg.speak
‘I don’t speak.’
b. tomat’injai fara ne nangrihi/tomat’injai nangrihi
that.man neg neg speak / that.man speak
fara ne
neg neg
‘That man doesn’t speak at all.’
The claim about the relative age of the two negators in doubling is a hypothesis
purely based on the assumptions that doubling results from a Jespersen cycle and
that the NEG-FIRST principle is relevant for the first negator. However, as we have
made clear in Section 1, doubling may also be interpreted as a case of NEG3 …
Frens Vossen & Johan van der Auwera
NEG2: in that case NEG2 results from a first Jespersen cycle, NEG3 from a second
one, and NEG FIRST puts NEG3 earlier in the sentence. We have at least one lan-
guage that is interesting in this respect. Biak, an Eastern Malayo-Polynesian lan-
guage of West New Guinea, normally uses va as a postverbal negation. However, it
can combine this with a preverbal bukan. The latter is borrowed from Indonesian/
Malay (van den Heuvel 2006: 131), so it is most plausibly of later date.
What should be less common is the pair illustrated in (36). In Nagu, a Temotu
language of the Santa Islands (politically part of the Solomon Islands, but closer
to Vanuatu), the simple preverbal negator is te/tö/to, but as we can see in (35b),
repeated as (36b) that te marker is postverbal in Nese.
Table 9 lists some results. Since the probability that look alikes are true cognates
is higher when the languages are more closely related, for each marker, languages
belonging to the same family (of the depth shown in Table 4 and 5) are listed first,
separately, and marked with boldface, italics or underline. Within each family or
group languages are listed alphabetically.7
Preverbal Postverbal
. We do not include Biak and Malay bukan, for we know that Biak borrowed bukan from
Malay (see the previous section). We also exclude doubling languages in which the embracing
negator is the same. The North and Central Vanuatu language Bieria uses se on both sides of
the verb. So it would to have to be entered in both the ‘preverbal’ and ‘postverbal’ NEG V NEG
columns.
The Jespersen cycles seen from Austronesian
Preverbal Postverbal
Postverbal Preverbal
9. Emphasis
When, in a (typical) Jespersen cycle, the initial single negation doubles, it is for the
sake of emphasis. The original meaning then bleaches, and it becomes a negative
marker, but the emphasis may remain. We thus expect to find languages with a
doubling construction with an optional NEG that express an emphatic negation.
There is some evidence in five languages: Uripiv, a North and Central Vanuatu
language spoken in Malakula, Kambera, a Central Malayo-Polynesian language
spoken in the lesser Sunda Islands, Moklen, the one Malayo-Polynesian language
. Crowley (1982: 140–142) argues that the r- and t- forms are cognate.
The Jespersen cycles seen from Austronesian
of Thailand, and formal as well informal Eastern Cham from Vietnam, illustrated
in (37) to (41). On this occasion we will gloss the optional emphatic negative the
way the source linguists have done it. For Uripiv the gloss suggests that it is not
inherently negative, but in the text Early (1994b: 432) calls it a ‘NEG2’.
(37) Uripiv (Early, 1994b: 432)
ete nu-majing te lelingen
neg 1sg-work “emphasis” today
‘I’m not working today.’
(38) Kambera (Klamer, 2005: 723)
nda na=pi=a=nya ndoku Nipong
neg 3sg.nom=know=mood=3sg.dat neg.emph Japan
ma=tobu=nja
rel=slaughter=3pl.dat
‘The Japanese didn’t know [at all] (who) slaughtered them.’
(39) Moklen (Larish, 2005: 523)
cəy laʔ kɯtɛ:n hãh
1s emph.neg lie neg
‘I (am) not lying.’
(40) Formal Eastern Cham Lee (1996: 306)
urang lingiu ôh tamu’ hu dalǎm hum ô
person outside neg1 enter have inside shelter neg2 (intens.)
‘An outsider is not permitted in the shelter at all.’
(41) Informal Eastern Cham (Lee, 1996: 304)
nu’k di pǎng amek amu’ ô
child neg1 (intens.) listen mother father neg2
‘The child doesn’t listen to his parents at all.’
Note that whereas in Uripiv, Kambera and formal Eastern Cham, the emphatic
negator is postverbal – what one would expect – in Moklen and informal Eastern
Cham it is preverbal.
Fehan, already illustrated in (29), and repeated below, is interesting, too. In
this language there is a preverbal negative la and a postverbal one ha’i. The first
one is much more frequent. La can bear no stress, so when stress is needed, speak-
ers use ha’i. Therefore, the use of ha’i is more emphatic than la, but when both are
used, emphasis is stronger still.
(29) Fehan (Klinken, 1999: 229)
a. ha’u la k-bá
1s neg1 1sg-go
‘I’m not going.’
Frens Vossen & Johan van der Auwera
This is an interesting variation on the Jespersen cycle. In the French variant, the
addition of pas initially served to emphasize the negation, but there is no evidence
that pas by itself was more emphatic than ne by itself or that ne … pas was more
emphatic still. But like in French, the postverbal negation is more associated with
emphasis than the preverbal one.
Something similar is found in Muna, a South-Eastern Celebic language spo-
ken off the south-eastern coast of Sulawesi. Its present and past time negator is
miina. In unstressed negation it is preverbal, in emphatic negation postverbal and
in some cases they are both present.
(42) Muna (van den Berg, 1989: 63, 208, 208)
a. miina nae-ala kapulu
neg 3sg.I-take machete
‘He didn’t take a machete.’
b. maka a-(m)afa-ane? a-mbaraka(-a) miina
then 1sg.i-do.what-it 1sg.i-climb.well(-cl) neg
‘What can I do about it? I cannot climb well!’
c. miina bhe ka-lele-ha dua miina
neg be nom-cross-loc also neg
‘There was no place to cross either.’
Fairly direct evidence for assuming a Jespersen cycle concerns the origin of the
second negator. For Austronesian we usually lack direct historical evidence, but in
some cases we see the item that is a second negator in one language with a non-
negative meaning in another language or, rarely, with a non-negative meaning in
the same language. From the data in Europe and Africa we have now reached
The Jespersen cycles seen from Austronesian
In Paamese the relevant element can have both uses. Crowley calls it a partitive
negative, and it is used in transitive verbs with non-generic objects ((44a)) and
with intransitive generic verbs and non-verbal sentences ((44b)). Transitive verbs
with generic objects are negated by the preverbal negator ro alone ((44c)). Intran-
sitive verbs do not have an object, so ‘a part of ’ is not applicable here. In (44a) tei
can function as a partitive, but in the other examples that is not possible.
(44) Paamese (Crowley, 1982: 140, 141, 141)
a. kai rongadei veta
kaie ro-ngani-tei vetae
3sg 3sg.r.neg-.eat-neg2/prt breadfruit
‘He didn’t eat the breadfruit.’
b. būsi rogukulutei
buusii ro-gu-kulu-tei
cat 3sg.r-neg1-redup.swim-neg2
‘Cats don’t swim.’
Frens Vossen & Johan van der Auwera
A negator can also come from another language. We have mentioned the case of
Biak borrowing bukan from Malay (see (34)). We also see borrowing in Vietnam,
with the Cham languages sharing negators with the surrounding Mon-Khmer
languages. Possible correspondences are shown in Table 12.
There are four languages that show triple negation, one on the Solomon Islands,
two on Vanuatu, and one on Easter Island. And one of these even allows quadruple
negation. In Natügü, spoken on the Reefs Islands (part of the Solomon Islands),
double negation is obligatory, one marker is preverbal, a verb-initial prefix tö-, and
the other postverbal, which also encodes aspect, and the most common one is the
negative completive -u.
On Vanuatu we find Nese, spoken on Malakula Island, and Lewo, spoken on Epi
Island. Nese has a realis-irrealis alternation in the preverbal negative, with se being
for realis and be for irrealis.
(51) Nese (Crowley 2006: 70, 70)
a. ne-se-rij-te
1sg.neg1r-talk-neg2
‘I did not speak.’
b. rrise-be-yat-te
1pl.incl.imm.fut-neg1irr-stay-neg2
‘We will not stay’
The two NEG1 prefixes can be concatenated, thus filling the first slot twice, and as
a result, we get three markers.
No reason is given for the concatenation of the two pre-verbal negative markers.
There is no indication in the grammar that triple negation is stronger than dou-
ble negation. It is this structure that prompted Crowley to the hunch, quoted in
the introduction, that Central Vanuatu will have tripling languages. He explicitly
refers to Lewo, and indeed, it is there as well, but so far no other Vanuatu language
has been found with tripling.
In Lewo, then, we again find a realis/irrealis alternation in the preverbal nega-
tive, respectively pe and ve. In the realis, a third negative, po(li), is added in sen-
tence final position.
(53) Lewo (Early 1994b: 411)
naga pe Ø-pa re poli
3sg r.neg1 3sg.subj-Rgo neg2 neg3
‘He hasn’t gone.’
There is tripling in the irrealis as well, but in the probihitive, and since this is not
standard negation, we will not discuss it here (but see van der Auwera 2006 for a
Jespersen cycle discussion).
Interestingly, Early (1994b: 419) reports that young speakers tend to skip the
preverbal negative, which takes the structure back to a doubling construction,
with two postverbal negatives – and for this reason Lewo was also counted as
doubling language.
(54) Lewo (Early 1994b: 419) Young speakers
yuwa (pe) Ø-kove re po
rain neg1 2sg. subj-fall neg2 neg3
‘It didn’t rain.’
This makes it similar to an ordinary Jespersen cycle, which is a matter of adding
a marker, but also of deleting one, and even typically the preverbal one. But older
speakers do something that is not what we see in a normal Jespersen cycle. They
do what we find in Nese: they can fill the first slot with two negators. Since there
are still two additional negators, this process makes for a total of four negators.
(55) Lewo (Early, 1994b: 420) Older speakers
pe-re a-pim re poli
neg1-neg2 3sg.r.come neg2 neg3
‘They didn’t come.’
We have little to say about Rapanui, the language of Easter Island. It seems to be a
system different from anything we focused on in this paper. Negation is expressed
by the particles eko, ‘ina, kai, ta’e ‘o, and combinations thereof, i.e. ‘ina…eko, ‘ina…
kai, ‘ina…ko, ‘ina…’o, kai…’o and ta’e ‘o. Most of them are clause-initial. The combi-
nation of three negation markers expresses the strongest negation, and two nega-
tions are stronger than one (du Feu 1996: 89–92).
Frens Vossen & Johan van der Auwera
Note that even though we have only 4 languages, there is a great variation to
the position where tripling happens. In Rapanui it is at the left periphery of the
clause, which is also the place for single and double negators. In Lewo, it is at the
right periphery, a position that is different for the other negators. In Nese and
Natügü the tripling position is relative to the verb, preverbal in the case of Nese
and postverbal in the case of Natügü. The case that lends itself most easily to an
explanation in terms of a Jespersen cycle is Lewo tripling. For Lewo poli Early
(1994b: 432–433) suspects an original negative meaning: plausible cognates in
related languages mean ‘absent, non-existent’. So speakers may be assumed to
have added a negative word to their negative strategy, not unlike what speakers
of Brabantic have done (see (10) to (12)) – except that the B rabantic speakers
used the second negator again, and the Lewo speakers used a new one. And like
speakers of French, progressive speakers of Lewo can drop the first negator. There
is, however, no evidence that Lewo tripling has or had an emphatic effect: maybe
what prompted triple negation was a desire for clarity. Interestingly, for Brabantic
too, there is a suggestion that what prompts tripling is not emphasis but clarity
(van der Auwera 2009: 50–52). Whether or not one can subsume multiple expo-
nence that is due to a desire for clarity rather than emphasis under the notion
of Jespersen Cycle is, at least to some extent, a terminological matter. In van der
Auwera (2009) the answer is a positive one. If that is the appropriate answer, then
the other cases of tripling becomes Jespersenian too. For Natügü, a neighboring
vowel changes the phonetic substance of the second negator, and a third one can
be argued to make the negation clearer. Doubly filling the first negative slot in
Nese and Lewo, however, may be something else: maybe speakers are just con-
fused. For Rapanui, a Jespersen Cycle account has little promise, but we do see
the role of emphasis here, in a seemingly simple iconic way (the more markers,
the more emphasis) and since most of the negators are clause-intial, Rapanui also
illustrates the force of the NEG FIRST principle.
The Jespersen cycles seen from Austronesian
12. Conclusion
Abbreviations
References
Adelaar, Alexander & Himmelmann, Nikolaus P. (eds). 2005. The Austronesian Languages of
Asia and Madagascar. London: Routledge.
Bernini, Giuliano & Ramat, Paolo. 1992. La frase negativa nelli lingue d’Europa. Bologna: Il
Mulino. English revision as 1996. Negative Sentences in the Languages of Europe. Berlin:
Mouton de Gruyter.
Blancquaert, Edgard. 1923. Over de dubbele ontkenning en nog wat. Handelingen van het
Vlaamse Filologencongres 6: 60–69.
Blood, Evangeline & Blood Henry F. 2009. Vietnam Word List (Revised): Mnong Rolom. Ms.
Blust, Robert. 2009. The Austronesian Languages. Canberra: Department of Linguistics, the
Australian National University.
Boerger, Brenda H. In preparation. Natügu Grammar.
Breál, Michel. 1897. Essai de sémantique. Paris: Hachette.
Clark, Ross, 2002. Ifira-Mele. In The Oceanic Languages, John Lynch, Malcolm Ross & Terry
Crowley (eds), 681–693. Richmond: Curzon Press.
Crowley, Terry. 1982. The Paamese Language of Vanuatu. Canberra: Australian National
University.
Crowley, Terry. 2006. Nese: A Diminishing Speech Variety of Northwest Malakula (Vanuatu).
Canberra: Australian National University.
Crowley, Terry. 2002. Southeast Ambrym. In The Oceanic Languages, John Lynch, Malcolm
Ross & Terry Crowley (eds), 660–670. Richmond: Curzon Press.
Dahl, Östen. 1979. Typology of sentence negation. Linguistics 17: 79–106. DOI: 10.1515/ling
.1979.17.1-2.79
de Swart, Henriëtte. 2010. Expression and Interpretation of Negation. An OT Typology. Dordrecht:
Springer. DOI: 10.1007/978-90-481-3162-4
Drabbe, Petrus. 1926. Spraakkunst der Fordaatsche Taal. Eerste Stuk. ‘s Gravenhage: M. Nijhoff &
Batavia: Albrecht & Co.
Dryer, Matthew. 1998. Univerals of negative position. In Studies in Syntactic Typology [Typologi-
cal Studies in Language 17], Michael Hammond, Edith Moravcsik & Jessica Wirth (eds),
93–124. Amsterdam: John Benjamins.
Dryer, Matthew S. 2011. Order of negative morpheme and verb. In The World Atlas of Language
Structures Online, ch. 143, Matthew S. Dryer & Martin Haspelmath (eds), Munich: Max
Planck Digital Library. 〈http://wals.info/chapter/143〉 (22 December 2012).
du Feu, Veronica. 1996. Rapanui. London: Routledge.
The Jespersen cycles seen from Austronesian
Early, Robert. 1994a. Lewo. In Typological Studies in Negation [Typological Studies in Language
29], Peter Kahrel & René van den Berg (eds), 65–92. Amsterdam: John Benjamins.
Early, Robert. 1994b. A Grammar of Lewo, Vanuatu. Ph.D. dissertation, Australian National
University.
Fast, Lesley.1990. Tungak grammar essentials. In Two Grammatical Studies, John R. Roberts
(ed), 1–72. Ukarumpa: Summer Institute of Linguistics. 〈http://www.sil.org/pacific/png/
abstract.asp?id=24466〉
Flaming, Rachel. 1982. Cohesion in Wandamen narrative. Linguistic Studies in Indonesian and
Languages in Indonesia, Vol 15, 41–49. Jakarta: Badan Penuelenggara.
François, Alexandre. 2002. Araki. A Disappearing Language of Vanuatu. Canberra: Australian
National University.
François, Alexandre. 2003. La sémantique du prédicat en Mwotlap (Vanuatu). Leuven: Peeters.
Gregerson, Kenneth. 1979. Predicat and Argument in Rengao Grammar. Arlington TX: Summer
Institute of Linguistics.
Greninger, David Eugene. 2009. Vietnam Word List (Revised): Kơho Chil [čil]. Ms.
Heine, Bernd & Kuteva, Tania. 2003. On contact-induced grammaticalization. Studies in Lan-
guage 27: 529–572. DOI: 10.1075/sl.27.3.04hei
Hill, Deborah. 2002. Longgu. In The Oceanic Languages, John Lynch, Malcolm Ross & Terry
Crowley (eds), 538–561. Richmond: Curzon Press.
Horn, Laurence. 1989. A Natural History of Negation. Chicago: University of Chicago Press.
Hovdhaugen, Even & Ulrike Mosel (eds) 1999. Negation in Oceanic languages: Typological Stud-
ies. München: Lincom.
Ivens, Walter George. 1921. Grammar and Vocabulary of the Lau Languages, Solomon Islands.
Washington DC: Carnegie Institution of Washington.
Jespersen, Otto. 1917. Negation in English and Other Languages. København: A.F. Høst & Søn.
Klamer, Marian, 2005. Kambera. In The Austronesian languages of Asia and Madagascar,
Alexander Adelaar & Nikolaus P. Himmelmann (eds), 709–734. London: Routledge.
van Klinken, Catharina Lumien. 1999. A Grammar of the Fehan Dialect of Tetun, an Austrone-
sian Language of West Timor. Canberra: the Australian National University.
Larish, Michael D. 2005. Moken and Moklen. In The Austronesian Languages of Asia and
Madagasca, Alexander Adelaar & Nikolaus P. Himmelmann (eds), 513–533. London:
Routledge.
Lee, Ernest W. 1996. Bipartite negatives in Chamic. Mon-Khmer Studies 26: 291–317.
Lynch, John, Ross, Malcolm & Crowley, Terry (eds). 2002. The Oceanic Languages. Richmond:
Curzon Press.
Lynch, John, Ross, Malcolm & Crowley, Terry. 2002. Typological overview. In The Oceanic Lan-
guages, John Lynch, Malcolm Ross & Terry Crowley (eds), 34–53. Richmond: Curzon Press.
Lynch, John & Ross, Malcolm 2002. Banoni. In The Oceanic Languages, John Lynch, Malcolm
Ross & Terry Crowley (eds), 440–455. Richmond: Curzon Press.
Martineau, France & Mougeon, Raymond. 2003. A sociolinguistic study of the origins of ne
deletion in European and Quebec French. Language 79: 118–172.
Meillet, Antoine. 1912. L’évolution des formes grammaticales. Scientia 12: 384–400. Reprinted
in Meillet, Antoine. 1926. Linguistique historique et linguistique générale : 130–148. Paris :
H. Champion.
Miestamo, Matti 2005. Standard Negation. The Negation of Declarative Verbal Main Clauses in a
Typological Perspective. Berlin: Mouton de Gruyter.
Frens Vossen & Johan van der Auwera
Edith Pineda-Bernuy
Australian National University
* I wish to thank Cynthia Allen for the invaluable feedback and our discussions on this paper.
I would also like to acknowledge my several Quechua speaking collaborators who patiently
helped me to validate my data, as well as the audiences at my presentation at the Centre
for Research on Language Change (Australian National University), at the 20th International
Conference on Historical Linguistics (Osaka) and at the 3rd Conference of the European
Network for the Study of Andean Languages at the Max Planck Institute (Leipzig). I am
especially grateful to Manuel Delicado, Willem Adelaar, Rodolfo Cerrón-Palomino, César
Itier, Liliana Sánchez, Johan van der Auwera, Frens Vossen and the two anonymous reviewers
for their outstandingly useful comments, objections, questions, discussions and hints on spe-
cific aspects of my proposal which, taken together, have helped me to shape this paper. I am
the only responsible for any mistakes.
** The main ideas expressed in this paper are drawn from my doctoral dissertation, which is
in preparation, and where they are examined in more depth.
Edith Pineda-Bernuy
1. Introduction
negation in three extensive Quechua speaking areas. Section 5 deals with the dia-
chronic approach to the development of standard negation in Quechua under the
Jespersen Cycle. Section 6 provides an overview of the sources of negation renewal
in Quechua. Section 7 finalises the paper with some conclusions. The more expan-
sive sections are required to adequately describe a topic not previously addressed
in relation to Quechua and to support the main claims of this paper.
Finally, I advise that I have not addressed the influence that the Quechua-
Aymara contact could have in the development of negation in Quechua due to the
limitations of the objective of this paper.
The only exception to this general rule is in Ancash Quechua, which has the suffix
-ku for interrogation and the clitic -tsu as the main negation marker of sentential
negation. See the examples from Huaraz (Ancash) (HZ):
The development of standard negation in Quechua
COLOMBIA
Quito
NORTHERN
ECUADOR
QUECHUA
South
America
PERU
QUECHUA
PERUVIAN
NORTH
BRAZIL
QUECHUA
CENTRAL
Lima
Cuzco
BOLIVIA
SOUTHERN
QUECHUA
Central La Paz
Aymara: Pac
Jaqaru ific
O cea Cochabamba
n
Southern
Aymara
CHILE
LEGEND
Quechua language family
Santiago
Aymara language family
del Estero
0 1000
Kilometres ARGENTINA
. Sánchez (2010) describes the suffix -chu in Cuzco Quechua as a syncretic morpheme able
to express focus and negation or focus and interrogation. Muysken (1995) identifies the suffix
-chu as a non-factual marker.
The development of standard negation in Quechua
variation, which concerns the internal variation of a basic pattern due to the
specialised functions of mana and the suffix -chu in different types of negation
within one variety. For this purpose, the concept of standard negation is indis-
pensable (see next Section 3.2). The sentences in (2), (4) and (6) are instances of
standard negation. My assumption here is that the current variation is a result
of the diachronic development of standard negation in the Quechua language
family.
NEG1, NEG2, NEG3 and NEG4 indicate some of the types of negation pat-
terns found in Quechua. Equivalences among the forms per column are approxi-
mated. NEG1 shows that all dialects have the same bipartite negation pattern
(disregarding their function in the specific area); NEG2 also shows bipartite nega-
tion, but with an alternative form for mana (appearing in certain contexts), NEG3
lists the minimum forms to express simple sentential negation (standard negation)
and NEG4 presents some (bipartite) forms for emphatic negation.
In the literature, it has been widely reported that all Quechua variet-
ies present both morphemes mana…-chu in the formation of sentential nega-
tion (Cerrón-Palomino 1976; Quesada 1976; Parker 1976; Coombs et al. 1976;
. The suffix is -chu in Junín-Huanca Quechua and -su in Cajatambo Quechua.
Edith Pineda-Bernuy
Cusihuaman 1976; Soto 1976; Adelaar 1986; Weber 1996; Nardi 2002; Cole
1985). Alternations of this assumed main pattern consisting of only mana or only
-chu have been randomly noticed mostly in unspecified contexts, and recorded
in a number of Quechua grammars (cf. Adelaar 1977; Cerrón-Palomino 1976;
Coombs et al. 1976; Parker 1976; Weber 1996; Soto 1976; Muysken 1977). The
reconstruction of the Quechua negation pattern is based on the presence of these
two morphemes across Quechua varieties, but not on the acknowledgement of a
number of current negation patterns.
Negation is cross-linguistically expressed by means of several strategies, and
this is reflected in Quechua languages as well. Types of negative sentential patterns
are due to a number of factors, including: sentence structure, emphasis, focus,
doubt, archaic strategies (diachronic evolution), etc. Drawing on Payne (1985),
I identify four general types of Quechua negation, introduced here in opposite
pairs:
The oppositions are based on structural characteristics of the sentence and the
scope of negation which is related to the type of structure affected by negation:
. This example (5) is from Imbabura Quechua (Ecuador), a Peripheral Quechua variety (Cole 1985: 87).
. This sentence is from Cuzco Quechua (Lira 1941: 110), but currently Quechua speakers from this region do not know the deriva-
tive use of -mana.
. Weber (1996: 450) says that in this sentence the subordinator -na ‘irrealis’ makes the clause a negative because it expresses that
the event has not occurred.
The development of standard negation in Quechua
Edith Pineda-Bernuy
3.3 O
verview of negation patterns for sentential negation in the Quechua
language family
Considering the notion of standard negation, the Quechua language family pres-
ents four patterns, as shown in first column of Table 3. Non-standard negation
includes strategies for emphatic negation and/ or for focus of negation (second
column). The third column of Table 3 presents enclosing negation with shortened
forms of mana. See these details in Table 3:
Table 3. Types of standard negation and non-standard negation in the Quechua
language family
Regional varieties Standard negation Non-Standard negation patterns
The reader might expect that varieties with bipartite negation could have
non-standard sentential negation expressed by mana alone and perhaps even by
-chu alone. However, it should be observed that in these varieties, mana alone
cannot express sentential negation. Mana is the negation marker in subordinate
clauses across Quechua varieties. The lone suffix -chu does not express negation
since it is the interrogative marker of polar questions in this large dialectal group.
When its meaning is negative, it occurs in limited contexts and can be found in the
Ayacucho dialect (see also Section 4.2.1). Southern Quechua expresses sentential
negation with predominantly bipartite negation.
The development of standard negation in Quechua
4. V
ariations in the standard negation in the Quechua
language family
. San Martin Quechua is considered in this group because negation is made with only
mana pre-posed to the verb in simple sentences (cf. Coombs & Weber 1976: 186).
. From this group, Otavalo Quechua (Ecuador) is the only dialect in which single nega-
tion is less common than double marking. In this case, the suffix -chu is typically attached
to the verb.
Edith Pineda-Bernuy
4.1.2.1 Focus of negation. Here I mention some strategies to indicate the focus
of negation. One is enclosing the negative phrase with [+focus] by mana…-chu.
This pattern is used in most EK varieties, including Napo. Observe the wh-word
question in (11) and the answer in (12):
(11) ¿Pi-tak latu-ta paki-rka? (NP)
Who-int dish-ac break-pst.3
‘Who broke the dish?’
(12) Mana ñuka-chu ka-rka-ni. Pedro-mi paki-rka. (NP)
neg i-neg be-pst-1 Pedro-as break-pst
‘It wasn’t me. Pedro broke it.’
In (12), the clitic -chu expresses the focus of negation in contrast with -mi in
Pedro-mi of the positive sentence. In this context, the lack of -chu makes the sen-
tence ungrammatical (cf. Weber & Thiesen 1976: 16).
The development of standard negation in Quechua
In SE, the focus of negation can be shown by fronting the verbal phrase, plus
attaching the clitic -chu to the verb. These strategies to answer a question are
optional since the application of -chu can also be understood as emphasis, irony or
incredulity. See the answer in (14).
(13) Pedro-chu llamka-n?(SE)
Pedro-int work-3
‘Does PETER work?’
(14) Mana llamka-n-chu Pedro. (SE)
neg work-3-neg Pedro
‘Peter DOES NOT work.’
4.1.2.2 Emphatic negation. Here, the strategy is the addition of suffix -chu and/
or subject displacement. Emphatic negation is evident in existential sentences in
Napo Kichwa. Positive and negative existential sentences display the subject at
initial position.
Sentences with emphatic negation show the verbal phrase raising (in bold) in
addition to the suffix -chu as in (17).
. Notice that this method is not accepted by all of the Kichwa speakers in Ecuador.
Edith Pineda-Bernuy
SE only uses verbal phrase raising to indicate negative emphasis in sentences with
predicative verbs. This syntactic strategy applies to positive sentences as well.
Observe the emphasis in the negative statement in (18):
Like Napo, all the other EK varieties have similar constructions where the clitic
-chu is optional according to the strength of the advice or prohibition:
SE has this pattern as well, but apparently it is more unusual than in Napo Kichwa.
Other contexts which are not covered here are the suffix -chu in negative sentences
in the future tense and other pragmatic uses.
1 Declaration +
2 Future +
3 Focus + + +
4 Emphasis + + + +
5 Prohibition +
6 Request +
7 Warning +
8 Advice +
9 Assertion + + +
10 Conviction +
The development of standard negation in Quechua
Similarly, Solís & Chacón (1989) include negative sentences of Ayacucho Quechua
having only -chu as the main negation marker. As shown by the following example,
there is no topicalizer -qa and there is no indication of which factor prevents the
sentence from being interpreted as interrogative:
(34) Pay yaku-ta mayu-manta apamu-rqa-chu. (AY)
He water-ac river-abl bring-pst-neg
‘He didn’t bring water from the river.’
(Solís & Chacón 1989: 186)
While Quechua speakers from Ancash and Junin consider the suffix alone as
sufficient to indicate negation in the sentence (see Section 4.3.1), my Ayacucho
Quechua speaking collaborators recognise negative sentences with both markers
(mana-m and the suffix -chu) and rejected those with just the suffix. My interpre-
tation of these data is that single negation is an innovation in this variety. This
makes Ayacucho Quechua a variety with a more evolved standard negation in
comparison with the rest of this large dialectal group. I also conclude that, since
native speakers vary in their judgments, their opinions should not be a primary
factor in determining standard negation.
There are two Central Quechua dialects with bipartite negation that present
variations of this main pattern. Huallaga Quechua (Huánuco) is similar to the
case of Ayacucho Quechua regarding the sporadic drop of the negation operator
mana, but there is no further elaboration in this point (cf. Weber 1996: 55–56).
The other dialect is Pacaraos, characterised by “the tendency to eliminate the s uffix
-su ‘interrogative-negative’ more than mana”, according to Adelaar (1986: 78).
It follows that mana on its own can be the main negation marker. Adelaar also
reports that mana occurs in verbal predicates followed by the suffix -pa (*-pas)
‘additive’, but does not provide the context for -su alone as negator. Huallaga and
Pacaraos Quechua do not possess another polar interrogative marker other than
-chu or su.
The remaining dialects of this large group (those spoken in North Peru) pres-
ent bipartite negation with some variations as well but in non-standard negative
sentences.
4.2.2 N
on-standard sentential negation in Southern Quechua, some central
dialects and Northern Peruvian Quechua
As discussed in Section 4.2.1, standard negation is bipartite in this Quechua
group. Non-standard negation sentences also contain bipartite negation, but are
characterized by displaying additional suffixes and/or structural modifications
to express supplementary information related to negation. If the suffix -chu were
eliminated, the negative sentence would become non-sentential. If mana were
The development of standard negation in Quechua
4.2.2.1 Focus of negation. This feature is marked with strategies almost identi-
cal to those observed in the Peripheral Quechua varieties (see Section 4.1.2.1).
First, mana and the suffix -chu enclose the noun phrase, indicating the focus of the
negative answer as in (36). There is verbal deletion but the sentence is contextually
understood. This strategy is one of the possible negative answers to a wh-word
question as (35):
(35) ¿Pi-taq p’aki-ra-n latu-ta? (AQP)
Who-contr break-pst-3 dish-ac
‘Who broke the dish?’
(36) Mana ñuqa-chu. Aswanpis Pedro p’aki-ra. (AQP)
neg i-neg. On the contrary, Pedro break-pst
‘It wasn’t me. On the contrary, Pedro broke it.’
A second possible strategy is constructed with the topicaliser -qa (instead of -chu)
attached to the verb in a full sentence, with mana heading the phrase. This is
shown in (37):
(37) Mana ñuqa-qa p’aki-ra-ni-chu latu-ta.(AQP)
neg i-top break-pst-1-neg dish-ac
‘It wasn’t me who broke the dish.’
In Ferreñafe Quechua, sentences can drop the clitic -chu when the subject is
marked with the topicalizer -qa as in (37), so that these sentences are negated only
with mana. The reader will observe that this is the reverse of what happens to simi-
lar sentences in Ayacucho Quechua (see Section 4.2.1) where mana is dropped,
and the clitic alone marks negation.
A similar structural effect of emphasis in the sentence was observed from a speak-
ing collaborator from Cuzco Quechua. Some Quechua speakers (mainly from
urban areas) might consider this emphasis to be incorrect.
Another way to express emphatic negation is with the suffix -puni ‘certainty’
plus -mi (n in Cuzco) ‘assertive’ when they are attached to mana. In this context, the
statement increases the strength of the negative declaration as definitive, as in (39):
(39) ¡Mana-puni-n pay-qa hamu-n-chu kay-man-qa! (AQP)
neg-crt-as he-top come-3-neg here-dir-top
‘HE never comes here!’ (‘I’m convinced that…’)
4.2.2.4 Negation and evidentiality. When negation interacts with the evidential
-mi ‘assertive’, this may result in emphatic negation as seen in (40). This can occur
only if the evidential -mi is attached to ama or mana. In the case of the combina-
tion mana-mi, the speaker’s assertion expressed through the evidential makes the
negated expression stronger, as the speaker’s declaration is his personal convic-
tion. See Example (41)
(41) Mana-mi kama-ka-n-chu.(HGA)
neg-as order-psv-3-neg
‘It is not fair.’ (‘I’m convinced that…’)
(Weber 1996: 444)
One of the best known meanings of the evidential -mi is attested evidence by
the speaker, who makes assertions based on his experience, events depicting
facts, or on his (general) knowledge (cf. Cerrón-Palomino 1987: 287–288, Faller
2011: 663–664). If so, the question is what an assertive evidential like -mi could
mean in sentences like (40) and (41) that do not describe facts. In these cases,
its function is validational, clearly more related to the expression of the speaker’s
conviction in his own statement (cf. Cusihuamán 1976: 240–241, Adelaar 1977: 79,
Solís & Chacón 1989: 150; Weber 1996: 546–548). This is approximately Nuckolls’s
(2008: 84) description of -mi as ‘assertion of speaking self ’.
The development of standard negation in Quechua
Quechua dialects present the interaction between negation and the evidential
-mi in a great variety of forms. While most dialects with double marking negation
incorporate the evidential -mi to strengthen the negative statement, some dialects,
such as Ayacucho, have this evidential integrated in the adverb, so that mana is
mana-m. This development is relevant to further understanding the evolution of
the standard negative pattern in Quechua.
The contrast between an affirmative sentence and a sentence with standard negation
is shown in (42) and (43). In Junín-Huanca Quechua (J-H), spoken in Huancayo,
Concepción and Jauja, standard negation -simple negative sentence according to
Cerrón-Palomino’s (1976: 106–107) terminology- is built with only the suffix -chu
attached to the verb:
(42) Chay kwintu-kaq-ta uyali-lqa-a-mi. (J-H)
That story-det-ac listen-pst-1-as
‘I listened to that story.’
(43) Chay kwintu-kaq-ta uyali-lqa-a-chu. (J-H)
That story-det-ac listen-pst-1-neg
‘I didn’t listen to that story.’
(Cerrón-Palomino 1976: 106–107)
Similarly, the suffix -tsu or -su is the main operator in standard negation in
Ancash Quechua and in the neighbouring regions of Huanuco (Dos de Mayo) and
Cajatambo. Most informants of this area are unable to distinguish the differences
Table 5. Summary of sentential negation strategies in varieties with bipartite standard negation
Edith Pineda-Bernuy
SN Non-Standard negation
NEGATIVE mana… -chu mana… -chu mana-n… mana-puni ama…-chu ama-m… fronting inversion mana…-qa
-chu -m…-chu -chu -chu mana -V-chu
1 Simple declaration +
2 (Future)
3 Focus + (+) + +
4 Emphasis + + + (+) +
5 Prohibition + +
6 Request +
7 Warning +
8 Advice +
9 Assertion + + +
10 Conviction + + +
The development of standard negation in Quechua
between sentences with single negation and bipartite negation. However, they are
confident assuming that the clitic is enough to indicate negation.
It appears that there are more conservative areas within this region, evi-
denced by the more frequent use of double marking in Chacas and Vicus Quechua
(Ancash) compared to Huaraz (Ancash).
4.3.2 N
on-standard sentential negation in central varieties having standard
negation with the clitic
These Quechua varieties employ bipartite marking in a number of contexts in
which the clitic is essential in all instances of sentential negation, while manam
can even be optional in the non-standard form.
There are some similarities with the other two large varieties (Peripheral and
Southern Quechua), but also important differences. While Peripheral varieties
incorporate the suffix and central varieties incorporate the adverb in the non-
standard negation, the contexts of these additions are not paralleled.
4.3.2.1 Focus of negation. The reader will observe that all of the strategies to
express focus of negation are based on the clitic (-chu, -tsu or -su), but the adverb
mana-m is optional. Any sentential constituent can be the focus of negation. The
question in (44) has focus on the direct object. The answer in (45) has focused on
this constituent by enclosing and also fronting the affected phrase:
(44) ¿Wamra-a-ta-tsuraq José rika-rqa-n? (CHS)
Child-1pos-ac-int.dub José look.after-pst-3
Has José looked after my child?
(45) (Mana-m) wamra-yki-ta-tsu José rika-rqa-n.(CHS)
neg-as child-2pos-ac-neg José look.after-pst-3
‘It wasn’t your child that José looked after.’
There are two other possible ways of expressing focus of negation. In the most
common method, focus is marked in situ, so that the phrase with [+focus] remains
in the original syntactic position of the phrase. In the alternate method, mana-m
can be omitted, so the phrase with [+focus] remains unenclosed. It is reasonable
to suppose that fronting the phrase for focus of negation could be an indication of
emphasis, the same as the addition of manam.
4.3.2.4 Negation and evidentiality. The interaction between negation and the
evidential -mi has two effects in the sentential negation. First is the emphatic
feature that the clitic -mi impregnates in mana as in (48). The second effect is
the assertive feature that -mi transmits to the whole negative sentence as in (49).
In this case, the clitic -mi replaces the suffix -su to indicate negative assertion in
agreement with mana:
(49) Nuqa mana gusta-a-mi marka-a-pita ka-q papa-kuna-ta.(TCS)
i neg like-1-as town-pos1-abl be-ag potato-pl-ac
‘I don’t like the potatoes grown in my town.’
Similarly, Pacaraos Quechua occasionally presents sentences with the evidential -mi,
but attached to the adverb mana and without the clitic -su. It seems that the evidential
The development of standard negation in Quechua
adds assertive feature rather than emphatic strength. The type of interaction mana…
-mi is more common in Peripheral Quechua (see further Section 4.1.2.4).
4.3.2.5 Sentential negation without the suffix -chu or -tsu. There are several cas-
es of negative sentences presenting mana as the sole marker of negation. It would
appear that some of them are frozen ‘compounds’ of mana plus a clitic: manapis,
manataq, manaqa, etc. Most of these combinations of mana require that the main
verb present the conditional -man:
(50) Mana-pis aywa-shaq. (CHS, HZ, JH)
neg-ad go-1fut
‘Maybe I will not go.’
(51) Qam mana piña ká-nki-man. (TCS)
You neg angry be-2-cond
‘You shouldn’t be angry.’
(52) Mana-taq watuka-nki-man. (CHVN)
neg-contr visit-2-cond
‘I’m warning you about not visiting him.’ (‘You should visit him.’)
Also, the prohibitive adverb ama can express warning in combination with the
conditional -man, but without the suffix -chu:
(53) Ama miku-yku-rqu-nki-man. (TCS)
proh eat-dir-rp-2-cond
‘Hey, don’t start eating.’
Ni-mu-n-taq-chu chay-naw.
Say-dir-3-contr-neg that-comp
‘It doesn’t say that!’
(Weber 1996: 452)
Table 6. Summary of sentential negation patterns in varieties with clitic for single standard negation
Edith Pineda-Bernuy
SN Non-Standard negation
NEGATIVE -tsu mana… mana… mana-mi… mana-pis… mana-taq… mana… ama… ama… fronting inversion -taq-ku
-mi -tsu -tsu (-tsu) V-man V-man V-man -tsu -tsu manam
Simple
+
1 declaration
2 (Future)
3 Focus (+) + + + + +
4 Emphasis + + + + + + + +
5 Prohibition (+) +
6 Request + +
7 Warning + +
8 Advice + +
9 Doubt +
10 Assertion + + +
11 Conviction +
The development of standard negation in Quechua
COLOMBIA
OT KEY to Place Names
ECUADOR Quito
TNG NP ANC Ancash
NORTHERN
CHBZ
QUECHUA
AQP Arequipa
PTZ
South AY Ayacucho
America CJM Cajamarca
FF SM
CJM BRAZIL CHVN Chavin
CRG CHBZ Chimborazo
CHS
VS LL CRG Corongo
HZ CHVN
TCS CZ Cuzco
QUECHUA
Ancash
CENTRAL
HGA Ecuadorian
CJT EK
PCR TMA Kichwa
Lima HYO FF Ferreñafe
HCA
AY Cuzco HGA Huallaga
Central La Paz
HZ Huaraz
Aymara:
Jaqaru Pac J-H Junín-Huanca
ific
Oc Cochabamba LL Llata
ean
NP Napo
OT Otavalo
PCR Pacaraos
Southern PTZ Pastaza
Aymara Santiago del
SE
Estero
Patterns of standard negation
in the Quechua language family SM San Martín
CHILE
TMA Tarma
Single negation: -chu ~ -tsu ~ -su Tercero
TC
Catecismo
Bipartite negation: mana...-chu TCS Ticllos
Santiago TNG Tungurahua
Single negation: mana del Estero
VS Vicus
0 1000
ARGENTINA
Kilometres
Figure 2. Map of geographical distribution of patterns of standard negation in the Quechua
language family
Edith Pineda-Bernuy
EK speakers do not add the suffix -chu to a sentence as (55). Some EK and SE
Quechua speakers shorten mana as na and má ~ maa, respectively, in this context.
Solís & Chacón (1989: 188), on the other hand, describe these types of sentences
only with the suffix -chu in an Ayacucho variety that displays a single negation
pattern:
(57) ¿Pi-taq yaku-ta mayu-manta apamu-rqa-chu?(AY)
Who-int water-ac river-abl bring-pst-neg
‘Who didn’t bring water from the river?’
(Solís & Chacón 1989: 188)
Solís and Chacón’s (1989: 186–188) account is consistent with their assertion that
Ayacucho Quechua forms negative sentences only with -chu. As discussed in 4.2.1,
single negation could be an innovation in this Quechua variety.
The development of standard negation in Quechua
4.4.2.2 Central Quechua. Both Huallaga and Pacaraos Quechua present bipar-
tite negation and this pattern is required when adding negation in the wh-inter-
rogative sentence.
(58) ¿Pi-m mana miku-q-s-áw?12(PCR)
Who-as neg eat-hb-neg-tag.q
‘Who didn’t use to eat (anything)?’
(Adelaar 1986: 84)
Other Quechua dialects of this group, Cajamarca and Ferreñafe, present double
negation marking mana…-chu without the clitic -mi.
The only apparent exceptions to this are Tarma and San Pedro de Cajas
(Central Quechua). Adelaar (1977: 78, Note 2) notices that San Pedro de Cajas
and Tarma, which are described as bipartite negation dialects, occasionally present
single negation with chu. In Tarma in particular, Adelaar observes this exclusively
in wh-word negative questions. One explanation for this situation is that Tarma
and San Pedro de Cajas have, indeed, the single negation pattern marked with
-chu, and it is therefore reasonable to suppose that the occasional single negation
marking and the wh-interrogative negative sentence are indications of that. Tarma
and San Pedro de Cajas are Quechua speaking locations neighbouring the Junín-
Huanca Quechua area which presents single negation pattern with chu, meaning
that influence from dialectal contact is possible. In addition, Tarma is part of the
Yaru dialectal group (Torero 1974: 26–27) along with Cajatambo and Pasco, dia-
lects with single negation marking.
In all of these Quechua dialects, speakers acknowledge that they could add mana
to the wh-interrogative negative sentence, but it is unnecessary except for adding
emphasis to the negation. On the contrary, if the suffix is omitted the sentence is
not negative.
. Notice that in this large Quechua group, Pacaraos Quechua is the only one that has
reduced the suffix -su to -s before a vowel.
Edith Pineda-Bernuy
As seen in the preceding section, there are three main standard negation patterns
in the Quechua language family:
Type I: mana (single negation)
Type II: mana(-m)…-chu (bipartite negation)
Type III: -tsu (single negation)
An interpretation of this situation is that these three patterns represent the diver-
sification of a main standard negative pattern over time in Quechua, where suc-
cessive splits occurred at different stages and in diverse regions. Some regions,
such as central and Peripheral Quechua speaking areas, preserve microvariation
entailing two or more negation strategies, where at least one of them is standard
negation and the others are non-standard sentential negation. It is also possible to
observe the coexistence of more than one standard pattern, somewhat similar to
the case of the Ayacucho dialect, a likely transitional area between bipartite nega-
tion and single negation with the clitic –although it is always possible to recog-
nise one main form among standard patterns. In addition, it appears that Central
Quechua could be another transitional area between the oldest pattern of single
negation with mana and bipartite negation. From a diachronic point of view, that
coexistence may also be an indication of a transitional period between compet-
ing negation patterns towards a main standard pattern. Generally a non-standard
negation form can, at some point of the development, turn into standard negation
by yielding to the renewal of the negation pattern strategy.
on. It also means that the sole clitic -chu is the most developed form of standard
negation.
(60) Stage 1 > Stage 2 > Stage 3
mana mana…-chu -chu > -tsu > -su
5.1.2 From double negation marking to single negation: Either mana or -chu
The starting point was double negation marking, similar to that of Southern
Quechua. Central Quechua languages then dropped mana, while Peripheral
Quechua dropped the clitic -chu. This assumption appears in most Quechua
grammars and studies on Quechua.
(61) Stage 1 > Stage 2
mana…-chu -chu > -tsu > -su
-mana
As van der Auwera (2009: 35–66) observes, Jespersen’s ideas focused on the for-
mal properties of negation, involving stages of weakening and strengthening of
the negative marker(s). Van der Auwera’s (2009) analysis of Jespersen’s account
led him to identify various possible types of Jespersen Cycle, and alternative
ways to understand the renewal of negative exponents in languages. In the case
of Quechua, I consider phonetic erosion did not in any way cause the renewal
of negation. On the contrary, the preverbal negator mana evolves incorporating
clitics that strengthens its negative meaning. Neither the renewal could be caused
by the phonetic erosion of the clitic -chu (> -tsu > -su) that has become the main
negator in Quechua I. The characteristics of these negation markers must be due
to the nature of the source of the renewal, which is a non-factual marker that
seems not to have been an original strengthener.
The Negative Cycle in Quechua is marked by the stages of the grammaticaliza-
tion of the non-factual marker -chu. As a part of the negative sentence, this clitic
first became an emphasizer, then it turned into an obligatory part of the bipartite
negation. In this context, it acquired negative meaning, as it is interpreted as the
. Van der Auwera (2009: 42–43 and Footnote 3) notes that before Jespersen’s account
(1917), other scholars, including scholars of Egyptian, Coptic, Berber and Arabic as well as
Gardiner in 1904, Meillet in 1912, had developed similar ideas on the evolution of negation.
Edith Pineda-Bernuy
bearer of negation in the sentence. This change pushed the original negator manam
out of the neutral sentence. Mana-m is the component of the bipartite negation,
considered too strong as to indicate simple negation (standard negation). There is
not semantic bleaching of the original negator manam as this remains as part of
the emphatic expression of sentential negation.
The following statements based on van der Auwera (2010)’s discussion are
considerations included in the analysis of the Negative Cycle in Quechua:
The stages here include the strategies of standard negation and non-standard
negation. The model of changes has more stages than those outlined in 5.1, and
this help to visualize the development of the standard patterns more realistically.
Suffixes other than -chu and -mi are part of the sentential negation, but they do not
take part of the cycle.
5.3.1 Stage 1
This stage represents hypothetical negation strategies of a previous period before
the beginning of a new cycle with mana …-chu. The current adverb mana seems
to have been composed from *ma ‘no’ and *-na ‘irrealis’. The particle *ma ‘no’ is a
feature of negation marking in several Amerindian languages (Payne 1990). Cur-
rently, the Quechua derivational suffix -na is associated with meanings of desire,
need, possibility, obligation, negation and the future (Cerrón-Palomino 2008: 231;
Cusihuaman 1976: 220–222, Taylor 1996: 42–43, Weber 1996: 72, 449–450). Some
of these meanings, such as future and negation, are expressed by -na as a subordi-
nator. The current shortening of mana in Peripheral Quechua is further evidence
The development of standard negation in Quechua
5.3.2 Stage 2
Mana is the main marker of sentential negation. Currently, this pattern is the stan-
dard form in Peripheral Quechua (see Section 4.2.1). Most importantly, Central
Quechua still exhibits some evidence of single negation that could be interpreted
as relics of an earlier pattern. Adelaar (1986: 78) says that Pacaraos Quechua tends
to drop the suffix su.15 Suffix dropping for Adelaar occurs particularly when the
predicate is followed by pa (*-pas). Similarly, Weber (1996: 164, 441) comments
that for some unknown reasons “dropping of the suffix -chu” occurs in Huallaga.
See some examples below:
(64) Mana-m puñu-si-ma-n-pa.(PCR)
neg-as sleep-caus-31-3-ad
‘He doesn’t even let me sleep.’
(Adelaar 1986: 78)
(65) Mana-m ñuqa-nsi rima-nqa-nsi-naw. (PCR)
neg-as i-1incl speak-fut-1incl-comp
‘That is not how we speak’ or ‘we don’t speak like that.’
(Adelaar 1986: 29)
(66) Kondor kada aywa-y-nin yaku puyñu-n-ta
Condor every go-inf-3 water pitcher-3-ac
mana kachayku-q. (HGA)
neg leave-np16
‘Every going the condor didn’t leave its pitcher with water.’
(Weber 1996: 164–165)
. Normally it has been assumed that ni comes from Spanish. However, there are traces of a
native origin. Marcos Ferrel suggested this idea in personal communication to me.
. Adelaar (1986: 78) also indicates that mana omission can occur, but it is less frequent.
Pacaraos Quechua is a variety with double negation marking.
. Weber uses NP for ‘narrative past’, also known as ‘habitual’.
Edith Pineda-Bernuy
Quechua dialects with standard negation expressed with the clitic, such as Ancash
(HZ, CHS, TCS) and Junín (J-H) also present cases of single negation with mana.
Sentences containing conditional mood, doubt and conjecture are common con-
texts for single negation with mana. Observe these cases:
(67) Mana-pis aywa-shaq. (HZ & CHS)
neg-ad go-1fut
“Maybe I won’t go.”
(68) Qam mana piña ká-nki-man. (TCS)
You neg angry be-2-cond
‘You shouldn’t be angry.’
(69) Mana-tr-aa chay-pi lula-paaku-n (J-H)
neg-conject-em that-loc work-pl-3
‘Maybe they don’t work over there.’
These instances of single negation could reflect an archaic strategy in which the
suffix -chu would not be obligatory at all.
Similar relics of single negation are found in the 16th century Colonial litera-
ture of Quechua. Religious texts of this period are mostly from (central) South-
ern Quechua and based on or written in a Quechua variety called lengua general,
a lingua franca of that period. Taylor (2003: 13–15) assumes that, even though
there are several varieties of lengua general, they have all evolved from the same
dialect. According to Taylor, the lengua general could originate from Central
Quechua, similar to those dialects currently spoken in southern Yauyos (Lima).
Itier (2011: 64) maintains that the lengua general existed and was in widespread
use by the Indians.17 If so, it is likely that colonial texts reflect some of the nega-
tion features prevailing at that time. I consider that, because of the content of reli-
gious texts, negative sentences usually express prohibition, warning, advice and
so forth, and therefore, they predominantly present bipartite negation type. This
could make it appear that the bipartite negation was the norm. We can still see
some examples of single negation in other contexts:
(70) Yacha-ŝpa-taq-mi mana waqaycha-n[chu].18 (TC, 1585)
Know-sub-contr-as neg obey-3
‘But, although they know, they don’t obey.’
(Taylor 2003: 82)
. There are discussions on the Quechua variety that was called lengua general (Itier
2011: 63–85) and particularly that described by Domingo de Santo Tomás (see “Estudio
Introductorio” by Cerrón Palomino in Santo Tomás [1560] 1995).
. Amended suffix between brackets (Taylor 2003: 82).
The development of standard negation in Quechua
Again, single negation appears with -paŝ in (71) as seen in Pacaraos Quechua and
within a wh-word interrogative negative sentence as in (72) and (73), a type of
sentence in which negation is expressed with the form of standard negation (see
Section 4.4).
The analysis of old and current data indicates that single negation with mana
must still have been the main marker of standard negation in Central Quechua in
the 16th century, despite a shift towards double negation marking. I assume that
Peripheral Quechua has this feature for retention of an older form of negation,
and that it spread there directly or indirectly from some point in Central Quechua.
This feature could first have spread to the north of Peru and from there could have
been dispersed to Ecuador. Modern Central Quechua still preserves some relics of
single negation, but no variety in this area has standard negation with mana alone.
On the other hand, non-standard negation has bipartite negation. The clitic
-chu complements mana in the sentence for focus of negation, emphatic negation
and future tense (see Section 4.1.2). There is also another clitic -mi ‘assertive’ tak-
ing part in the negative sentence in a similar context corresponding to the clitic
-chu ‘negative assertion’ (see Section 4.1.2.4).
Peripheral Quechua must to some extent represent the use of the suffix -chu
in negative sentences in a phase when it was not yet obligatory. In 4.1.2, the suffix
-chu appears under a number of contexts. In the case of the colonial Quechua, pre-
sumably the suffix had spread to more contexts, similar to present-day Otavalo, an
EK variety, which is another reason why bipartite negation is more frequent in the
literature from the colonial period. The spread of the clitic -chu to other contexts
could lead to the interpretation of this as the part of the negation pattern.
Stage 2 Stage 3
Next stage: mana > mana…-chu
Edith Pineda-Bernuy
5.3.3 Stage 3
Bipartite negation with mana …-chu becomes obligatory. This innovation must
have spread from Central Quechua to the South and northern Peru. This is the
current form of standard negation of this large region (see Section 4.2.1), except
in the Ancash, Cajatambo, Dos de Mayo and Junín dialects, which have more
advanced changes. In these dialects, the suffix -chu indicates that negation has
sentential scope.
Non-standard negation is expressed with additional suffixes for different
meanings. Emphatic negation requires at least the evidential -mi attached to mana
(see Section 4.2.2.3). Observe the contrast between (74) and (75):
(74) Mana muna-ni-chu.(CZ)
neg want-1-neg
‘I don’t want.’
(75) Mana-m muna-ni-chu.(CZ)
neg-as want-1-neg
‘I absolutely don’t want this.’
Stage 3 Stage 4
Next stage: mana…-chu mana-m…-chu
5.3.4 Stage 4
Bipartite negation mana-m…-chu, including -m(i), is now the standard form, but
it coexists with the form mana…-chu. The addition of the clitic -m(i) to mana
converts the sentence into a negative assertion. This change must have covered the
areas of Ayacucho, Junín, Ancash, Dos de Mayo and Cajatambo. Currently, Aya-
cucho Quechua presents this characteristic for which mana-m…chu are obligatory
in the standard negation. It also occurs in the other mentioned varieties, but it has
emphatic strength.
It is also possible that some sentences having mana-m…-chu still can be inter-
preted as emphatic, depending on the context. Soto (1976: 122–123) asserts that
speakers incorporate the clitic -mi to mana in order to express emphasis.
In Ayacucho Quechua, mana-(m) dropping can occur, so that the senten-
tial negation is achieved with the sole clitic -chu, although it is contextually con-
strained (see Section 4.2.1).
Sentences containing non-standard negation present various strategies: the
suffix -puni ‘certainty’ in mana-puni-m adds emphatic strength. Similarly, fronting
the negative phrase is another strategy. Enclosing the negated phrase is for focus
(see 4.2.2.1).
Stage 4 is the bridge to single negation marking with -chu.
The development of standard negation in Quechua
Stage 4 Stage 5
Next stage: mana-m…-chu > -chu > -tsu > -su
5.3.5 Stage 5
Apparently, it is not possible to add more suffixes to mana-m to express emphatic
negation. The bipartite negation splits into two negation strategies. The sequence
mana-m…-chu is for emphatic negation and -chu alone is for neutral negation.
Such a reduction is not a result of a weakening of manam so that it can never
again express negation as seen in the last stage of the Jespersen’s cycle. Manam
alone can be the short negative answer to a question, but it cannot express negation
in a full sentential negation without the clitic -chu. Since the clitic -mi strength-
ens the negative meaning of mana, the resulting manam is the word which bears
the emphasis in an emphatic negation, while -chu is reinterpreted as the bearer
of negation (see Section 4.3.2.2). Therefore, the clitic alone can be the marker of
the sentential negation. This interpretation is confirmed by some speakers who
analyse mana-m…-chu as a double ‘no’, assuming that manam is unnecessary in
the simple negation.
5.4 Summary
The evolution of the standard negation of Quechua has been driven by the intro-
duction of the clitic -chu as a complement of mana. Judging by the current dia-
lectal distribution of this non-factual suffix, this appears as an interrogative that
spread gradually to the negative sentence, another irrealis domain, until it reached
all of the contexts of sentential negation.
Even though the origin of the clitic -chu does not correspond to an emphatic
marker, the contexts in which it appears are infused with a strengthening of the
negation meaning as seen in Section 4. In this function, the clitic -chu competes
at some stage with the evidential -mi ‘assertive’ (in combination with mana) to
become, in the final stage, the bearer of the negation in the standard form.
I will mention some traces of the sources of the renewal of negation in Quechua by
looking at current data. The morphemes mana and -chu participated in the men-
tioned Negative Cycle, but other morphemes and strategies were also candidates
for a new negation marker. All indicate that Quechua negation marking has been
linked to irrealis over time.
Edith Pineda-Bernuy
If it is the case that the clitic -ku was replaced by the clitic -chu, the implementa-
tion did not conclude in Ancash Quechua and neighbouring areas. The clitic -ku
is an irrealis.
The suffix -chu indicates contempt when added to nouns because it is irrealis
although nobody has related this to the suffix -chu ‘interrogative-negative’.
(79) lipi ‘rheum’ lipi-chu ‘rheumy’(CJM)
(Quesada 1976: 152)
7. Conclusions
. In any case, SE could represent the frozen spread of the clitic -chu, which is becoming
obsolete in certain contexts.
The development of standard negation in Quechua
References
Adelaar, Willem F.H. 1977. Tarma Quechua: Grammar, Texts, Dictionary. Lisse: Peter de Ridder
Press.
Adelaar, Willem F.H. 1986. Morfología del Quechua de Pacaraos [Documento 53]. Lima:
Universidad Nacional Mayor de San Marcos, Centro de Investigación de Lingüística
Aplicada.
Adelaar, Willem F.H. & Muysken, Pieter. 2004. The Languages of the Andes. Cambridge: CUP.
DOI: 10.1017/CBO9780511486852
Adelaar, Willem F.H. 2012. Modeling convergence: Towards a reconstruction of the history of
Quechuan-Aymaran interaction. Lingua 122: 461–469. DOI: 10.1016/j.lingua.2011.10.001
Beresford-Jones, David & Heggarty, Paul. 2012. Broadening our horizons: Towards an inter-
disciplinary prehistory of the Andes. In Archaeology and Language in the Andes. A Cross-
Disciplinary Exploration of Prehistory, David Beresford-Jones & Paul Heggarty (eds),
57–84. Oxford: OUP. DOI: 10.5871/bacad/9780197265031.003.0003
Bravo, Domingo. 1965. Estado Actual del Quichua Santiagueño. Tucumán: Universidad Nacio-
nal de Tucumán.
Catta, Javier. 1994. Gramática del Quichua Ecuatoriano. Quito: Abya-Yala.
Cerrón-Palomino, Rodolfo. 1976. Gramática Quechua Junín-Huanca. Lima: Instituto de Estudios
Peruanos.
Cerrón-Palomino, Rodolfo. 1987. Lingüística Quechua. Cuzco: CERA Bartolomé de Las Casas.
Cerrón Palomino, Rodolfo. [1994] 2008. Quechumara. Estructuras paralelas del Quechua y del
Aimara. La Paz: Plural Editores.
Cole, Peter. 1985. Imbabura Quechua. Melbourne: Croom Helm.
Coombs, David, Coombs, Heidi & Weber, Robert. 1976. Gramática del Quechua de San Martín.
Lima: Instituto de Estudios Peruanos.
Cusihuamán, Antonio. 1976. Gramática Quechua Cuzco-Collao. Lima: Instituto de Estudios
Peruanos.
Dedenbach-Salazar, Sabine. 1993. Una gramática colonial del Quichua del Ecuador: Transcrip-
ción e interpretación de un manuscrito del Archivo Histórico Nacional de Colombia. Bonn:
Bonner Amerikanistische Studien & Institute of Amerindian Studies.
Faller, Martina. 2011. A possible worlds semantics for Cuzco Quechua evidentials. Proceedings
of SALT 20: 660–683.
Fasola, Carlos. 2006. A unified semantics for the Quechua Question and negation marker -chu.
Paper presented in Questions Seminar at Rutgers University and SULA 4.
Edith Pineda-Bernuy
Harris, Alice & Campbell, Lyle. 1995. Historical Syntax in Cross-Linguistic Perspective.
Cambridge: CUP. DOI: 10.1017/CBO9780511620553
Heine, Bernd & Kuteva, Tania. 2002. World Lexicon of Grammaticalization. Cambridge: CUP.
DOI: 10.1017/CBO9780511613463
Hintz, Daniel. 2000. Características distintivas del Quechua de Corongo. Perspectivas histórica y
sincrónica. Lima: Instituto Linguístico de Verano.
Instituto Nacional de Cultura. 2009. Chopccam Kanni. Published with introductory study by
Pedro Roel Mendizábal. Lima: INC.
Itier, César. 2011. What was the lengua general of colonial Peru? In History and Language in
the Andes, Paul Heggarty & Adrian J. Pearce (eds), 63–85. New York NY: Palgrave. DOI:
10.1057/9780230370579.0010
Jespersen, Otto. 1917. Negation in English and Other Languages. København: A. F. Høst & Søn.
Landerman, Peter. 1991. Quechua Dialects and their Classification. Ph.D. dissertation, Univer-
stity of California, Los Angeles.
Lira, Jorge. [1941] 1982. Diccionario Kkechuwa-Español. Bogotá: Cuadernos Culturales Andi-
nos, No. 5.
Mercier, Juan Marcos. 1979. Nosotros los Napo-runas. Napu Runapa Rimay. Mitos e Historia.
Iquitos: Publicaciones CETA.
Miestamo, Matti. 2005. Standard Negation. The Negation of Declarative Verbal Main Clauses in a
Typological Perspective. Berlin: Mouton de Gruyter.
Moya, Ruth & Jara, Fausto. 2009. Taruka la Venada. Literatura Oral Kichwa. Quito: Ministerio
de Educación Ecuador.
Muysken, Pieter. 1977. Syntactic Developments in the Verb Phrase of Ecuadorian Quechua. Lisse:
Peter de Ridder Press.
Muysken, Pieter. 1995. Focus in Quechua. In Discourse Configurational Languages, Katalin É.
Kiss (ed.), 375–393. Oxford: OUP.
Nardi, Ricardo. 2002. Introducción al Quichua Santiagueño. Lelia Inés Albarracín, Mario C.
Tebes & Jorge R. Alderetes (eds). Buenos Aires: Editorial Dunken.
Nuckolls, Janis B. 2008. Deictic selves and others in Pastaza Quichua evidential usage. Anthro-
pological Linguistics 50(1): 67–89.
Parker, Gary. 1976. Gramática Quechua Ancash-Huailas. Lima: Instituto de Estudios Peruanos.
Payne, John. 1985. Negation. In Language Typology and Syntactic Description, Timothy Shopen
(ed.), 197–242. Cambridge: CUP.
Payne, David L. 1990. Some widespread grammatical forms in South American languages.
In Amazonian Linguistics: Studies in Lowland South American Languages, Doris L. Payne
(ed.), 75–87. Austin: UTP.
Pearce, Adrian J. & Heggarty, Paul. 2011. Introduction. History, linguistics, and the Andean
past: A much-needed conversation. In History and Language in the Andes, Paul Heggarty
& Adrian J. Pierce (eds), 1–16. New York NY: Palgrave. DOI: 10.1057/9780230370579.0006
Perroud, Clemente. 1961. Gramática quechwa: dialecto de Ayacucho. Lima: Seminario San
Alfonso de los Padres Redentoristas.
Pineda Bernuy, Edith. 2011. Derivación negativa en Quechua: El caso de la gramaticalización
de Mana. In Estudios sobre lenguas Andinas y Amazónicas. Homenaje a Rodolfo Cerrón-
Palomino, Willem Adelaar, Pilar Valenzuela & Roberto Zariquiey (eds), 233–255. Lima:
Fondo Editorial PUCP.
The development of standard negation in Quechua
Abbreviations
Hui-Ling Yang
National Quemoy University
This study provides a historical account for several issues not yet fully addressed
in the Taiwanese Southern Min (TSM) V bo XP construction, one of which is an
ambiguity over two interpretations, and another of which involves an absence
or a presence of negation in the post-verbal order. Many of these analyses might
be attributable to the historical development of Chinese negation. The term V2
negation is used in this current paper to describe the topic under investigation.
Assuming that the diachrony of bo is parallel to that of Mandarin mei, I analyze
Southern Min negative bo as originated in the V and reanalyzed as an aspect
marker in V bo XP. Other than the diachrony of Chinese negation, previous
treatments of one particular V2 negation are also reviewed. Contra Huang (2009),
who analyzes the semantics of bo in V bo DP as lexically determined in the
abstractness of the nominal phrase following bo, I argue that this bo occupies two
positions: while the aspectual head bo yields an episodic interpretation in the
nominal argument, bo in a higher head gives rise to genericity. Verb inner aspect
is adopted as the major theoretical framework.
1. Introduction
. Non-canonical is used here as a relative term, given that negation in an SVO language is
typologically pre-verbal.
. Unless noted otherwise, all examples in this paper are Taiwanese Southern Min. I use the
Taiwanese Romanization System for Southern Min sentences and Hanyu Pinyin for modern
Mandarin data.
Hui-Ling Yang
Other than (3) and (4), there is yet another type of negation in TSM, namely V bo
DP as shown in (7) and (8), which is less discussed in the literature. With a dif-
ference in word order, (7) and (8) are distinct semantically. In (7), the activity of
catching fish is not initiated, whereas the expected result is not (possibly) achieved
in (8a). Not only does the semantics differ, but there bears two possible readings in
(8), as suggested by Huang (2009): episodic in (8a) as opposed to generic in (8b).4
While bo allows two word orders, its quasi-counterpart mei in Mandarin only per-
mits one, as in (9), and (10) is disallowed. Rather, another construction with bu as
. I refer the reader to Lien (2013) for a wider range of uses of bo.
. Sentence (7) is modified from (8). I gloss bo unspecified for now.
Taiwanese Southern Min V2 negation
in (11) is utilized. The comparison of (9) and (11) is not parallel to that of (7) and
(8), however.5 Unlike modern standard Chinese (MSC hereafter), aka Mandarin,
which is more restricted in the use of mei, Southern Min allows for these two word
orders for bo.
(7) A-bing bo liah hi-a.
A-bing BO catch fish
‘A-bing did not catch the fish.’
The contrast becomes even more marked when the second predicate changes to
resultatives, as in (12)–(15). In the TSM data, the first order as in (12), where the
negative bo precedes the verbal string, is the same as that of (14) in MSC. The other
order, as in (13), has bo immediately preceding the element over which bo scopes,
namely pa. In (12) and (13), bo is predicated by a stative verb pa ‘full’. No coun-
terpart in MSC can be found; see (15). Despite the two different word orders, the
semantics of (12) is identical to that of (13) in TSM. By contrast, this is not appli-
cable to the Mandarin data in (14) and (15). A diachronic account may explain the
above-noted discrepancy. Historically, mei might have been used postverbally, as
in (10) or (15), despite their ungrammaticality in MSC.
. The morpheme dao is often regarded as a phase marker in Chinese linguistics (cf. Li &
Thompson 1981), which is a less grammaticalized unit than aspect. DAO is used here for
convenience.
Hui-Ling Yang
Several related issues with respect to the post-V negation are first addressed in
Section 2, followed by a historical development of Chinese negation in Section 3.
This study attempts to associate the diachrony of Chinese negation with the issues
under investigation. The later part of this paper (Sections 4 and 5) focuses on the
ambiguity of the post-V nominal interpretation in (8). The questions raised are:
(a) what causes the asymmetry in the semantics of bo between preverbal and V2
negation, as in (7) and (8), respectively?; and (b) how can we accommodate the
two readings of V2 negation syntactically?
Notice that the term post-verbal negation is better understood V2 as nega-
tion instead, for the use of bo as in (8), since bo does not negate the verb which it
immediately follows. The Sinitic language is presumably SVO and thus negation
is expected to precede the verb phrase. On the surface, the negative bo is placed in
between the verb and the shared object argument, thus called post-verbal negation
by some scholars. I use the two terms, V2 negation bo and post-verbal bo, inter-
changeably in this study only for convenience.
I argue for an inner aspect (cf. Travis 2010) within the VP to accommodate
the two readings and the representational approach in the tree is a modified ver-
sion of cartography (cf. Rizzi 1997; Cinque 1999). In a bi-clausal VP, the inner
aspect is to host the episodic reading. The negated verb bo ‘not.have’ is assumed to
initiate in the big V and reanalyzed in the small v, which indicates inner aspect.6
The aspectual ‘bo’ provides the verb with a change of state to ‘didn’t catch the fish’;
see the bridged tree diagram in (16). As for the generic reading, bo is reanalyzed as
Mod, sitting in a different head.
. Several names are used such as lexical aspect, telicity and Aktionsart.
Taiwanese Southern Min V2 negation
(16) …vP
v AspP
bo
Asp VP
bo
V DP
bo
hi-a ‘fish’
Intriguingly, the semantics varies based on predicate types. Unlike the identical
semantics in (29) and (30) as resultative verb compounds (RVCs), (33) and (34)
have a scopal difference. Examples (33) and (34) are often referred to as serial
verb constructions, SVC, with a continuity of the action ‘climbing’ to a change
of state ‘crossing over here’. In (33), the negative bo scopes over the resulting
state kau-lai ‘come.over’ only, whereas the same negative in (34) scopes over the
entire VP.
(33) kau-a pe bo kue-lai.
dog climb not come.over
‘The dog attempted but did not succeed in climbing over here.’
More striking facts about post-V negation can be found in negatives other than
bo in TSM. A closer examination of other SM pre- and post-verbal negative sets
reveals more asymmetries. The scope of negation may differ depending on the
word order. First, the modality of be in (35) and (36) changes from abilitive to
Taiwanese Southern Min V2 negation
Second, the negative m ‘not.want’ does not permit the post-verbal position; see
(37) and (38). The same applies to the negative bian ‘need not’, as in (39) and (40).
Both m and bian have a closer relationship with the speaker and grammatical sub-
ject argument, which may prevent them from the second predicate position. MSC
patterns this unavailability of (37) and (39).
Finally, another TSM aspectual negative bue is rather free in both pre- and post-
verbal positions, as in (41) and (42), and their semantics is seemingly identical.
This flexibility does not occur to MSC, either.
The above are synchronic data in both TSM and MSC. The exceptional con-
struction (i.e. post-V negation) is arguably a historical remnant from serial verb
constructions (SVCs). Shi (2002) observes that the V resultative construction
undergoes a grammaticalization path like (43), where ADV/NEG is fronted and
V–R becomes a verb compound with the course of time.
Shi attributes the fronting of negation in modern MSC to a byproduct of the estab-
lishment of the V–R construction in the history of the Chinese language. This
study applies his idea to the developmental process of bo, in particular the syntax
of bo in the post-V position. This diachronic process is remarkable as it explains
many of the synchronic divergences in SM negation. I do not intend to cover all of
the above-mentioned negatives. This current paper focuses on bo.
and brackets to indicate. He suggests that bo is attached to the main verb tshue to
form an RVC, in this case V1-V2 = tshue-bo.8
(48) i [tshue-bo] thau-loo.
3sg look.for-not.have job
‘He didn’t find a job.’
(49) i thau-loo [tshue-bo] thau-loo.
3sg job look.for-not.have job
‘He, jobs, didn’t find any.’
The above issue will be made clearer in the following section, where I will argue
that the post-V NEG construction of modern TSM is in between SVC and RVC,
not yet reaching the status of RVC.
Modern corpus data show that Southern Min bo exhibits fairly similar syntactic
behaviors to Chinese mei 沒. Since mei is well recorded in written texts, I there-
fore look at the historical development of mei in order to trace a possible gram-
maticalization path of bo. Many believe that 無 wu is used as ‘not have’ in modern
Southern dialects; therefore, the character 無 is often used to represent bo. The
diachronic development of wu is also examined.
. A reviewer points out that Example (46) can be viewed as a looser resultative construc-
tion. My decline however is the V compounding, V-bo, suggested by Huang. Whereas bo may
indicate a resultative state, an RVC account for V-bo does not seem to be stabilized in modern
TSM, as V and bo are still separable.
. Chinese writing is not a phonetic system. Note that the transcripts are based on modern
standard Mandarin. These negatives might have various pronunciations in different periods
of the history.
Hui-Ling Yang
The two negatives 無 wu and 沒 mei, as in (53), are related to Southern Min bo.
As shown, both mei and wu used to serve as full-fledged verbs for nominal predi-
cates. Shi and Li (2004) argue that the grammaticalization of mei has led to the
death of wu and other negatives when mei extended its usage to include adjectival
and verbal predicates. This marks a big shift in that the labor distribution of these
negatives has become no longer clear-cut, and mei has become multi-functional.
As suggested by the two authors, the modern mei system was established no later
than the 16th century and the disappearance of wu took place approximately in
the Yuan-Ming dynasties. Pan (2002), however, believes that 無 wu changed to
沒 mei due to phonological weakening. Regardless, an investigation of the gram-
maticalization path of mei and wu may provide some insight into the diachrony
of SM bo.
During the first stage, the original verbal mo/mei ‘to sink, to die’ evolved into
non-possession of ‘not have’. Sentence (54) shows that mo was used to mean ‘to
Taiwanese Southern Min V2 negation
sink’ before the eighth century. Sentences (55) and (56) illustrate cases where mei
means ‘not have’. Note that in (55) mei predicates a nominal phrase.10
(54) 夢為魚而沒於淵。 Huainanzi, Zhenxun; 125 BCE
meng wei yu er mo/mei yu yuan.
dream become fish and sink into abyss
‘(He) became a fish in his dream and sank into an abyss.’ (Shi 2002: 199)
(55) 深山窮穀沒人來。 Poem by Liu Shangyin; 800 CE
shen shan qiong gu mei ren lai.
deep mountain poor valley not.have people come
‘The deep mountains and poor valleys do not have people who know
them.’ (Shi 2002: 200)
(56) “車子有麼?” “車子沒。” Laoqida; 1325 CE
chezi you me? chezi mei.
carriage have q carriage not.have
‘Do (you) have a carriage?’ ‘(We) don’t have a carriage.’ (Shi 2002: 200)
During the second stage, the negated verb changes to a pure negator. In (57), mei
negates the verb you ‘to have’. Mei alone is a functional head.
(57) 如何沒有鮮魚? Shuihu zhuan; 1550 CE
ruhe mei you xianyu?
why neg have fresh.fish
‘Why don’t you have fresh fish?’ (Shi 2002: 200)
During the last stage, mei began to be used with verbs other than you ‘to have’.
Mei began to appear with non-nominal predicates. This change probably occurred
in the 16th century (Shi & Li 2004: 249). For instance, the negative mei in (58)
negates a non-have verb shang ‘to serve’.
(58) 這一日沒上過鐘酒。Jin ping mei; BCE 1550
zhe-yi-ri mei shang-guo zhong jiu.
for.a.while neg.asp serve-exp cup wine
‘Wine has not been served for a while.’ (Shi 2002: 200)
. Transcriptions of the examples in this section are all in Hanyu Pinyin.
Hui-Ling Yang
By this time, mei had become a negative perfect marker (Shi & Li 2004: 197). Mei
as a negative perfective marker came to be used later than mei.you as a negated
verb. The aspectual mei has then functioned as a bounded negative marker since
the 15th century, although its original verbal usage ‘not have’ still stays productive
to this day.
4) mei + you > Aspectual mei.you
There is also a stage where mei became fused with you to form mei.you (one unit),
which as a whole is reanalyzed as a negative perfect marker. It took a couple of
centuries for the reanalysis of aspectual mei.you to take place from the negative
verb mei you (Shi & Li 2004: 248). Shi and Li (2004) postulate a possible syntactic
environment for this analysis to take place. In (59), mei is the first verb in an SVC.
There are two VPs.
(59) [VP V1 = mei + NP] + [VP V2 + NP]
When the first NP is empty, the structure becomes (60). The next step is for mei to
become a functional head, taking the VP as its complement, as in (61). Also, we
see that two events are condensed into one.
(60) [VP mei + …] + [VP V2 + NP] >
The historical development of the negative mei can be summarized as (62). It pro-
vides us with a possible pathway of the grammaticalization of Southern Min bo.
(62) mei = verb ‘to die’
> mei ‘not’ + you ‘to have’
> negative perfective: mei + V
> negative perfective: mei.you + V
From the above three examples, we learn that wu appears as the verb in the VO
composition. Wu as a verb ‘not.have’ often appears in a verb sequence connected
by the conjunction er ‘and’, such as (68). Wu in (68) also shows the two-verb
sequence although there is no explicit conjunction marker between the verb ju
and a second verb wu. In other cases with wu, however, er ‘and’ is intensively used
in Shishuo Xinyu.
(68) 居無幾何而周舍死。《世說新語》(Shishuo Xinyu; 220–258 CE)
ju wu ji he, er Zhoushe si
live not.have some year and name die
‘…lived for some years and Zhoushe died’
Per Laoqida in (69), wu is the second component in the verb sequence with an
adverbial hou ‘later’ in between the first verb kong ‘worry’ and wu. The nominal
phrase is indefinite. The indefinite bare noun is easily lexicalized with the previ-
ous verb wu into one unit, such as wu-ping (not.have-proof) ‘having no evidence’.
It is rather difficult to find instances in this text where wu is placed as the second
verbal element. Only three out of the very few wu entries (seven) are found. Most
tokens have wu as the main verb such as the case in (70). Interestingly, mei 沒 is
also attested as a verb ‘not have’ in Laoquda.12 Both mei and wu are overlapped in
functions. As a matter of fact, the number of mei sentences is larger than that of
wu. One begins to see the rise of mei and the fall of wu.
(69) 恐後無憑。《老乞大》(Laoqida; 918–1392 CE)
kong hou wu ping
worry later not.have proof
‘worry that there will be no proof later’
(70) 這房裏無人。《老乞大》(Laoqida; 918–1392 CE)
zhe fang li wu ren
this room inside not.have person
‘There is no one in the room.’
The above text is relatively more colloquial. In the poetic style, wu is often used to
contrast with you ‘to have’, as in (71).
(71) 荷盡已無擎雨蓋,
he jin yi wu qing yu gai
lotus die already not.have hold rain cover
‘When the lotus dies, its sheltering leaves are already gone.’
菊殘猶有傲霜枝。 蘇軾《贈劉景文》(Su Shi; 1036–1101 CE)
ju can you you ao shuang zhi
chrysanthemum die still have elegant frost branch
‘When the chrysanthemum dies, its elegant frost branches still remain.’
. The historical text Laoquda served as a textbook at the time (918–1392 CE) for Korean
learners of Chinese.
Taiwanese Southern Min V2 negation
We now proceed to Lijing ji, the earliest available Min text. Intriguingly, the V
NEG DP construction is still scarcely observed, with (73) as one such example.
The nominal phrase mih ‘thing’ here is indefinite.
(73) 我畏無物通度汝食。《曆鏡記》(Lijing ji; approximately 1566 CE)
gua kiann bo mih tang hoo li tsiah.13
1sg dare not.have thing can give 2sg eat
‘I’m afraid not to have food for you to eat.’
So far, we have not seen a case where the nominal phrase is definite. This in fact
needs some syntactic environments. For instance, we can say that kiann ‘dare’ and
bo ‘not.have’ are two independent verbs in (73), but they can be in one bi-clausal
event in certain situations, such as ‘I don’t have the things that you requested the
other day.’ This is only postulation. When the syntactic relationship becomes
closer between the verbs, the object nominal argument may become definite. This
involves a change of SVC into RVC, a topic to which we now turn.
. I transcribe this line in modern Southern Min as Lijing ji is a Southern Min text.
. Examples are from Shi and Li (2004: 238–239); transliteration is mine.
Hui-Ling Yang
(78) 住了沒兩日就下起雪來。 紅樓夢 The Dream of the Red Chamber (the 18th cy.)
zhu le mei liang ri
live asp neg two day
jiu xia qi xue lai
then fall up snow come
‘(Someone) stayed (somewhere) for less than two days when it
began to snow.’
Briefly, MSC loses the post-V negation structures, yet contemporary TSM pre-
serves them, e.g. bue for ‘not.yet’ and bo for ‘not.have’. Note that the post-V
negation in (76) and (77) are still available in MSC, but wei in (75) and mei in (78)
are excluded. MSC mei also serves the function of ‘not.yet’ in the pre-V position.
This study does not attempt to delve into the intra-linguistical divergences; rather,
the pre-MSC data provide a piece of evidence that the historical development of
Chinese negation may provide an answer to the intricate syntax of bo in TSM.
The construction in (43) originates as two serial verbs with a negative scoping
over the second verb: V1 + NEG+ V2, as in (79).15 The negative is then fronted
before the main verb and the second verb becomes a secondary predicate of the
first verb, i.e. NEG+ V1 -V2, as in (80). We now see NEG + V–R in MSC, as in
(81), which indicates a closer relationship between V and R. According to Shi
(2002), at the last stage in the process, due to disyllabification and a syntactic
reanalysis, the verb (chi ‘eat’) and the resultative complement (bao ‘full’) are com-
bined to form a V–R unit, thus leaving the negative separated from the compo-
nent that it originally negates. In (81), the negative mei scopes over the second
predicate of the VP: chi bao, as shown in the transliteration: ‘(someone) ate but
the consequence is that s/he did not feel full’. Shi and Li (2004: 234) refer such to
boundedness in RVC.
(79) V1 + neg + V2
食 未 飽
shi wei bao
eat not.yet full
(80) neg + V1 + V2
未 食 飽.
wei shi bao
not.yet eat full
(81) neg + V–R
mei chi-bao MSC
neg eat-full
‘(Someone) ate but did not feel full’
. One may argue that wei is different from mei. Recall that mei was once for nominal
predicates only. The change in negation is as complex as the expansion of mei to other types of
predicates and the fronting of negation. I leave this issue for further research.
Hui-Ling Yang
Now we turn to the target language variation, Taiwanese Southern Min. Supposing
that Southern Min bo mirrors the aforementioned syntactic pattern of negation, a
postulated process for bo to evolve from SVC to RVC should look like (83). The V1
+ bo-R is tsiah + bo-pa, as in (85). The fronted bo with V–R to form ‘bo + tsiah-pa’,
as in (86), is a later development. Both (85) and (86) are available in contemporary
TSM. MSC mei can only fill the position of NEG in (86).
(84) V1 + NEG + V2
tsiah bo pa
eat BO feed.well
(85) V1 + [NEG + R]
tsiah bo pa
eat BO full
The above addresses the issue of syntax from SVC to RVC in the history of Chinese.
The purpose is to show how syntax changes where two independent events (two
verbs in a sequence) have come to a closer relationship and where the two events
(headed by the verbs) become more dependent syntactically. The dependency
interacts with semantics, in particular when an NP argument participates. One
character to be brought up later in this paper is the episodic vs. generic interpreta-
tion in the nominal phrase.
Taiwanese Southern Min V2 negation
Neg AspP
mei
Asp VP
mei
V …
As argued by van Gelderen, a loss of semantic features accounts for the reanalysis
of a lexical head to a higher head, as the conceptualized figure in (88) shows. As a
lexical negated verb, mei ‘not have’ carries a bundle of semantic features as nega-
tivity, possession, and agency. When mei loses its semantic features, it becomes a
negative marker with interpretable features of negativity and perfectivity, labeled
as [i-F: neg; perf]. This also means that mei becomes reanalyzed to a different
head. When mei gets further reanalyzed to a pure negative, i.e. the Neg head, the
aspectual features in mei have become extinct and its interpretable feature [i-F]
is solely negativity. There is in fact another head, which is interrogative in the
CP/discourse layer (labeled as [iF: int]).
(88) Feature loss during the process of grammaticalization
IntP
mei …
[iF: int] NegP
mei AspP
[iF: neg]
mei VP
[iF: neg; perf]
mei
[neg; poss; agent]
Hui-Ling Yang
The remainder of the paper focuses on one particular V2 neg construction in TSM,
namely V bo DP, to elucidate how the previous historical account is at play in the
synchronic syntax. This section examines two previous works on the V bo DP
construction, examining their plausibility while pointing out possible problems
from a historical point of view.
Huang relates these two readings to the property of the nominal phrase following
bo. She claims that the two types of readings of u/bo are determined by the refer-
entiality of the nominal phrase following u/bo. As stated by Huang, the nominal
Taiwanese Southern Min V2 negation
The central issue now is that there can be a definite reading in the bare noun. Bare
nouns can indicate plurality, definite singularity, and indefinite singularity. Thus,
(93) is also possible, with u/bo being read as abilitive deontic modality. H owever,
a non-abilitive reading can also be derived from the same sentence, as illus-
trated in (94). The same distinction (definite versus abilitive) applies to (91). This
phenomenon indicates a deficiency of Huang’s rule of “abstract/concrete” real-
world experience to determine the quality of the object nominal phrase. Huang’s
analysis is in fact about bare nouns, given that a non-bare noun such as (95) is
beyond her discussion (Huang 2009: 17).16
(93) A-bing liah u/bo hi.
A-bing catch U/BO fish
‘A-bing can/cannot catch the fish.’
(94) A-bing liah u/bo hi.
A-bing catch U/BO fish
‘A-bing failed to/didn’t catch the/a fish.’
(95) A-bing thak bo siann tsheh.
A-bing study not.have what book
‘Abing is unable to study well.’
The interaction between inner aspect and the NP argument is fairly complex and
is often treated as a debate of semantics. It also involves grammaticalization of cer-
tain components within the nominal phrase. Here I intend to indicate that syntax
has to be taken into consideration as well.
. The interrogative pronoun siann ‘what’ here serves as an indefinite marker and yields a
universal interpretation.
Hui-Ling Yang
NP
Abing V VP
DO
liah ‘catch’
V NP
GET
bo hi-a ‘fish’
To relate the NP to the event, Huang claims that whether a sentence is read as
generic or episodic is determined by the nominal phrase. She further explains
that the object nominal phrase (referential or non-referential) entails resultatives
from the activity verb of a GETTING of something, either a concrete or abstract
object. For perception verbs, it is “a GETTING of the understanding, sight, hear-
ing or selling of the NP” (Huang 2009: 63–65). But, the problem is if the GET
head is made use of, as explained by her, telicity is achieved and there should be
no generic reading as a result. Next, Huang seems to suggest that the GET V head
serves to turn the activity verb into change-of-state type in the lower VP. However,
it’s not clear how the change of state comes about. I will address how this telicity
aspect is connected to the DP in terms of (in)definiteness later. Also, in Huang’s
bi-clausal event in (96), I assume that the second verbal element bo has telicity,
and it becomes a state of accomplishment. If this is true, how do we get the generic
reading as in (93)? From her tree diagrams, I assume that a cartographic approach
is taken. However, her referentiality suggestion resorts to the speaker’s perception
toward the world. It is not clear whether she wants to say the parameters are in the
lexicon (Chomsky 1995: 60) or they rely more on pragmatics.17
. I ignore other features denoted by D. I do not intend to argue for or against the notion
whether D denotes definiteness at all. I simply assume that definiteness is in the DP.
Taiwanese Southern Min V2 negation
DP
Li-ei
I VP
tshue SC
DP VP
ti
NegP VP
bo
V NP
ACHIEVE thau-loo
The above analyses echo partially the development of word order change in nega-
tion, addressed in Section 3. Much of the interim process is evident in TSM post-V
bo construction. What is undergone by MSC may be delayed in another Sinitic
Hui-Ling Yang
language variation. For instance, as argued by Gillon and Yang (2010), bo is not yet
developed into a resultative verb compounding. The split VP approach provides a
good solution to the syntactic change of negation, yet a new model is expected to
cover as many characteristics of post-V bo and address the restrictions of different
complements in V bo XP. An unsolved puzzle so far is the generic interpretation in
V bo DP. Before closing this section, I would like to review an analysis of the MSC
V-de/bu-R construction, which also expresses abilitive modality.
Modal′
Modal VP
Dei Modal V′
V DeP
[kan-[De de/bu]i-wan]j De
De RP
ti
R′
R NP
tj
zhe-ben shu
Taiwanese Southern Min V2 negation
Her argument about the ModP above the VP comes from the two sentences below,
which she argues share identical semantics.18
(100) Lisi kan-BU-dao zhe-ke shu. MSC
Lisi chop-DE/BU-fall this-clf tree
‘Lisi cannot chop the tree down.’ (Wu 2004: 272)
(101) Lisi bu-neng kan-dao zhe-ke shu. MSC
Lisi not-can chop-fall this-clf tree
‘Lisi cannot chop the tree down.’ (Wu 2004: 272)
Based on the diachrony of Chinese negation, I argue for two heads responsible
for two different interpretations for the V bo DP construction. The frameworks
adopted here include verb inner aspect and cartography.
. Note that bu in (100) carries modality, but is a pure negator in (101), where neng ‘can’
denotes abilitive modality.
Hui-Ling Yang
I analyze this u/bo as inner aspect. Based on the diachronic analysis in section
three, the affirmative u is postulated to originate in the lowest V and get reanalyzed
in the aspect head, combining with Neg to form bo.19
(106) The aspectual bo; (104b)
vP
v NegP
lia
‘catch’
Neg AspP
V DP
u
hi-a ‘fish’
One piece of evidence that bolsters the claim that bo is aspectual is its use with tioh
as in (105). The morpheme tioh originates as a verb, meaning ‘to obtain; to attach’
. I ignore the exact site of NegP for the sake of simplicity. The rest of the tree diagrams
does not include a NegP node for the same reason.
Taiwanese Southern Min V2 negation
and now is used to show verb inner aspect, similar to MSC dao. I label the ASP
head with an abstract OBTAIN head in the spirit of Lin (2004), who utilizes two
verbal heads to distinguish an activity from a stative event. In the grammaticaliza-
tion framework adopted here, this yet is another reanalyzed negative bo, being
non-aspectual. When this non-aspectual bo is selected in the lexicon coupled with
aspectual tioh, the bare noun in the object DP position is interpreted as episodic.
The minimalist feature economy (Chomsky 1995) is adopted.
v NegP
lia
‘catch’
Neg AspP
bo-tioh
hi-a ‘fish’
Asp VP
OBTAIN
tioh
V DP
hi-a ‘fish’
The above perspective offers multiple advantages. First, it tracks the diachrony of
Chinese negation: the reanalysis process (van Gelderen 2011). Like mei, bo pos-
sibly undergoes a process from a full-fledged verb to aspect marker (and further to
other functional categories). Second, this syntactically driven approach avoids the
challenges faced by the principle of “abstractness” proposed by Huang (2009), who
analyzes the distinction between her two readings as lexically determined by the
abstractness of the nominal phrase. The aspectual bo in (106) associates the bare
nominal argument with an episodic reading. Though the syntax of bo provides two
available orders (one in the verb head and the other in the aspect head) at LF, the
linear order for both happens to be the same in the case of (106) and (107). With
this simpler model, there is no need to refer to a rather abstract principle: that the
absence or presence of a particular reading in the NP comes from the speaker’s real-
world experience. Finally, my analysis accommodates other types of post-verbal bo,
illustrated in (108)–(111). When the DP position is substituted with XP (X = resul-
tative verbs or adjectival stative verbs), there is no need to c reate an a dditional tree
Hui-Ling Yang
diagram. The telic marker liao/uan in (108) fills the inner AspP, and bo precedes a
resulating complement in (109)–(111), which also indicates telicity.20
(108) i lim bo liao/uan hit uann thng.
3sg drink neg finish that bowl soup
‘He didn’t finish drinking that bowl of soup.’
(109) kau-a pe bo kue-lai.
dog climb neg come.over
‘The dog attempted to climb over but did not make it here.’
(110) i tsiah bo pa.
3sg eat neg full
‘He ate but didn’t feel enough.’
(111) hue khui bo ang.
flower bloom neg red
‘The flower did not bloom as red as it would.’
v AspP
turn
the lights
Asp VP
off
V DP
turn the lights
. The fronting of the DP, i hit uann tng lim bo liao/uan, may be preferred by some speakers.
Taiwanese Southern Min V2 negation
v AspP
turn
Asp VP
off
V DP
turn some lights
. Huang (2009) uses generic instead for one of the multi-interpretations. My emphasis
here is on the proposal of a different head for bo.
Hui-Ling Yang
(117) generic bo
ModP
ModP AspP
bo
Asp VP
V DP
hi-a ‘fish’
Mod AspP
be
Asp VP
tioh
V DP
hi-a ‘fish’
The cross-out of tioh in (119) indicates where the morphemes are possibly base-
generated. Historically, be ‘can’ originates in the V, and tioh can be aspect or modal-
ity (cf. Yang 2001; Lien 1997). Sentences (116) and (118) potentially share the
same semantics. An advantage for the ModP head bo is that it bridges semantics
(modality) with syntax. As noted, the previous treatment of the post-V negation
can only host a single type of syntax, namely V bo DP. This alternative proposal
extends to other Sothern Min negatives, such as be ‘not.able’ in (118). Crucial are
the apparently multiple categories found in the same morpheme of bo. Bo has
Taiwanese Southern Min V2 negation
6. Conclusion
The semantics of the DP under the scope of negation is a richly debated issue in
the field of morphology. One analysis of the DP is related to two interpretations
of the DP argument, one being episodic and the other generic (Huang 2009). For
the episodic vs. generic distinction in the TSM V bo DP construction, an inner
aspect and a higher head are suggested in this paper. Southern Min negation sys-
tem is far more complex than what has been presented here. As a matter of fact, in
modern TSM not every single negative morpheme can participate as V2 negation.
The abilitive modal be ‘not.able’ and another aspectual bue ‘not.yet’ are qualified
for this position, but the volitional m ‘not want’ and deontic bian ‘need.not’ are
excluded from this V2 position. Many factors other than syntax are intertwined
within this negation system, which is beyond the discussion of this current paper.
I leave them for future research.
Acknowledgements
I would like to thank Elly van Gelderen for her suggestions on the work discussed
here. I am grateful to the audience at ALC-4 held at the University of Arizona,
and to two anonymous reviewers for their comments on an earlier version of this
paper. All errors and omissions are my responsibility. I also wish to acknowledge
the generous support of the Chiang Ching-Kuo Foundation for International
Scholarly Exchange for awarding me a doctoral dissertation fellowship during the
academic year of 2011–2012.
Abbreviations
References
Shi, Yuzhi & Li, Charles N. 2004. A History of Grammaticalization in Chinese. Peking: Peking
University Press.
Tang, T.-C. 1994. On the semantics and syntax of negatives in Southern Min. Studies on Chinese
Morphology and Syntax 5: 119–169. Taipei: Student Book Co.
Travis, Lisa. 2010. Inner Aspect: The Articulation of VP [Studies in Natural Language and Lin-
guistic Theory 80]. Dordrecht: Springer. DOI: 10.1007/978-90-481-8550-4
Wu, Ching-Huei Teresa. 2004. On de/bu and the syntactic nature of resultative verb compound-
ing. Language and Linguistics 5(1): 271–329.
Wang, Chyan-an. 2010. The Microparametric Syntax of Resultatives in Chinese Languages. Ph.D.
dissertation, New York University.
Wang, Yu-yun Iris. 2008. Postverbal negation in Taiwanese. Paper presented at the 11th Inter-
national Symposium on the Chinese Languages and Linguistics. Taiwan: National Chiao
Tung University.
Yang, Hsiu-fang. 2001. The forms and meaning of the word JIE: A historical perspective.
Language and Linguistics 2(2): 261–297.
Berber negation in diachrony*
Vermondo Brugnatelli
University of Milan-Bicocca
The Berber languages are a linguistic family1 that is scattered right across Northern
Africa, from the western oases of Egypt to the Atlantic Ocean and from the
Mediterranean coast to the southern borders of the Sahara. They make up a branch
of the Hamito-Semitic (Afroasiatic) macro-family and are still spoken by some
forty million people, although dialects of the Arabic language are nowadays also
spoken in many regions that were previously only Berber-speaking.
* In this paper, Arabic is written according to the standard transcription, while Berber is
written according to the standard orthography of Kabyle, whenever appropriate. The differ-
ences between both systems are as follows: Berber 〈 c 〉 stays for Arabic 〈 š 〉, 〈 ɣ 〉 for 〈 ġ 〉, 〈 x 〉
for 〈 ḫ 〉, 〈 ɛ 〉 for 〈 ʿ 〉. Moreover, the interdental fricatives [θ] and [ð] are represented by 〈 ṯ 〉 and
〈 ḏ 〉, schwa [ə] is represented by 〈 e 〉 and Tuareg [e] is written 〈é〉. In Medieval Berber, what
I transcribe 〈ǧ〉 may represent both [ǧ] and [g]. Abbreviations in the examples are: aor =
aorist, aorist particle; aux = auxiliary verb; neg = negator, negative particle; npf = negative
perfective; op = orientation particle; part = participle, mark of participle; pf = perfective;
pl = plural; pred = predicative particle. The sign = marks an amalgam, a dash separates clitics,
while underscore marks the existence of sandhi phenomena.
. From a linguistic point of view, Berber is more a linguistic family than a language.
However, many people, especially in countries where it has little or no recognition, still regard
it as one language, for ideological and political (extra-linguistic) reasons involving prestige.
Vermondo Brugnatelli
With regard to negation, the Berber languages display several clear-cut com-
mon features but also many innovations that are peculiar to isolated dialects or
shared by a small number of languages. A comprehensive description of such a
vast and manifold subject has not yet been written, although many partial sketches
and studies are already available.2 As far as comparative linguistics is concerned,
an overall outline of Berber negation within a Hamito-Semitic framework is given
in Brugnatelli (2006).
From a diachronic perspective, there are many interesting phenomena to be
observed in the domain of negation in relation to both historic and pre-historic
times. Given that it would be both over-lengthy and superfluous to deal with all of
these in the current paper, I have chosen to focus on the key aspects that appear to
offer the most potential for cross-linguistic comparison, namely the origin of the
negative particles, the negative forms of the verb and the question of “discontinu-
ous negation” and how it relates to the issue of Arab-Berber interference.
The Berber languages possess many different negative particles. Some of these are
widely attested and quite likely to have derived from a common element, while
others are restricted to smaller areas and may nearly always be considered innova-
tions. The creation of new negative elements took place above all in the domain of
verbless sentences, possibly as a consequence of contact with Arabic, a language
that tends to use different negative strategies for nominal and verbal sentences
(Brugnatelli 2006).3
The negation used with verbs often consists of circumfixes:4 one particle (neg1)
precedes the verb and another (neg2) follows it. However, neg2 is not used in the
. For a general picture of negation in the Berber languages, letting aside the specific issues
that will be dealt with below, see the contributions in Chaker & Caubet (1996) and the papers
of Galand (1994), Brugnatelli (2006), Lucas (2007), Mettouchi (2009), Taine-Cheikh (2011).
. Some of these new negative tools are dealt with in Kahlouche (2000), Brugnatelli (2010),
and Lafkioui (1996, 2007: 234–236; 2011: 62–69). For the sake of convenience, the present
paper will deal mostly with verbal negation.
. Circumfixes are frequent in the Berber grammar: in addition to their use in negation,
they are normally used in nouns (for which gender and number are expressed by prefixes +
suffixes) and in the conjugation of verbs (see Footnote 12 below). The sporadic “doubling”
of clitics before and after the verb could also be viewed as the development of circumfixes;
it occurs in Libya (Zuara), Algeria (Mzab, Aurès, Ngouça): Brugnatelli (1993: 234–237) and
Morocco (Tarifit: Lafkioui 2007: 128).
Berber negation in diachrony
. Some of these languages have innovated the negator. See below, § 1.1.
. As a general rule, most of the Berber languages have a threefold verbal system based on
the opposition between a perfective and an imperfective stem, while a third stem, called aorist,
is unmarked as far as aspect is concerned and has distinctive, mostly modal, functions.
. An interesting restriction in the use of this particle is the fact that in most languages it
cannot be used with the aorist, but only with perfective or imperfective forms. To examine
the origin of this state of affairs would be a lengthy process and outside the scope of this
paper, given that it would also require a wider investigation of the genesis of the modern
tripartite TAM system (aorist, perfective, imperfective), which is not that originally in use but
appears to be the result of long-term processes that reshaped the entire verbal system (Galand
2010: 202–207).
. Most of these particles have not yet been investigated. With regard to mi in Yefren, see
Brugnatelli (2014: 130). The origin of the preverbal negator ak used in Ghadames as an alter-
native to wel is puzzling and difficult to account for. Mettouchi (1996:191) considers it to be
a cognate of the most widespread words used for neg2 (discussed below): “Il peut arriver que
l’élément généralement réalisé après le verbe prenne la place de ur. C’est le cas pour le parler
de Ghadames” (“it may happen that the element usually uttered after the verb replaces ur. This
is the case with the variety of Ghadames”). A misinterpretation of this statement leads Lucas
Vermondo Brugnatelli
This similarity has led some scholars to posit a link between the two particles
(Loubignac, 1924: 177; Basset 1940: 221; Chaker 1996: 12), but two facts challenge
this view: firstly, the privative particle has a feminine form tar- implying that the
initial wa- sounds should rather be regarded as the reflex of a masculine pronomi-
nal marker (wa, which has a feminine counterpart ta); secondly, the presence of
the prefix war- in languages that use wel and not wer as a negator, points to an
independent origin. See, for instance, an example from an unpublished old Ibād�ite
text9 containing both particles:
(2) d iwalen ǧǧ wel yelli war elh�eqq
pred word=pl in neg 3ms=be=npf lacking truth
‘they are utterances in which there is nothing untruthful’ (f. 82b, l. 8)
In any case, Lionel Galand has repeatedly stressed the fact that, given the multiple
forms of the negator, it seems preferable to consider only the beginning sound
we-/u- as the “basis of the negative particle”, because it is the common element
shared by all the languages (Galand 1994: 171; 2010: 279–280). This implies that -r,
-l and -d were added during a later phase, and this could account for the different
forms attested.
to erroneously list the language of Ghadames among those that only possess a post-verbal
negator (2013: 412). More convincingly, Galand (1994: 172) points to another pan-Berber
particle, akw/akkw ‘all’, which is often used to strengthen other particles. In the Moroccan
dialect of the Ayt Youssi of Enjil it precedes the preverbal negator ur when the meaning is ‘not
at all’. More details in Galand (2011: 2nd part 66). An example:
This discovery, first presented and more thoroughly explained and discussed in
Brugnatelli (2011), reopens, on a new basis, a long-debated issue in Berber his-
torical studies, namely the proposed verbal origin of the negative particle. This
hypothesis was put forward by Loubignac in the work cited earlier (1924: 177), in
which he suggested bringing together under one and the same root R the negative
particle wer, the privative prefix war and the verb ar ‘be void, desert’ of Central
Morocco; later on, A. Basset took up the debate in a detailed study concluding that:
. For instance, in Nafūsī wel yuc we dd-yusu-c ‘he has not yet come’, lit. neg1 3ms=aux
neg1 op-3ms=come=pf-neg2. Other verbs are used in other regions. Elsewhere, for instance
in Jerba, this verb is lexicalised and has become an invariable particle wecci. More details in
Brugnatelli (2011: 521–524).
. The Berber personal indices may be: prefixes (3ms y-, 3fs t-, 1pl n-), suffixes (1s -ɣ,
3mpl -n, 3fpl -nt) or circumfixes (2s t-…-d, 2mpl t-…-m, 2fpl t-…-mt). In Chenoua dialect,
morphological t usually becomes h.
Vermondo Brugnatelli
“We would be strongly tempted to recognize in war an old 3rd pers. masc. sg.
preterite of a quality verb whose characteristics are already fully determinable. In
this case, the root would be biliteral, WR; tar would be due to reshaping (…) and
wər, whose vocalism remains obscure, would probably be a frozen masc. sg.”12
What he considered important was the fact that most nouns following the priva-
tive war are in the free state instead of the annexed state required by most prepo-
sitions.13 The free state is typical of the direct object and could be easily explained
if the original construction was a verb phrase and not a prepositional phrase (see
also Prasse 1972: 244).
On the other hand, the argument put forward by Prasse (1972: 244; 1973: 12) –
quite similar to the point of view of Marcy (1936, 1940/41) – took into consider-
ation the behaviour of negation in the relative and wh-interrogative clauses, which
require a special form of the verb, the so-called participle. In these constructions,
the morphemes of the participle are not suffixed as in positive sentences, but
placed before the verb and after the negator. As a consequence of this shift, these
morphemes may be seen as the participial mark placed after the negator, and the
method adopted to transcribe a negative expression in relative or interrogative
clauses depends on whether this theory is accepted or rejected:
(5) a. uren yufil
neg=part 3ms=be tanned=npf
b. our en ioufil
neg part 3ms=be tanned=npf
‘(which is) not tanned’ (newer and older spelling of the same phrase in
Drouin 1996: 244)
The case of Chenoua negation examined above suggests an earlier scenario some-
where in between the reconstructions of those positing an ancient verb meaning
‘not being’ and those who have not accepted this theory, considering the negator
to have been an invariable particle from the earliest stages of Berber. Actually, it
seems that only the last part of the negator may be traced back to a verb, while
. “Nous serions fort tentés de retrouver en war une ancienne 3e pers. masc. sg. de prétérit
d’un verbe de qualité aux caractéristiques dès maintenant parfaitement déterminables. En ce
cas, la racine serait bilitère, WR ; tar serait dû à une réfection (…) et wər, dont le vocalisme
reste obscur, serait sans doute un masc. sg. figé” (1940: 222). Chaker (1996: 12–14) accepts this
theory and assumes that the different vocalism of wər and war points to a contrast between
aorist and “preterite” (i.e. perfective).
. The opposition of two “states” in nouns (“free” vs. “annexed”) is a well-known feature of
Berber grammar and does not require in-depth presentation here. The best updated descrip-
tion of the phenomenon, with a critical bibliography on the topic, is offered by Galand (2010),
Chapters 4.1–4.3 and 5.3.
Berber negation in diachrony
the initial part u/we was a basic uninflected negator, as previously suggested by
Galand.
According to Galand (2010: 284), the conjugated element of the final part of
the negator derives from a monoconsonantal verb meaning ‘will, want’. This verb
still exists as iri in Moroccan Berber (Tashelhit in the South as well as dialects of
central Morocco), and as әr or äru in Tuareg. This is fully consistent with the form
of the negative particle in all those dialects where it appears as ur/wer. However,
some dialects display a lateral [l] in the negator, which appears as ul/wel (Ouargla,
Ghadamès, Jerba (partially), Mzab, Nefousi, Central Morocco: Zemmour, Iziyan).14
In this respect, it is worth noting that the verb itself may have a variant with [l]. As a
matter of fact, in a Medieval Berber text in which the preverbal particle is wel, I was
able to single out a verb il meaning ‘will, want’: in some instances, the word yil is
translated by an Arabic gloss yurīdu ‘he wants’. For instance:
(6) Arab text: yurīdu iqtit�ā ʿa māli-ka
Berber translation: yil [Gloss: yurīdu] s weɣres n weḏrim-ik
‘he wants to take possession of your money’ (f. 286b, l. 2)15
In any case, this phonetic issue requires further investigation, given that the Berber
varieties in which the preverbal negator is ul/wel usually also display l-variants
of other morphological elements, such as the preposition ɣel/al ‘towards’ (vs. the
more frequent ɣer/ar) or the particle ala (vs. ara) used before relatives.16
. Concerning ud, at the beginning the consonant d might have been an element added
in order to avoid a hiatus. For instance, in Jerba neg1 is usually u/we (seldom wel), but a free
choice exists between w- and weḏ(ḏ)- before a vowel: w-uḏifeɣ-c / weḏḏ-uḏifeɣ-c ‘I did not
come in’.
. The spelling is sure, since the word is fully vocalised: the consonant y bears the sign of the
i vowel, and the final l has a sukūn, which means lack of a following vowel.
. For a more detailed discussion of this issue concerning the preposition, see Brugnatelli
(2006: 69). As for the other particles, see below (§ 1.2.2.).
Vermondo Brugnatelli
to consider neg2 a borrowing from Arabic, but this is not always the case (Brug-
natelli 1987a). In reality, the most widely used particles for neg2 may be traced
back to one and the same lexical item, “the proto-Berber forms *kyăra ~ (h)ăra(t)
‘thing’ ” (Kossmann 2013: 332). The original forms have undergone divergent devel-
opments in different areas, and neg2 may appear as: kra/ḵra/cra, ara, ka/ḵa/ca or k/
ḵ/c.17 Thus, the palatalisation of velars occurring in certain Berber varieties (above
all in the so-called Zanata dialects), has led to a form of neg2 comparable to the
corresponding Arabic morphemes (šay, ši or š from the classic word šay ʾ ‘thing’).18
As pointed out in the most recent studies, this tripartite classification is somewhat
sketchy and would benefit from the inclusion of subtypes based on the degree of
coexistence in a given language of the structures of each of the three main stages
(van der Auwera 2009, 2010: 75–85).
The similarity between the negative structures of North-African modern
Arabic dialects and those of Berber has prompted scholars to explore the ques-
tion of the origin of discontinuous morphemes within the framework of linguistic
interference:
“The fact that those varieties of Arabic and Berber which have reached stage II or
III of JC are spoken in largely the same geographical area raises the question of
whether the stage II construction was spread from one language to the other via
contact, and, if so, which was the source and which the target language as far as
this structure is concerned.” (Lucas 2007: 401)
. Other particles used for neg2 will not be examined here. Kossmann (2013: 332) suggests
that ani, used in Eastern Kabylia (Algeria), is derived from ani ‘where’. Lafkioui (2013a and
2013b) examines the Riffian neg2 bu, which has also spread to the neighbouring Arabic dia-
lects. Another particle attested in Eastern Kabylia is ula (Rabhi 1992 and 1996).
. Kossmann (2013: 333) set up a table summarizing the forms of neg2 in the principal
Berber varieties and regrouping them into three columns according to what he considers
surely Berber, surely Arabic and ambiguous.
Berber negation in diachrony
It is difficult to identify a clear-cut solution, given the lack of material from the
earliest stages of spoken Arabic and Old Berber. For this reason, Lucas’ suggestion
that, in Berber, Stage II “developed under the influence of Arabic” (2013: 402) is
not conclusive (see also Lafkioui 2013a for a critical discussion of Lucas’ hypoth-
esis). The main reason put forward concerns the areal distribution, “consistent
with a gradual spread westwards and southwards of the cycle in the local con-
tact varieties of Arabic” (Lucas 2013: 413). But the areal data may also be read in
the opposite way: in terms of the loss of a redundant feature in peripheral areas.
For instance, both the easternmost language, Siwa, and the westernmost, Zenaga,
no longer possess state opposition in nouns, but this alone does not justify the
straightforward assumption that this is an innovation they never shared with the
other Berber languages (Brugnatelli 1987b). Furthermore, the sporadic presence
of neg2 in Ancient Ibād�ī Berber, in both a more archaic form (-cra) and a phoneti-
cally reduced one (-c) seems consistent with viewing it as an ancient construction
that is losing ground, rather than as a newly acquired innovation. The influence
of Arabic might be seen, rather, as a stimulus to preserve neg2 in the Berber
varieties in which it had become similar to the Arabic equivalent, while most of
the so-called kem dialects, where neg2 did not undergo a palatalisation, have lost
it (Brugnatelli 1987a: 58).
A decisive argument in favour of a very early stage characterised by a twofold
negator across the whole Berber area derives, in my opinion, from the wide diffu-
sion of negative stems in all the verbal systems. For a more in-depth look at this
subject, see § 2 below.
. A different opinion on the origin of this particle is expressed by Chaker (1983: 121): “as
a matter of fact, ara is the amalgam of ay ‘this’ + ad ‘non-real’ ” (“ara est en fait l’amalgame de
ay ‘ce’ + ad ‘non-réel’ ”) (see also ibid. 159).
Vermondo Brugnatelli
As a matter of fact, a specific aorist particle that is used in relatives and wh-
interrogatives is also found outside of Kabylia. In the language of Figuig this
particle is ala (7). It has also been documented elsewhere, for instance in the
old language of Jerba (in a religious text composed at the beginning of the 19th
century) (8) and even in Medieval Berber from the Ibād�ite regions (Tunisia/
Libya) (9).
(7) Figuig: wi stt ala nawey?
who her ala take=aor=part
‘who will marry her?’ (Kossmann 1997: 268)
The similarity of forms and functions makes it probable that ala and Kabyle ara
are cognate. The only slight phonetic difference lies in the quality of the liquid,
much in the same way as the negative particles ur and ul or the prepositions ar
and al mentioned above.20 The Medieval Berber evidence is important not only
because it shows that this particle is old (while Mettouchi’s theory frames it as a
recent development), but also because it makes clear that it has nothing to do with
neg2, attested as c or cra.21
Most probably, another cognate is the particle la of Zuara (Libya), which is a
connective (“relateur”: Galand 2005: 192–193) introducing a relative that may be
either verbal or verbless:
(10) wuh la yemmut
that one la 3ms=die=pf
‘the one who is dead’
knečč la d_tamet�t�ut
me la pred_woman
‘I, who am a woman’
. Kabyle ara has several phonetic variants: aɣa, ɣa, aɛa, aa, ara. Other possible cognates
(such as Gourara ɣa or Mzab aɣa) are discussed in Brugnatelli (2011: 529–530).
. In the Medieval text, cra is also used as an indefinite pronoun in non-negative expres-
sions, such as cra n ‘some of ’, ‘a little of ’, in the same way as the Kabyle kra.
Berber negation in diachrony
At this early stage of the language, the particle was also used in cleft sentences, as
in the following example:
(14) d elqimeṯ en yelemlemen ǧ essuq ala yuc
pred price of y. in market ala 3ms=give=aor
‘it is the market price of y.-s [unknown word] that he will give’ (f. 403b, l. 8)
The fact that all these particles are strongly connected with relatives raises the
question of a possible link with the Tuareg particle éré, which is not a connective
but a pronoun – ‘anyone who’ – that is solely used as an antecedent of a relative
clause. Prasse (1972: 189) proposed the following etymology of the last-mentioned
term: “i-irâ, i-irê ‘un qu’il (on) veut’ ”, i.e. a pronoun i (“pronom d’appui”) followed
by the verb ăr ‘will, want’. Interestingly, this same construction is widely accepted
as the source of a “future” particle in Tashelhit (Southern Marocco), namely ra, rad
/ara, arad.24 An in-depth analysis of the usages and functions of all these gram-
matical devices falls outside the scope of this paper, but this brief comparative
outline is sufficient to support the claim that the Kabyle ara used as neg2 should
be regarded as independent of the aorist particle ara.
. Such cases of use with the perfective prevent us from considering ala a simple “variant
of the prospective preverb under ‘attraction’ [i.e. in relatives and wh-interrogatives: V.B.]”
(“variante du préverbe prospectif en attraction”, Kossmann 1997: 442).
. Interestingly, the same construction bə d
ʿ la ‘après que’ “after (that)” appears in Takrouna
Arabic (Tunisia): Galand (2005: 193). Maybe in this case a borrowing from Berber is more
probable than the other way round. See bə dʿ la with the same meaning.
. A possible link between the Kabyle ara and the Tashelhit rad has already been proposed
by Taine-Cheikh (2009: 98).
Vermondo Brugnatelli
Another distinctive feature of the Berber languages is the use of special forms of
the perfective and imperfective stems in negative constructions. Negative stems
are seldom used without the presence of other negators, yet their negative value is
indisputable and in some – admittedly few – cases they may be the only means of
expressing negation. For instance, in Kabyle (Dallet 1982: 530):
(15) a. mazal yet�t�es
still 3ms=sleep=pf
“he is still sleeping”
b. mazal yet�t�is
still 3ms=sleep=npf
“he is not yet sleeping”
The most widespread form is the negative perfect, which appears in nearly all
the varieties, while the negative imperfective is less generalized but nonetheless
scattered across the whole area and should probably also be considered a com-
mon form. This is further confirmed by the fact that the negative imperfective is
attested in ancient texts even in areas in which it is not currently in use.
Both perfective and imperfective stems undergo similar modifications in the
negative form. A survey reported in Brugnatelli (2002: 168) indicated that these
changes may be summarised as follows:
. For the changes in vocalism of the perfective in the negative form as well as in the so-
called resultative, see Brugnatelli (2005: 376–378).
Berber negation in diachrony
b. – Jerba
yez�wa yez�wi iz�ugga yez�uggi ‘go down’
yebbes yebbis yetbessa yetbessi ‘be switched off ’
yeweṯ = yeččaṯ yeččiṯ ‘strike’
From a diachronic point of view, these forms are easily explained as the result of
phonotactic changes involving the final part of the stem, possibly under the influ-
ence of a suffixed negative particle. This would account for the shortening of the
beginning vowels and the lengthening of the final ones (via a left-to-right stress
shift) as well as for the fronting of (final) vowels (as a consequence of Umlaut,
assuming that the original particle contained front vowels).
An interesting parallel comes from the Arabic dialects of Egypt (Dakhla
Oasis), in which negative verbal forms have arisen from positive ones, displaying
a vocalic difference most likely provoked by “consonant clustering and heavy syl-
lable formation”, due to the affixation of neg2 (Woidich 1995–97: 14–15):
(17) Western dialects: i ʾgoːm > ma-ti ʾgaːm-š ‘do not speak Cairene!’
Central dialects: si ʾaːn > ma-si ʾin-š/ma-si ʾeːn- š «he did not ask».
. “Une ancienne forme intensive qui devait être employée dans des environnements à
forte modalisation: énoncés négatifs (interdiction), de souhait, d’hypothèse irréelle, etc.”
Vermondo Brugnatelli
The existence of negative verb stems in all Berber languages could be viewed
in itself as a pan-Berber strategy of double-marking the negation (Schmitt-Brandt
1979: 235; Brugnatelli 1987a: 59; Lafkioui 2013a: 87). The fact that such forms
probably derived from elements placed towards the right end of the verbal com-
plex strengthens the hypothesis that Berber achieved Stage II of the Jespersen cycle
in very ancient times, earlier than any contact with Arabic. The actual shape of the
added element is difficult to reconstruct. It is even possible that the attested forms
of neg2 are innovated forms which replaced or were added to earlier morphemes,
given that the most important phonetic change is palatalisation, which in principle
entails the presence of a front vowel. We find similar phenomena in many other
world languages, such as the Old Irish genitive maicc (from macc ‘son’), in which
a final -i, preserved in Ogam maqqi, had completely disappeared leaving only a
phonetic vestige in the palatalisation of the final consonant, or the well-known
phenomenon of Umlaut in German, in which final vowels undergo fronting under
the influence of i-endings that have now disappeared.
3. Conclusion
References
Basset, André. 1940. Quatre études de linguistique berbère: 1. gar ‘mauvais’; 2. Touareg әrk
‘mauvais’; 3. war; 4. bla en berbère. Journal Asiatique 232: 161–291.
Boumalk, Abdallah. 1996. La négation en berbère marocain. In Chaker & Caubet: 35–48.
Brugnatelli, Vermondo. 1987a. La negazione discontinua in berbero e in arabo-magrebino. In
Atti della 4. Giornata di Studi Camito-semitici e Indeuropei, Bergamo 28 November 1985,
Giuliano Bernini & Vermondo Brugnatelli (eds), 53–62. Milano: Unicopli.
Brugnatelli, Vermondo. 1987b. Deux notes sur l’état d’annexion en berbère. In Proceed-
ings of the Fourth International Hamito-Semitic Congress [Current Issues in Linguistic
Theory 44], Hermann Jungraithmayr & Walter W. Mueller (eds), 349–359. Amsterdam:
John Benjamins.
Brugnatelli, Vermondo. 1993. Quelques particularités des pronoms en berbère du Nord. In À
la croisée des études libyco-berbères. Mélanges offerts à Paulette Galand-Pernet et Lionel
Galand, Jeanine Drouin & Arlette Roth (eds), 229–245. Paris: Geuthner.
Brugnatelli, Vermondo. 2002. Les thèmes verbaux négatifs du berbère: Quelques réflexions. In
Articles de linguistique berbère. Mémorial Werner Vycichl, K. Naït-Zerrad (ed.), 165–180.
Paris: L’Harmattan.
Brugnatelli, Vermondo. 2005. Voyelles et accents dans l’histoire du berbère. In Proceedings of
the 10th Meeting of Hamito-Semitic (Afroasiatic) Linguistics, Florence, 18–20 April 2001.
[Quaderni di Semitistica 25], Pelio Fronzaroli & Paolo Marrassini (eds), 371–380. Firenze:
Dip. di Linguistica-Università.
Brugnatelli, Vermondo. 2006. La négation berbère dans le contexte chamito-sémitique. In Les
langues chamito-sémitiques (afro-asiatiques), Vol. 2 [Faits de Langue 27], Atoine Lonnet &
Amina Mettouchi (eds), 65–72. Paris: Ophrys.
Brugnatelli, Vermondo. 2010. Problème de la négation en berbère: À propos de l’origine d’ulac,
ula, ula d. In Camsemud 2007. Proceedings of the 13th Italian Meeting of Afroasiatic Linguis-
tics held in Udine, May 21st–24th, 2007, Fredrick Mario Fales & Giulia Francesca Grassi
(eds), 401–405. Padova: S.A.R.G.O.N.
Brugnatelli, Vermondo. 2011. Négations, participes et figement en berbère: Nouvelles hypothèses.
In Parcours berbères. Mélanges offerts à Paulette Galand-Pernet et Lionel Galand pour leur
90e anniversaire, Amina Mettouchi (ed.), 521‒532. Cologne: Rüdiger Köppe.
Brugnatelli, Vermondo. 2014. Typology of eastern medieval Berber. STUF – Language Typology
and Universals 67(1): 127–142.
Chaker, Salem. 1983. Un parler berbère d’Algérie (Kabylie). Syntaxe. Aix-en-Provence: Publica-
tions Université de Provence.
Chaker, Salem. 1996. Quelques remarques préliminaires sur la négation en berbère. In Chaker &
Caubet (eds), 9–22.
Chaker, Salem & Caubet, Dominique (eds). 1996. La négation en berbère et en arabe maghrébin.
Paris: L’Harmattan.
Vermondo Brugnatelli
Dallet, Jean-Marie. 1982. Dictionnaire kabyle-français. Parler des At Mangellat. Paris: SELAF.
Drouin, Jeannine. 1996. Les formes participiales en berbère. Essai de dialectologie comparée.
Littérature Orale Arabo-Berbère 24: 233–260.
Galand, Lionel. 1994. La négation en berbère. Matériaux Arabes et Sudarabiques 6: 169‒81.
Galand, Lionel. 2005. Quelques traits du parler berbère de Zouara (Libye). Studi Maġrebini 3:
187–195.
Galand, Lionel. 2010. Regards sur le berbère. Milano: Centro Studi Camito-Semitici.
Galand, Lionel. 2011. Deux mille phrases dans un parler berbère du Maroc. Application et évalu-
ation de la méthode d’enquête linguistique d’Henri Frei. Rabat: Institut Royal de la Culture
Amazighe.
Kahlouche, Rabah. 2000. Le présentatif négatif ulac ‘il n’y a pas’, est-il de souche berbère ou
un emprunt à l’arabe? In Études berbères et chamito-sémitiques. Mélanges offerts à Karl-G.
Prasse, Salem Chaker & Andrzej Zaborski (eds), 233–236. Louvain: Peeters.
Kossmann, Maarten. 1989. L’inaccompli négatif en berbère. Études et Documents Berbères 6:
19–29.
Kossmann, Maarten. 1997. Grammaire du parler berbère de Figuig (Maroc oriental). Louvain:
Peeters.
Kossmann, Maarten. 2013. The Arabic Influence on Northern Berber. Leiden-Boston: Brill. DOI:
10.1163/9789004253094
Lafkioui, Mena. 1996. La négation en tarifit. In Chaker & Caubet, 49–77.
Lafkioui, Mena. 2007. Atlas linguistique des variétés berbères du Rif. Köln: Rüdiger Köppe.
Lafkioui, Mena. 2011. Études de la variation et de la structuration linguistiques et sociolinguis-
tiques en berbère. Köln: Rüdiger Köppe.
Lafkioui, Mena. 2013a. Reinventing Negation Patterns in Moroccan Arabic. In African Arabic:
Approaches to Dialectology [Trends in Linguistics. Studies and Monographs 25], Mena
Lafkioui (ed.), 51–94. Berlin: De Gruyter Mouton.
Lafkioui, Mena. 2013b. Negation, grammaticalization and language change in North Africa:
The case of the negator NEG___bu. In Tilelli. Scritti in onore di Vermondo Brugnatelli,
Giorgio Francesco Arcodia, Federica Da Milano, Gabriele Iannàccaro & Paolo Zublena
(eds), 113‒130. Cesena: Caissa.
Loubignac, Victorien. 1924. Étude sur le dialecte berbère des Zaïan et Aït Sgougou. Paris: Leroux.
Lucas, Christopher. 2007. Jespersen’s cycle in Arabic and Berber. Transactions of the Philological
Society 105(3): 398–431. DOI: 10.1111/j.1467-968X.2007.00189.x
Lucas, Christopher. 2013. Negation in the history of Arabic and Afro-Asiatic. In The History
of Negation in the Languages of Europe and the Mediterranean, Vol. I: Case Studies, David
Willis, Christopher Lucas & Anne Breitbarth, 399–452. Oxford: OUP.
Marcy, Georges. 1936. Note sur le pronom relatif-sujet et le pseudo-participe dans les parlers
berbères. Bulletin de la Société de Linguistique de Paris 37: 45–57.
Marcy, Georges. 1940–41. Observations sur le rélatif futur en touareg Ahaggar. Bulletin de la
Société de Linguistique de Paris 41: 129‒133.
Mettouchi, Amina. 1996. La négation dans les langues du Maghreb: Synthèse. In Chaker &
Caubet (eds), 177–195.
Mettouchi, Amina. 2001. La grammaticalisation de ara en kabyle, négation et subordination rela-
tive. In Travaux du CerLiCO 14, Gilles Col & Daniel Roulland (eds), 215–235. Rennes: P.U.
Mettouchi, Amina. 2009. The system of negation in Berber. In Patterns of Negation in West African
Languages [Typological Studies in Language 87], Norbert Cyffer, Erwin Ebermann &
Georg Ziegelmeyer (eds), 287–306. Amsterdam: John Benjamins.
Berber negation in diachrony
Mitchell, Terence F. 2009. Zuaran Berber (Libya). Grammar and Texts. Köln: Rüdiger Köppe.
Picard, André. 1957. Du prétérit intensif en berbère. In Mémorial André Basset (1895–1956),
107–120. Paris: Adrien Maisonneuve.
Prasse, Karl-G. 1972. Manuel de grammaire touarègue (tăhăggart), I–III: Phonétique Écriture-
Pronom. Copenhagen: Éditions de l’Université.
Prasse, Karl-G. 1973. Manuel de grammaire touarègue (tăhăggart), VI–VII: Verbe. C openhagen:
Éditions de l’Université.
Provotelle, Paul. 1911. Étude sur la tamazir’t, ou zénatia de Qalaât es-Sened. Paris: Ernest Leroux.
van Putten, Marijn. 2013. A Grammar of Awjila Berber (Libya). Based on Umberto Paradisi’s
Material. Ph.D. dissertation, Leiden.
Rabhi, Allaoua. 1992. Les particules de négation dans la Kabylie de l’Est. Études et Documents
Berbères 9: 139–146.
Rabhi, Allaoua. 1996. De la négation en berbère: les données algériennes. In Chaker & Caubet
(eds), 23–48.
Schmitt-Brandt, Robert. 1979. Berberische Adstrateinflüsse im maghrebinischen Arabisch.
Folia Linguistica 13: 229–235. DOI: 10.1515/flin.1979.13.3-4.229
Taine-Cheikh, Catherine. 2009. Les morphèmes de futur en arabe et en berbère. Réflexions pour
une typologie. Faits de Langues 33: 91–102.
Taine-Cheikh, Catherine. 2011. L ’énoncé négatif en berbère zénaga. In Parcours berbères.
Mélanges offerts à Paulette Galand-Pernet et Lionel Galand pour leur 90e anniversaire,
Amina Mettouchi (ed.), 533–551. Köln: Rüdiger Köppe.
van der Auwera, Johan. 2009. The Jespersen cycles. In Cyclical Change [Linguistik Aktuell/
Linguistics Today 146], Elly van Gelderen (ed.), 35–71. Amsterdam: John Benjamins.
van der Auwera, Johan. 2010. On the diachrony of negation. In The expression of Negation,
Laurence R. Horn (ed.), 73–109. Berlin: De Gruyter Mouton.
van der Auwera, Johan, Vossen, Frens & Devos, Maud. 2013. Le cycle de Jespersen à trois ou
quatre négations. In La linguistique de la contradiction, Jacques François, Pierre Larrivée,
Dominique Legallois & Franck Neveu (ed.), 19–30. Bern: Peter Lang.
Woidich, Manfred. 1995–97. Negation in the Egyptian Arabic dialect of the Dakhla-Oasis: A
case of rule morphologization. Mediterranean Language Review 9: 13–28.
The grammaticalization of negative indefinites
The case of the temporal/aspectual n-words plus and
mais in Medieval French
This chapter traces the diachronic evolution in Medieval French of two temporal/
aspectual n-words of adverbial origin, the markers mais (< Lat. MAGIS ‘more,
to a greater degree’) and plus (< Lat. PLUS ‘more’), equivalent to English no
more/no longer, anymore/any longer, with a view to addressing two theoretical
issues: (i) whether the evolution of indefinites is unidirectional, as claimed
by Haspelmath (1997), or rather subject to a “random diachronic walk”, as
suggested by Jäger (2010); (ii) whether the evolution of the French n-words can
be described in terms of a general cyclical development parallel to that of the
standard clause negator (commonly known as Jespersen’s Cycle). It is argued that,
like those reported in Hansen (2012), the results of this study support the random
walk hypothesis, and weakens the case for a quantifier cycle in French. Moreover,
the results of the two studies suggest that functional categories, or paradigms, are
pragmatic, rather than linguistic, entities.
1. Introduction
. The notion of “cycles” is used loosely in this paper, following the terminology used in
Willis et al. (2013: 27f).. As will become clear below, even standard clause negation in French
has not completed an actual diachronic cycle, in as much as the postverbal negative marker
has so far failed to move to a preverbal position (except in infinitival clauses). The notion of a
“quantifier cycle” is adapted from Ladusaw’s (1993) term “argument cycle”, and should not be
taken to imply an expectation that the contemporary French n-words will someday become
polarity-neutral again.
Maj-Britt Mosegaard Hansen
Modern French developed as a group, along similar lines, and for similar reasons,
possibly as a result of prior developments in the expression of standard clause
negation, as proposed by Zeijlstra (2008).
Déprez and Martineau’s (2004) study of the n-words rien (‘nothing/any-
thing’), personne (‘nobody/anybody’), and aucun (‘no(ne)/any’), all of which are
of nominal origin and prototypically function as valency elements in negated
clauses, convincingly suggests, however, that the diachronic evolution of these
items was strongly influenced by developments elsewhere in the nominal system.
The account offered by Déprez & Martineau (2004) is therefore, as pointed out
by Hansen (2011: 273), inherently unable to account for n-words of non-nominal
origin. In addition, Hansen (2012) provides evidence that the temporal adverbial
n-word jamais (‘(n)ever’) developed in a way that runs counter to what one would
expect if a general quantifier cycle were operative in French. Further in-depth
studies of individual n-words thus appear to be called for.
More generally, it has been proposed by Haspelmath (1997: 230ff) that the
evolution of indefinite quantifiers is unidirectional, going from positive through
affective to negative uses. In counterdistinction to this, Jäger (2010) proposes a non-
unidirectional “random walk” model of the diachronic development of indefinites.
While aspects of Jäger’s model are problematic (cf. Willis 2011; van der Auwera &
Van Alsenoy 2011), Hansen’s (2012) study of jamais supports the idea of at least par-
tial non-unidirectionality, and again calls for further studies of individual n-words.
The present study traces the diachronic evolution in Medieval French of two
temporal/aspectual n-words of adverbial origin, the markers mais (< Lat. MAGIS
‘more, to a greater degree’) and plus (< Lat. PLUS ‘more’).2 Throughout the Medi-
eval period, mais and plus were in competition in contexts of quantifier negation,
where they were used in combination with preverbal ne to convey a temporal /
aspectual meaning equivalent to English no more/no longer, anymore/any longer,
as in (1)-(2):
(1) “Sire, je ne puis plus chi demorer,…”
(Merlin, p. 145, c. 1230–35 – from BFM)
‘Sire, I can no longer stay here…/I can’t stay here any longer,…’
(2) Et Gavains, qui avoit ja si avanchié son caup qu’il ne le puet mais retenir.
(Merlin, p. 166, c. 1230–35 – from BFM)
‘And Gawain, who had already gone so far in casting his blow that he can no
longer hold it back/… that he cannot hold it back any longer.’
. As Medieval French had no standard orthography, the exact forms reproduced in the
examples vary. This is particularly the case with mais.
The grammaticalization of negative indefinites
In Modern French, only (ne…)plus remains, the relevant sense of mais having
disappeared in the course of the 16th c.3 A tentative explanation of why this hap-
pened will be proposed.
It will be argued that the evolution of mais and plus, like that of jamais, pro-
vides evidence against the notion that developments in the quantifier domain are
unidirectional, and thus further weakens the idea of a general quantifier cycle in
French. Moreover, it will be argued that the evolution and contemporary status
of plus support Hansen’s (2008, 2012) view of functional categories, or functional
paradigms,4 as pragmatic rather than linguistic entities.
. This study adheres to the following commonly accepted periodization of the history of
the French language:
Table 1. The evolution of French clause negation (sample sentence: ‘I do not say…’)5
Stage 0. [Classical Latin] non dico The negator is preverbal
Stage 1. [Proto-French] je ne dis The preverbal negator is
phonetically reduced
Stage 2. [Old/Middle French] je ne dis (pas/mie/point) The preverbal negator is optionally
complemented by a postverbal
element
Stage 3. [Classical French] je ne dis pas The postverbal element
grammaticalizes as part of a
discontinuous negator embracing
the verb
Stage 4. [Modern/ je (ne) dis pas The original preverbal negator
Contemporary French] becomes optional
Stage 5. [Future French?] je dis pas The negator is postverbal
Stage 6. [Louisiana French mo pa di The previously postverbal negator
Creole]5 migrates to preverbal position
In the Medieval period, which is of primary interest to this study, Stage 2, i.e.
the mirror image of the present-day situation, obtained: the preverbal clitic was
the principal clause negator, which could be optionally reinforced by various post-
verbal items. Some such items, principally pas (< Latin PASSU(M) ‘step’), mie (<
Latin MICA(M) ‘crumb’), and point (< Latin PUNCTU(M) ‘point’), were used to
reinforce the most basic form of clausal negation (although, as shown by Hansen
2009, initially only in pragmatically marked contexts), (3)-(4) being thus synony-
mous at the truth-conditional level.
(3) Je nel vos dirai.
(4) Je nel vos dirai pas/mie/point.
‘I won’t tell you’, lit. ‘I will not say it to you.’
. Clearly, we cannot assume that Creoles necessarily represent future stages of the super-
strate language. Louisiana French Creole is thus adduced principally to illustrate the full
Jespersen Cycle. That said, French negated infinitives have in fact already completed the cycle,
as shown in (i)(cf. Martineau 1994):
clause in its entirety, as illustrated in (5). Just as Medieval speakers had a choice
of postverbal markers of standard clause negation at their disposal, so they had
a variety of ways of expressing zero quantification of particular types of constit-
uents in negative clauses. Thus, as seen in (5), the nouns rien and chose (both
‘thing’) could both be used with an essentially pronominal function, to represent
inanimate referents:6
(5) Je ne voy rien qui ne m’anuye, Et ne sçay chose qui me plaise (Ch. D.Orl.,
457, 2, early 15th c. – from Martin 1966: 234)
‘I see nothing that doesn’t upset me and I know of nothing that pleases me’,
Lit.: ‘I don’t see thing that doesn’t upset me, And I don’t know thing that
pleases me’
Similarly, in Medieval French, mais and plus could be used with adverbial func-
tion, to express quantification of a temporal/aspectual nature. However, just as the
choice of postverbal markers of standard clause negation was gradually reduced,
leaving pas as the only register-neutral option in contemporary French,7 so the
available quantifiers were gradually reduced, essentially leaving only one marker
for each relevant type of constituent.8 Thus, in Modern Standard French, tempo-
ral/aspectual plus forms part of a closed set of so-called n-words (Laka Muzarga
1990: 107), i.e. items that are used for purposes of quantification in both nega-
tive and certain weak negative-polarity (or ‘affective)’ contexts. Apart from plus,
the French n-word paradigm includes the items personne (‘nobody/anybody’, <
Latin PERSONA ‘mask, character’), rien (‘nothing/anything’, < Latin REM ‘thing’),
aucun (‘no(ne)/any’, < Latin ALIQUIS UNUS ‘some one’), nulle part (‘nowhere/
anywhere’, < Latin NULLA PARTE ‘in no place’), and jamais (‘(n)ever’, < Latin
. According to Martin (1966: 229), chose was principally used in contexts where it was fol-
lowed by a relative clause, as in (5).
. In principle, point remains possible alternative, but it is both archaic and marked for a
high degree of formality.
. In principle, the two quantifiers aucun and nul (< Latin NULLUS ‘no(ne)’) can still both
be used as animacy-neutral pronouns and determiners. Likewise, contemporary colloquial
French possesses a few expressions that function as alternatives to the ones listed in the main
text, the most prominent one being the expression que dalle (‘nothing’), as in (i). However, like
point, nul is confined to formal registers in contemporary French, while, conversely, que dalle
is strongly marked for informality. These expressions are therefore left out of consideration
here.
(i) Je pige que dalle, moi.
‘I understand nothing.’
Maj-Britt Mosegaard Hansen
IAM MAGIS ‘from now on more’). The first three function pronominally, as
valency elements, or alternatively, in the case of aucun, as a determiner, while the
last two have adverbial functions, respectively marking place and time.
The semantic status of the Modern French n-words is, however, problematic
for a number of reasons:
As mentioned above, they typically occur in negative contexts with seemingly
inherent negative meaning. In this, they resemble negative indefinites like those of
the Standard English no-series. The interpretation of the French n-words as inher-
ently negative is supported by two distributional facts:
Yet negative concord in Modern Standard French9 does not extend to combina-
tions of n-words with the standard clause negator pas. Such combinations invari-
ably result in DN interpretations, cf. (10), making French a “negative-spread”
language, rather than a strict-NC language (cf. de Swart 2010: 46):
. This is not true of all dialects. Quebecois, for instance, allows an NC interpretation of (10).
The grammaticalization of negative indefinites
Finally, the French n-words also occur, with non-negative meaning, in so-called
affective, or weak negative-polarity, contexts (cf. Klima 1964: 313) such as inter-
rogatives, conditionals, comparatives etc. cf. (11).
(11) Elle travaille mieux qu’aucun autre étudiant.
‘She works better than any other student.’
For the reasons above, these items cannot be clearly identified as either inherently
negative quantifiers or negative-polarity items (NPIs).
Further complicating the picture, the individual n-words form a cline in
terms of their potential to appear in weak negative-polarity contexts. As Table 2
shows, jamais is acceptable in the widest range of such contexts, while plus is
clearly acceptable in only one (and marginally so in two others). Thus, plus is
common after the preposition sans (‘without’), as in (12). The status of sans
as a strong vs weak negative polarity environment will be discussed further in
Section 6 below:
(12) Il est parti sans plus attendre.
‘He left without waiting any longer.’
. Interestingly, Hoeksema (2007) shows that meer, the closest Dutch equivalent of plus,
is similarly quite restricted in its distribution (as are English anymore and German mehr,
according to the data presented in Hoeksema 2013), but that it may occasionally be parasiti-
cally licensed in other environments. Why this should be so remains to be elucidated.
Maj-Britt Mosegaard Hansen
Table 2. Negative polarity uses of French n-words (adapted from Muller 1991: 265)11
N-word jamais rien/aucun/ nulle part plus
Negative polarity context personne
Given that, like the standard clause negator pas, all the n-words listed above,
with the exception of nulle part, are derived from items that originally had polar-
ity-neutral meanings, it is tempting to posit the existence of a quantifier cycle,
as in Table 3, parallel to the basic Jespersen Cycle, with Modern/Contemporary
French at Stage 3 (cf. Ladusaw 1993).12 Such a cycle would be compatible with
. An anomymous referee asks if this takes into account the nature of the matrix verb,
Guerzoni & Sharvit (2007) having shown that NPI licensing in English indirect interrogatives
is variable, such that interrogatives embedded under predicates like wonder accept NPIs more
easily that those governed by predicates like know, cf. (i)-(ii)(where % indicates cross-speaker
variability):
(i) Claire wonders which students have any books on Negative Polarity.
(ii) %Claire knows which students have any books on Negative Polarity.
It is not clear from Muller’s (1991) discussion whether he has explicitly considered this issue.
However, while it seems intuitively that, indeed, not any indirect interrogative will accept
n-words with a non-negative interpretation in French, the question appears to be relevant
only for jamais, none of the other n-words (incl. plus) being admissible at all. For that reason,
I leave the matter out of consideration in the remainder of this paper.
. Note that Ladusaw (1993) does not speak of a “quantifier cycle” as such, but of an “argu-
ment cycle”. Strictly speaking, he is thus making no claims about the diachronic evolution of
indefinites with non-argument functions, such as plus, jamais, and (in most of its uses) nulle
The grammaticalization of negative indefinites
Nor is there any doubt that these indefinites have undergone grammaticaliza-
tion, in as much as their gender has changed from the more marked feminine
to the unmarked masculine, and they have undergone decategorialization, losing
their capacity to accept both determination and direct modification (cf. Déprez
2011: 251ff). With respect to aucun, Déprez & Martineau (2004) show that,
as a pronoun, this item was predominantly used in polarity-neutral and weak
negative-polarity contexts in Renaissance French, with negative contexts con-
stituting a clear minority. The proportions were, however, gradually reversed,
such that aucun is now favored in strong negative contexts. Just as the noun rien
part. Ladusaw does, however, argue that his proposed cycle is intimately bound up with the
development of negative concord. If that is the case, it is unclear why the adverbial n-words
should be excluded from the proposed cycle, given that they participate in NC on an equal
footing with the argument terms rien, personne, and aucun.
Maj-Britt Mosegaard Hansen
eventually disappeared, positive uses of aucun are completely absent from the
contemporary language, apart from the frozen (and stylistically marked) expres-
sion d’aucuns (‘some’). The evolution of the three indefinites rien, personne, and
aucun thus supports the hypothesis of positive > negative unidirectionality in the
evolution of quantifiers.
However, as shown in Hansen (2012), the adverbial n-word jamais pres-
ents a problem for the hypothesized quantifier cycle and for Haspelmath’s
hypothesis about the grammaticalization of indefinites. The data adduced in
that paper strongly suggest that jamais was lexicalized as a negative marker
at an early stage, and only subsequently developed its uses in weak negative-
polarity contexts. Although its component morpheme ja was polarity-neutral
(Hansen 2014), no attestations of jamais in polarity-neutral contexts were found
in Hansen’s (2012) data, apart from sporadic instances of the frozen expression
à/pour jamais (‘forever’).
In combination with the observations regarding contemporary NPI uses of
the French n-words, summarized in Table 2 above, this raises the question of the
unity of the n-word paradigm in that language, and makes it relevant to take a
closer look at the evolution of mais and plus in negative-polarity contexts. Both
these markers are adverbial in nature just like jamais, and it is at least conceivable
that there might be a divide between nominal n-words on the one hand and adver-
bial n-words on the other. At the same time, however, contemporary plus behaves
quite differently from jamais in affective contexts, a fact which may or may not be
attributable to different developmental paths.
As noted above, mais and plus as n-words both originate in polarity-neutral Latin
adverbs expressing comparative degree and/or quantity, e.g. (16)-(17). Latin
MAGIS is related to the adjective MAGNUS (‘great’), and basically expressed
degree, while PLUS is the comparative of MULTUS (‘much’), and thus tended to
express quantity in a more concrete sense. That said, these two kinds of meaning
overlap in many contexts. Neither of the Latin adverbs appears to have been used
in negative contexts with a temporal/aspectual meaning similar to that of their
French descendants.
(16) Non magis Alexandri saevitiam quam Bessi parricidium ferre potuisse.
(Quintus Curtius Rufus, Historiae Alexandri Magnus 7, 6, 15)
‘That they had not been able to bear Alexander’s savagery any more (i.e. ‘to
any greater degree’) than the Bessian’s parricide.’
The grammaticalization of negative indefinites
The temporal/aspectual sense of mais and plus thus represents a later develop-
ment. The Latin uses are, however, preserved in the two Medieval French adverbs,
and in the contemporary language, plus remains prominently used as an adverb of
both degree and quantification, as shown in (18)–(19).13
(18) Ce vin est plus cher que l’autre.
‘This wine is more expensive than the other one.’
(19) On reçoit plus de pluie à Manchester qu’à Londres.14
‘We get more rain in Manchester than in London.’
Although highly unlikely to actually be uttered for reasons of euphony (as well as –
in the case of (21) – extralinguistic reasons), (20)–(21) would thus in principle be
grammatically correct sentences in Modern French:
(20) Ce vin n’est plus plus cher que l’autre.
‘This wine is no longer more expensive than the other one.’
(21) On ne reçoit plus plus de pluie à Manchester qu’à Londres.
‘We no longer get more rain in Manchester than in London.’
. While the morpheme mais still exists in Modern French, it has lost all the uses men-
tioned above, and functions only as an adversative connective corresponding to English but, a
use that is attested since Old French, and which, as suggested by Ducrot and Vogt (1979: 318)
can be plausibly derived from Latin structures like that in (i):
(i) Id, Manli, non est turpe, magis miserum est (Catullus, 68, 30)
‘That, Manlius, is not shameful, it’s more sad.’ >> ‘That’s not shameful, but sad.’
. Plus would tend to be pronounced with a final /s/, as /plys/ in this use, to distinguish it
from the n-word, which is always pronounced /ply/.
Maj-Britt Mosegaard Hansen
In weak negative polarity contexts, Medieval French mais and plus have continu-
ative meaning, corresponding to encore/toujours in Modern French. Thus, as in
their negative uses, they presuppose the validity of the state-of-affairs at some
point in time prior to topic time, and in addition they suggest the possibility of its
continued or renewed validity at topic time or beyond, without however, asserting
it (cf. (23)-(24)):
(23) Peu voy cité, peu voi chastel Ou il ait mes ancïen prince.
(G. de Coinci 4, p. 26, v. 653, c. 1218–27 – from BFM)
‘I see few towns, I see few castles where there is still an old prince/where
there is an old prince anymore.’
Both these meanings can be related to the original quantifying senses of the
markers if time is conceived metaphorically as a scalar phenomenon, such that
temporal continuity of a state of affairs consists in the addition of points or
intervals to an existing scale. Conversely, a state-of-affairs is discontinued when
t emporal points/intervals cease to be added to the scale, in other words when the
end of the scale has been reached.
Both mais and plus are also sporadically found in positive contexts. With
plus, this is only the case in my 16th-century data, while positive temporal mais is
attested in the corpus from the 12th to the 14th century. Neither of the two mark-
ers has phasal continuative meaning in positive contexts. While clearly referring
to temporal duration, the plus in (25) below is essentially comparative in mean-
ing, and would need to be followed by an adverb like longtemps (‘a long time’) to
be acceptable in this kind of context in Modern French. Positive mais has incho-
ative meaning similar to that of Modern French désormais (‘henceforth’), as seen
in (26):
(25) Ma dame, je viens de sainct François où j’ay sejourné (peut estre) plus que
mon devoir ne requeroit ;
(Pierre Boisteau, Histoires tragiques, 116, 1559 – from DMF)
‘My lady, I come from Saint Francis where I stayed (perhaps) longer than
my duty required;’
(26) “Vo pooirs, fait ele, et li nostres Doit mais estre tout une cose.”
(Escoufle, p. 272, v. 8377, c. 1200–02 – from BFM)
‘Your power, says she, and ours must henceforth be one and the same.”
4. Data
. Both the BFM and the DMF are evolving data bases. The information provided here is
valid for the time when the data for this study was downloaded, in January 2012.
Maj-Britt Mosegaard Hansen
has been subdivided into two periods of 50 years each, with the distribution of
texts and number of words likewise represented in Table 4 below.
In addition, Table 4 gives an overview of the total number of tokens of
(potentially) temporal uses of mais and plus, and of the overall occurrence of the
two items out of the total number of words in the sub-corpora.
A few important caveats must be noted with respect to the BFM data in
particular:
While all or most of the 16th c. DMF corpus can be assumed to be based on
printed texts, the data from BFM originate in critical editions of originally hand-
written manuscripts. Thus, the possibility that, where those critical editions are
based on more than one manuscript, there may be variation in the use of mais and
plus among those manuscripts cannot be excluded. The original Medieval manu-
scripts have not been consulted for the purposes of this study.
Furthermore, the BFM data represent a variety of Northern French (i.e. langue
d’oïl) dialects, incl. Anglo-Norman. Although from the 12th c. onwards, all but
three of the texts contain tokens of both mais and plus with (potentially) temporal
meaning, dialectal preferences for one or the other likewise cannot be excluded,
but have not been systematically investigated. There is, however, no indication
of such preferences to be found in a number of existing grammars of Medieval
French (Foulet 1965; Togeby 1974; Moignet 1976; Buridant 2000).
No. of texts 3 4 11 5 6 11 4
No. of words 5,171 32,981 756,230 331,724 357,233 483,076 320,771
Tokens of temporal mais 5 2 351 14 23 3 0
Tokens of temporal plus 0 19 247 67 118 194 104
Overall occurrence of .00096 .00006 .00046 .000042 .000064 .0000062 0
temporal mais per
1,000 words
Overall occurrence of 0 .00057 .00032 .0002 .00033 .0004 .00032
temporal plus per
1,000 words
(27)
Mais un de nus valt de ces cent
(Wace, Brut, p. 647, v. 12428, 1155 – from BFM)
‘But one of us is worth one hundred of those.’
(28)
Mes sires Gauvains fu navrez ou costé senestre,
(Graal, p. 152, c. 1220 – from BFM)
‘My Lord Gawain was wounded on the left side,’
In other words, the use of Latin MAGIS as a non-temporal degree adverb is not
attested at all as a possible use of mais in my corpus. In the DMF data, temporal/
aspectual mais is only found in the first half of the 16th c., and only a total of 3
tokens, or 0.2% of the total number of instances of the form mais, have this mean-
ing, all of them forming part of the fixed expression n’en pouvoir mais (‘to not be
able to go on any longer’). All the remaining 1,650 16th c. tokens represent the
adversative conjunction.
Potentially temporal/aspectual uses of plus constitute a slightly smaller, but
broadly comparable, 7.86% of the total number of tokens of that marker in the
BFM and 8.85% of the 16th c. DMF tokens. The remaining tokens are used as
adverbs of quantification or degree, as in (29)-(30):
(29) Reis Evander e Catellus E des altres cinc cenz u plus Furent ateint e abatu,
(Wace, Brut, p. 637, v. 12238, 1155 – from BFM)
‘The kings Evander and Catellus and five hundred or more of the others
were caught and killed,’
(30) Dunc li remembret de sun seinor celeste, Que plus ad cher que tut aveir
terrestre. (Alexis, p. 95, v. 58, c. 1050 – from BFM)
‘Then he remembers his heavenly lord, whom he holds more dear than all
earthly possessions.’
. In principle, mais could be interpreted as similarly ambiguous in certain examples, but
given that, as stated above, this marker is not unambiguously attested with a meaning of quan-
tity or degree, I take it that its intended meaning is continuative in the examples in question.
Maj-Britt Mosegaard Hansen
(31) A Dieu plaise que je ne vive gueres ! Au moins fussés vous quite de moy et
n’eussés plus de desplesir de moy. (QJM, p. 9, c. 1400 – from BFM)
‘May it please God that I not live much longer! At least you’d be rid of me
and would not have (even) more displeasure from me/and would no longer
have displeasure from me.’
In the case of plus, two further context types have been added:
iv. Negative ambiguous contexts, where the marker cooccurs with ne, but where
it cannot be determined with absolute certainty whether ne…plus is temporal/
aspectual in meaning (equivalent to Modern French ne…plus ‘no longer’) or
odern French ne…
indicates the absence of an added quantity (equivalent to M
pas plus ‘not more’);
v. Affective ambiguous contexts, where plus may either be read as meaning ‘any
longer’ or as indicating an added quantity, i.e. meaning ‘more’.
As Table 6 below clearly shows, up until the 16th c., a substantial portion of the
occurrences of plus in both negative and affective contexts are not clearly tempo-
ral/aspectual in meaning, but are potentially interpretable as adverbs of quantity
or degree. It is plausible that these ambiguous instances provide bridging con-
texts (Heine 2002) for the meaning extension that plus underwent. That assump-
tion is rendered initially problematic by the fact that there are no examples of
ambiguous contexts prior to the appearance of occurrences of plus that are clearly
The grammaticalization of negative indefinites
temporal/aspecual in meaning. For one thing, however, that may simply be due
to the sparsity of both 11th and 12th-century texts in the data base. Secondly, in
eight of the nine earliest unambiguous examples (from Philippe de Thaon’s Com-
put, ca. 1113–1119), plus occurs in connection with a speech act verb (cf. (32)),
while the six ambiguous examples from that same text all have plus in connection
with support verb constructions with speech act nouns as the direct object, as in
(33). While syntactically different, these two types of constructions are semanti-
cally equivalent (i.e. to produce additional speech ≡ to continue to speak); and it
would thus have been easy for speakers to extend the use of plus from the s upport
verb construction to speech act verbs proper.
(32) N’e[n] voil ore plus parler, char ore vus voil nuncer Des signes dum parlai
Quant des jurz traitai Que Egyptïen truverent Qui mult sage gent erent,.
(Comput, p. 16, v. 1181, 1113 or 1119 – from BFM)
‘I don’t want to speak of that any longer, for now I want to tell you about
some signs that I talked about when I dealt with the days that the Egyptians,
who were very wise people, found,’
(33) Mais de ceste raisun Ne ferai plus sermon, Kar ore voil cumencer Altre dunt
voil traiter. (Comput, p. 28, v. 2034, 1113 or 1119 – from BFM)
‘But about this topic I will produce no more speech/I’ll speak no longer, For
now I want to begin something else that I wish to deal with.’
The data do not appear to yield any patterns which might explain why the
frequency of use of temporal/aspectual mais drops so dramatically after the 13th
c., eventually resulting in the complete disappearance of the marker in the course
of the 16th c. In the 13th c., where the most substantial number of occurrences
of both mais and plus are found, the two markers are comparable in terms of the
range of tenses they co-occur with. From the 14th c. onwards, mais becomes con-
fined to the simple finite indicative tenses. However, it continues to occur across
present, past and future contexts, so the reduced range may be attributable simply
Maj-Britt Mosegaard Hansen
to the considerably smaller number of tokens contained in the Middle French data.
The two markers likewise turn up in similar proportions in main and subordinate
clauses across the centuries. Finally, in the 13th c., they co-occur with a compa-
rable range of verbs, including the verbs être (‘to be’), avoir (‘to have’), modal verbs
followed by infinitives, support verb constructions, and full verbs belonging to a
variety of semantic classes. The subsequent reduction in the frequency of mais is
not accompanied by a reduction in the basic types of verbal construction it can
occur in. The possibility remains that dialect competition may have been a factor.
As stated above, however, I have found no indication in the literature of different
preferences for either plus or mais in different langue d’oïl dialects.
For the present, I can therefore offer only a speculative attempt at explaining
why mais disappeared as an n-word: In general, most of the erstwhile competitors
of the present-day set of n-words seem to have been eliminated in the course of
Middle French, for reasons that have yet to be fully elucidated. If, in this particular
competition, it was mais rather than plus that was lost, that may be linked to the
polysemous nature of these two markers. As we have seen, French plus is polyse-
mous between a temporal and a quantifying sense, which are closely related, and
which, as shown in Table 6, frequently overlap in Medieval French. Unlike plus,
however, mais does not appear to have been used as a non-temporal degree adverb
in Medieval French (cf. Section 3 above), so the only alternative use of the French
descendant of Latin MAGIS was as an adversative conjunction. The temporal and
the adversative senses of mais are not directly related to one another, but only
indirectly, via the use of MAGIS as a degree marker (cf. Note 13 above), and it may
therefore have been more difficult for language users to reconcile the two as differ-
ent senses of one and the same item than it is to reconcile the two senses of plus.
The adversative sense of mais being vastly more frequent than the temporal sense,
and given that temporal mais had a strong competitor with apparently identical
meaning in plus and was thus essentially redundant, I suggest that temporal mais
may have been lost largely for reasons of cognitive economy.18
Although Latin MAGIS and both Latin and French quantitative PLUS/plus were/
are polarity-neutral, the figures in Tables 5–6 above do not seem to support
. As pointed out by an anonymous referee, there was, of course, no necessity for tem-
poral mais to be lost, even in these circumstances. An alternative scenario might have seen
a division of labor developing between plus and mais, for instance along the lines of English
anymore vs any longer (cf. I don’t want to see you anymore/?any longer vs If you stay in bed any
longer/*anymore, you’ll miss the show).
The grammaticalization of negative indefinites
Table 7. Contexts of use of plus in the TFA (1150–1155) and in the TFA + BFM (12th c.)
Context type TFA alone (1150–1155) TFA + BFM (12th c.)
While the addition of the TFA data confirms that temporal plus could, indeed,
be used in weak negative polarity contexts at an early stage, it fails to show that
NPI uses were diachronically prior to the use of the marker in strong negative
polarity contexts. Indeed, the fact that no examples of temporal plus were found in
positive contexts, paired with the significantly higher proportion of occurrences in
strong negative contexts (particularly if tokens that are ambiguous between a tem-
poral and a quantitative interpretation are removed, in which case strong negative
contexts account for 92.9% of all the 12th c. examples), arguably supports the idea
that it is the strong negative use which is central, whereas the affective uses are
likely to be derived.
Closer analysis of the data reveals some further patterns suggesting that
Medieval temporal plus may have had an inherently negative semantics. Firstly,
the marker is found in preverbal position in negative contexts (i.e. preceding
the negative marker ne) already in the earliest texts, and as Table 8 below shows,
preposing accounts for a substantial portion of all the examples of plus in negative
contexts. Preposing is indicative of a negative semantics, as – in European lan-
guages, at least – NPIs with temporal adverbial functions overwhelmingly tend to
be ungrammatical when preceding their trigger (cf. Hoeksema 2000: 132f). Inher-
ently negative adverbials of this type have no such restrictions (although prepos-
ing may be stylistically marked), cf. the contrast between (34) and (35):
(34) *Any longer he couldn’t bear to keep it up.
(35) No longer could he bear to keep it up.
10.8% (7) 27% (52) 19% (11) 59% (44) 29% (45) 8% (8)
(37) Si s’en aloit contremont vers les nues, et maintenant ovroit li ciex por lui
recevoir, et il entroit enz sanz plus demorer. (Graal, p. 131, c. 1220 – from
BFM)
Thus he ascended towards the clouds, and now heaven opened up to receive
him, and he went inside without delaying any longer.’
(38) …parquoy sur tous autres le vous devroie desconseillier et, qui plus est,
deffendre de plus vous mectre en telz perilz ; (Saintré, p. 146, 1456 – from
BFM)
‘…wherefore above all others I ought to advise you against it and, what’s
more, forbid you to put yourself in such peril anymore/again;’
12.5% (1) 58% (30) 89% (8) 70% (31) 74% (28) 83% (5)
If tokens of plus following sans are counted as negative rather than affective, the
proportion of affective uses of plus starts to appear considerably less important
overall, as seen in Table 10:
Maj-Britt Mosegaard Hansen
Table 10. Negative vs affective uses of plus (sans classified as affective >
sans classified as negative)
Period/ 12th c. 13th c. 14th c. 15th c. 1501–1550 1551–1600
Context type
Negative 87.7% (57) > 79% (195) > 86% (58) > 63% (74) > 80% (155) > 93% (97) >
89.2% (58) 84% (207) 90% (60) 84% (99) 91% (175) 98% (101)
Affective 12.3% (8) > 21% (52) > 13% (9) > 37% (44) > 19.5% (38) > 5% (6) >
10.8% (7) 16% (40) 10% (7) 16% (19) 9% (18) 2% (2)
In the case of mais, the evidence is somewhat weaker. This marker only rarely
occurs in preverbal position in negative environments, namely in 5% (15) of the
13th c. examples, and in 13% (3) of those from the 15th c. In the 13th c., which is
the only period where a significant number of tokens are found in weak negative-
polarity contexts, 34% (22) of these are found in clauses governed by explicit or
implicit negation, while the remainder occur in a range of other types of affective
contexts, viz. conditionals, interrogatives, comparatives, temporal clauses intro-
duced by avant que (‘before’), and relative clauses following a superlative anteced-
ent. Interestingly, mais never occurs after the preposition sans in my data. Still,
the proportions of affective uses are relatively small (except for the 12th c., where
only two tokens of temporal mais were found, however) or, in several periods,
non-existent.
In any case, in terms of assessing the hypothesis of a quantifier cycle in French,
plus is the crucial case, as it is the only one of the two markers that has survived
to form part of the closed Modern French set of n-words. Like that of the other
temporal adverbial member of that set, jamais (cf. Hansen 2012), the evolution
of plus fails to provide support for the generalized cyclical development set out
in Table 3 above. For the same reason, the two temporal n-words likewise fail to
support Haspelmath’s (1997) hypothesis that where the same indefinite is used in
both negative contexts and weak negative-polarity contexts, the latter uses must be
prior to the former. Rather, the evidence presented here and in Hansen (2012) sug-
gests that, at least within the domain of negative polarity, indefinites can develop
in different directions, from NPIs into negative indefinites but also vice versa.
Whether Jäger (2010) is correct in also allowing for developments from negatively
polar items to positively polar or polarity-neutral ones, on the other hand, is not
an issue that can be addressed on the basis of the data reported in either this study
or Hansen (2012).
The fact that jamais and plus both have adverbial functions, while those
n-words that seem to conform to the quantifier cycle, i.e. personne, rien, and aucun,
are all nominals, might seem to suggest that some essential difference between
The grammaticalization of negative indefinites
adverbial and nominal items could be responsible for the different evolutions of
the markers, and that the assumption of an “argument cycle”, at least (as originally
positied in Ladusaw 1993), could thus be upheld. That is, however, immediately
undermined by the fact that the nominal n-word nul, which is also a component
part of the locative adverbial nulle part, the only one of the French n-words to have
an actual negative etymology, also developed NPI-uses during the transition from
Latin to French (cf. Buridant 2000: 135; Ingham 2011), as exemplified in (41):
(41) Et s’il y trouve nul ne son mestre ne autre, … que il le fera savoir au
preudome. (13th c. – cited in Ingham 2011: 445 – his (6))
‘And if he finds anyone there either his master or someone else, … that he
will let the gentleman know.’
Nor is there strong reason to suppose that similar directions of evolution found
with jamais and plus can somehow be attributed to their adverbial status, for, while
it appears that it was in both cases their use in negative contexts that gave rise
to their being extended to weak negative-polarity contexts, the two items differ
markedly in the extent to which the NPI uses became entrenched. As shown in
Table 2 above, these two adverbials are situated at opposite ends of the spectrum
in Modern French, with the nominal n-words in the middle. Possibly, this differ-
ence is due to the fact that plus was and is polysemous between a temporal sense,
nowadays found (almost) exclusively in negative contexts, and a quantifying sense
(predominantly) found in non-negative contexts, these two senses having thus
more or less divided up semantic space between them. Jamais, on the other hand,
has only ever been used with temporal meaning, and is thus in less need of a strong
contextual clue such as presence/absence of negation to its precise sense.
7. Conclusion
The evidence reviewed in this paper shows that the functional category, or para-
digm, of n-words in French is neither a synchronically nor a diachronically homo-
geneous one. On the contrary, individual members of the category exhibit very
different developmental trajectories and different distributional behaviors that
cannot simply be attributed to argument vs non-argument status.
It is clear that, even if there is no unidirectional quantifier cycle in French,
the n-words of Modern Standard French have nevertheless undergone a pro-
cess of grammaticalization in sense that the language has witnessed the gradual
emergence of a closed set of items which are obligatorily used to express quanti-
fier negation, and which are syntactically confined to contexts of (strong, and to
varying degrees, weak) negative polarity. In other words, paradigmaticization has
Maj-Britt Mosegaard Hansen
taken place (cf. Lehmann 1985). There is every reason to believe that language
users perceive the n-words as forming a functional category, and thus as shar-
ing certain crucial grammatical and semantic features. Indeed, the perception of
shared properties among different linguistic items is presumably a prerequisite for
paradigmaticization to occur in the first place.
It appears equally clear, however, that alongside this level of generalization,
speakers can and do maintain an awareness of idiosyncratic synchronic proper-
ties of individual members of the paradigm. Moreover, a given item can simul-
taneously be perceived as a member of different types of functional categories or
paradigms. Thus, temporal/aspectual plus is not only a member of the category
of n-words in contemporary French, but also belongs to the category of phasal
adverbs in that language, along with déjà, encore, toujours and enfin (Hansen
2008).
The present study thus provides further support for the idea, already presented
in Hansen (2008, 2012) that functional paradigms are not essentially linguistic,
but rather pragmatic, entities. Paradigmatic properties are those that are saliently
perceived to be shared between members in many, if not all, communicative con-
texts. They may come in overlapping clusters, such that there is no requirement
that all members possess the exact same set of properties, nor are they required to
possess shared properties to exactly the same degree.
The relationship between a paradigm and each of its members, and between
different paradigms sharing some of the same items as members, can thus be con-
ceptualized as a Gestalt structure, where either the paradigm or any invidual mem-
ber can become foregrounded at any given time. The principle can be illustrated
visually by the widely known image in Figure 1 below. This image can be perceived
either as a young woman who is turning her face away from view, or as an old
woman tucking her chin into her fur coat. What you interpret as the young wom-
an’s nose becomes a wart on the nose of the old woman, while the old woman’s
mouth becomes a ribbon around the young woman’s neck.
Just as one and the same feature plays different roles within the two differ-
ent images, so one and the same linguistic item can have different statuses within
different functional categories. What distinguishes plus within the category of
n-words is principally, on the one hand, its adverbial function, whereby it forms
a subgroup together with jamais and nulle part, contrasting with the subgroup
consisting of personne, rien, and aucun, which are characterized by having argu-
ment functions, and on the other hand, its aspectual meaning, whereby it stands
in contrast to the purely temporal marker jamais. Within the category of phasal
adverbs, on the other hand, plus stands out mainly because of its association with
negative polarity.
The grammaticalization of negative indefinites
Moreover, just as the viewer will focus on a single image at a time and p
otentially
on a specific feature of that image at the expense of the whole, so – I argue – the
language user when using a given functional item in a given context, will focus on
that item as a member of one particular category, and s/he may focus on either the
item itself or on the paradigm.
In either case, the focus on one Gestalt or feature of a Gestalt at the expense of
other aspects of the whole means that certain properties of either the paradigm or
the individual item are ignored. Where idiosyncratic properties of individual items
are frequently ignored because a paradigm is foregrounded, analogical change is
likely to take place, but paradigmatic pressure can equally well be resisted, as, on
the one hand, paradigms do not have inherently greater cognitive prominence
than their members and, on the other, they may be in competition with other
paradigms.
The consequence for grammatical descriptions of the system of negation in
Modern French (and presumably in many other languages as well) is a certain,
crucially unavoidable, heterogeneity that cannot easily be reduced to the operation
of a few simple principles.
Acknowledgements
I would like to thank the two anonymous referees for their valuable comments and
Daron Burrows (Oxford) for helpful discussion of a number of examples from my
data base. The usual disclaimers apply.
Maj-Britt Mosegaard Hansen
Data bases
References
Heine, Bernd. 2002. On the role of context in grammaticalization. In New Reflections on Gram-
maticalization [Typological Studies in Language 49], Ilse Wischer & Gabriele Diewald
(eds), 83–101. Amsterdam: John Benjamins.
Hoeksema, Jack. 2000. Negative polarity items: Triggering, scope, and c-command. In Negation
and Polarity. Syntactic and Semantic Perspectives, Laurence Horn & Yasuhiko Kato (eds),
115–146. Oxford: OUP.
Hoeksema, Jack. 2007. Parasitic licensing of negative polarity items. Journal of Comparative
Germanic Linguistics 10: 163–182. DOI: 10.1007/s10828-007-9012-y
Hoeksema, Jack. 2013. Polarity items in Strawsonian contexts – A comparison. In Beyond Any
and Ever. New Explorations in Negative Polarity Sensitivity [Trends in Linguistics. Studies
and Monographs 262], Eva Csipak, Regine Eckardt, Mingya Liu & Manfred Sailer (eds),
47–78. Berlin: Walter de Gruyter.
Ingham, Richard P. 2011. Grammar change in Anglo-Norman and Continental French: The
replacement of non-affirmative indefinite nul by aucun. Diachronica 28(4): 441–467. DOI:
10.1075/dia.28.4.01ing
Jäger, Agnes. 2010. Anything is nothing is something. On the diachrony of polarity types of indef-
inites. Natural Language and Linguistic Theory 28: 787–822. DOI: 10.1007/s11049-010
-9113-1
Jespersen, Otto. 1917. Negation in English and Other Languages [Det Kgl. Danske Videnskaber-
nes Selskab. Historisk-Filologiske Meddelelser I, 5]. Copenhagen: Høst og Søn.
Klein, Wolfgang. 1992. The present perfect puzzle. Language 68(3): 525–552. DOI: 10.2307
/415793
Klima, Edward S. 1964. Negation in English. In The Structure of Language. Readings in the Phi-
losophy of Language, Jerry A. Fodor & Jerrold A. Katz (eds), 246–323. Englewood Cliffs,
NJ: Prentice-Hall.
Ladusaw, William A. 1993. Negation, indefinites, and the Jespersen Cycle. In Proceedings of the
Nineteenth Annual Meeting of the Berkeley Linguistics Society.General Session and Parases-
sion on Semantic Typology and Semantic Universals, Joshua S. Guenter, Barbara A. Kaiser &
Cheryl C. Zoll (eds), 437–446. Berkeley CA: Berkeley Linguistics Society.
Laka Muzarga, Miren Itziar. 1990. Negation in Syntax: On the Nature of Functional Categories
and Projections. Ph.D. dissertation, MIT.
Lehmann, Christian. 1985. Grammaticalization. Synchronic variation and diachronic change.
Lingua e stile XX(3): 303–318.
Marchello-Nizia, Christiane. 1997. La langue française aux XIVe et XVe siècle. Paris: Nathan.
Martin, Robert. 1966. Le mot rien et ses concurrents en français (du XIVe siècle à l’époque contem-
poraine). Paris: Klincksieck.
Martineau, France. 1994. Movement of negative adverbs in French infinitival clauses. Journal of
French Language Studies 4: 55–73. DOI: 10.1017/S0959269500001976
Moignet, Gérard. 1976. Grammaire de l’ancien français. Paris: Klincksieck.
Muller, Claude. 1991. La négation en français. Geneva: Droz.
Offord, Malcolm. 1976. Negation in “Berinus”: A contribution to the study of negation in four-
teenth century French. Zeitschrift für romanische Philologie 92(3–4): 313–385. DOI: 10.1515
/zrph.1976.92.3-4.313
Payne, John. 1985. Negation. In Language Typology and Syntactic Description, Vol. 1, Timothy
Shopen (ed.), 197–242. Cambridge: CUP.
Rowlett, Paul. 1998. Sentential Negation in French. Oxford: OUP.
Maj-Britt Mosegaard Hansen
1. Introduction
For this reason, French appears to resemble languages such as those of the Slavic
family and medieval Germanic, having a series of indefinites limited to occurrence
in contexts featuring a negative particle, a characteristic of a negative concord
(NC) language. Such items are commonly referred to as ‘n-words’ in the literature
on the syntax of negation. As is well-known, in earlier stages of the language they
rien, nul and aucun appeared freely in a range of non-assertive contexts such as
interrogative and conditional clauses, as well as in negative contexts (Foulet 1930;
Richard Ingham & Amel Kallel
Roberts 2000).1 In keeping with their non-assertive polarity status, they are not
attested conveying a negative meaning in elliptical contexts. They were thus not
n-items and OFr was not an NC language. As an initial generalisation, then, it
would appear that a typological change has taken place in French with respect to
the distribution of indefinites. The issue addressed in this article is how this change
took place.
Haspelmath (1997: 197) distinguishes two main types of negative indefinites:
those that always co-occur with verbal negation (NV-NI), and those that never
co-occur with verbal negation (V-NI).2 In the first case, as in (1), the indefinite is
accompanied by a sentential negative element, whereas in the latter case the nega-
tive indefinite pronoun expresses the negative sense by itself, without needing an
additional verbal negation. Support for this distinction comes from the criterion
of negative meaning in elliptical contexts. Here, n-items will convey a negative
meaning, unlike non-assertive polarity items such as the English any-series, which
cannot be used in elliptical answers. French indefinite items such as rien, aucun,
and personne, which we shall identify as the personne series, do indeed convey a
negative meaning when used elliptically.
Both indefinite types are distinct from a third kind, indefinites such as English
any that occur commonly in other contexts than negation, designated by Ladusaw
(1980) as downward entailing: I’ll see if John reads any books entails I’ll see if John
reads novels; the term downward entailing refers to the fact that novel is a hyponym
of book. This class of indefinites will be referred to here as non-assertive polarity
items.3 In Modern French n-words cannot usually be employed in such contexts:
(2) Je vais voir… /*s’il se trouve nulle part/*si personne est derrière/*s’il y a
rien derrière/*s’il y a aucun problème/*si je m’étais aucunement trompé/*s’il
existe plus.
‘I’ll see… if he’s anywhere/if anyone is behind, if there is anything behind/if
there is any problem/if I had been mistaken in any way/if it exists any more’.
. Though one or two, e.g. personne, were post-medieval innovations as indefinites (Hansen
2013).
. A hybrid type is represented by Italian, where a negative indefinite sometimes co-occur
with verbal negation and sometimes does not (hence (N) V-NI), depending on its position in
relation to the verb.
. Haspelmath (1997: 34) similarly suggests the term ‘scale-reversal indefinites’, but in prac-
tice retains what he calls the ‘not particularly felicitous’ term negative polarity items for ele-
ments such as any. We concur with his judgment, if not his actual practice, and therefore do
not make use of the ‘NPI’ label.
Evidence from a correspondence corpus for diachronic change in French indefinites 1450–1715
Essentially, then, Modern French limits the personne series of indefinites to nega-
tive contexts, and can thus be called a NC language.4
The issue addressed in this article is the nature of the change in French such
that the personne series indefinites came to take on a negative semantic value, or in
other words how it became a NC language.
One possible answer is that French indefinites were driven by a common pro-
cess of change in the representation of negation into assuming a negative char-
acter; another is that they moved in this direction independently. The former
approach has been dubbed ‘macro-parametric’ by Martineau & Déprez (2004). A
‘micro-parametric’ approach, on the other hand, as pursued by these authors, and
Déprez (2011), investigates the changes in the syntactic behaviour and semantic
nature of individual lexical items themselves, and notably of indefinites. If one
wished to argue for a macro-parametric change, the development at the sentential
level that might be linked to this outcome would presumably be the grammati-
calisation of pas as an inherently negative marker, accompanying the OFr negator
ne. This is thought by Eckardt (2006) and Déprez (2011) to have occurred by the
end of the medieval period. The question of how a clause-level change might affect
indefinites would remain to be determined, however.
Early discussions of how a language could gain negative concord (Delbrück
1910; Jespersen 1917) invoked a morphological change whereby a negative ele-
ment was incorporated into a previously non-assertive indefinite, e.g. nihein in
Old High German, gheen in Dutch, and none in English. In medieval Germanic
these incorporated negative indefinites continued to be used with the sentential
negator, ni or ne.5 It is clear that in the case of French n-items, overt incorporation
did not occur. The indefinite items plus and nul have kept their morphological
form unchanged from earlier stages of the language, but have taken on a negative
character.
Minimalist studies of negation in syntactic theory consider the difference
between NC and NNC languages in terms of featural properties. According
to Rowlett (1998), the syntactic change in clause level negation was that pas,
rather than ne, became the semantically negative particle (i.e. the one bearing
. As pointed out by Hansen (2012), an exception here is the item jamais (‘ever’) routinely
used in negative contexts, but which at least in educated usage can appear in non-negative
contexts as well, e.g. Si jamais tu le vois…(‘If you ever see him…’), A-t-on jamais vu un cas
pareil ? (‘Have you ever seen such a case?’)
. Ingham (2006) and Kallel (2007, 2011) have studied how at a later point English changed
from a NC language to a non-NC language, following the loss of the sentential negator ne in
the 14th century.
Richard Ingham & Amel Kallel
. However, we do not adopt the syntactic approach of Déprez (2011), which appeals to a
change in the syntactic position of the indefinite within the phrase structure of nominals. As
pointed out by Hansen (2011, 2012) this approach faces difficulties when the evolution of
non-nominal indefinites, such as plus or jamais, is considered.
Evidence from a correspondence corpus for diachronic change in French indefinites 1450–1715
(7)
direct
(4) (6) negation
question indirect
negation
(1) (2) (3)
specific specific irrealis
known unknown non-specific
(5) (8)
conditional comparative
(9)
free choice
. Hansen (2011) notes that even in contemporary French this usage is not entirely confined
to high register texts, however.
Richard Ingham & Amel Kallel
present article will based. Section 4 details the analysis used in data drawn from
this source. Sections 5 and 6 present and discuss the results obtained.
But already in OFr. they could combine in a single item with a free-choice mean-
ing, e.g.:
(4) a. Car mes sires m’en ocirra en quel que leu qu’il me truist.
La queste del Saint Graal, 1220
‘For my lord will kill me in whatever place he might find me’
In Middle French quelque began to be used as a non-specific indefinite, as in:
b. Et vous, mere Dieu debonnaire,/Jettez me hors de ceste haire/Par
quelque tour. Miracle de l’enfant ressuscité, 1353
‘And you, kind Mother of God, release me from this hair-shirt in some
way or other’
This use was common in a non-assertive context such as a conditional clause:
(5) Se auchunefois en quelque besongne desirons aide faire a cheulx qui
labeurent, il le convient faire par si grand humilité.
Jean Daudin, De la érudition ou enseignement
des enfans nobles, c.1360–1380, 277
‘If we want in any way to help those who work; it should be done with such
great humility’
A further use as a specific indefinite also developed in later Middle French, accord-
ing to Combetttes (2004), e.g.:
(6) Le roy eut quelque amy qui l’ en advertit. Commynes, Mémoires
‘The King had a certain friend who told him about it’
It should also be noted that in Middle Fr. quelque began to appear in negative
clauses:
(7) Comme dist Boece: «Il n’est quelque bien qui ne reluise plus plaisamment
quant il est approuvé par la congnoissance de plusieurs». Jean Daudin, 1360
‘As Boetius said: There is no good thing that does not shine more
attractively when it is approved by other people’s awareness’
Such instances appear to be cases of the emphatic free-choice use. As quelque lost
this sense, it seems to have lost the ability to appear in negative contexts.
To date, research into the diachrony of French indefinites has left a number
of issues without satisfactory resolution. In particular, it is not clear why, or when,
aucun acquired its status as a negative item, or how this comported with the rise
of the indefinite quelque. Déprez’s (2011) account of semantic change in French
indefinites is oriented exclusively towards nominal expressions. The steps taken by
aucun as part of that process, and the extent to which the change is consistent with
theoretical positions taken in the previous literature remain unclear, in particular
whether they accord with the rather precise predictions of the typological model
adopted by Haspelmath (1997); our aim in this study was to clarify these issues.
Richard Ingham & Amel Kallel
The data examined in this study were provided by private correspondence between
the later 15th and early 18th centuries. It is widely accepted as the written genre
nearest to informal spoken language where changes typically spread. Private letters
have often been found to be valuable sources of information available to diachronic
researchers. Linguistically, such letters occupy a special position as a record of the
speech and writing habits of a certain speech community. They may reflect less
concern with formal correctness and high style pre-occupations than do literary
genres of the relevant period, especially, as is indicated by the following evaluation
of one particular letter-writer of the 16th century, Marguerite D’Angoulême:
(8) ‘Marguerite écrivait très vite et très étourdiment, elle saute des mots, elle en
estropie, elle a deux ou trois manières d’écrire le même mot ; tantôt elle suit
les règles grammaticales d’alors, tantôt elle note simplement la prononciation,
qui a partir de cette époque surtout, se séparait de l’orthographe’ (Lettres de
Marguerite D’Angoulême, sœur de François Ier, Reine de Navarre).
The correspondence collections used for the present research were as follows,
divided into three periods, the first representing Middle French of the 15th cen-
tury, the second the Renaissance period of the mid-16th century, and the third the
Classical period beginning 1650, up to 1715 (full references for the words indi-
cated will be found in the bibliography):
Period I (1450–1499)
La Guerre du bien public; Lettres de Louis XI; Lettres de Charles VIII
Period II (1530–1575)
Lettres D’Antoine De Bourbon et de Jehanne D’Albret; Lettres de Marguerite
D’Angoulême ; Lettres de Blaise de Monluc
Period III (1650–1715)
Correspondance du Maréchal de Vivonne; Lettres du Duc de Bourgogne;
Lettres du Cardinal Mazarin a la Princesse Palatine 1651–1652
Unlike the Corpus of Early English Correspondence (Nurmi 1998) where from
the 15th century onwards one finds private correspondence that involves family
members and discusses everyday issues (Kallel 2011), the great majority of the
French correspondence collections used here were written (or more likely dic-
tated) by aristocratic or royal personages in high office, and might therefore be
seen as having been written in a formal public capacity, rather than representing
the informal register for which the use of private correspondence is held desirable
in diachronic linguistics. However, it should not be assumed that the royal or noble
provenance of a letter entailed an especially formal style. The correspondence style
Evidence from a correspondence corpus for diachronic change in French indefinites 1450–1715
of Louis XI, for example, was rather plain and direct. That said, there were many
letters from noble or royal correspondents whose stylistic choices would certainly
have differed very considerably from those of ordinary spoken communication.
We have tried to mitigate this problem by taking a selective approach to the avail-
able correspondence material, especially by excluding letters written to more than
one addressee, which tended to have a more official status. Likewise, among royal
letters addressed to individuals, those beginning ‘De par le Roy’ were excluded, as
inconsistent with being written in a private capacity.
Previous studies of the history of French, reviewed above, mention that quelque
came to replace aucun in positive and in most non-assertive contexts, but it
remains unclear when aucun took on its largely negative character, in other
words when it was decisively ousted from positive and non-assertive contexts.
Martineau and Déprez (2004) considered that this development occurred in
the Renaissance period but left the matter open for further research. At issue,
furthermore, is the aetiology of the change, whether quelque came to occupy
semantic spaces left empty by the retreat of aucun, or whether aucun was pushed
out of such spaces by the intrusion of quelque. This question may be compared to
one of a ‘pull’ chain versus a ‘push’ chain, in terms of phonological change. Did
aucun adopt a more clearly negative value, for a reason to be determined, and in
so doing require that another indefinite assume its former place in the positive
functions of indefinites identified by Haspelmath, or did quelque compete for its
non-negative functions?
The data categorisation procedure intended to address these issues was as fol-
lows: all instances were recorded of nul, aucun, and quelque occurring in one of
the following contexts:
. In practice, no interrogative clauses were identified in the correspondence data sources.
Richard Ingham & Amel Kallel
Table 1 below, showing data from Period I, indicates that all three indefinite
items occurred in all three different contexts, except for nul, absent from positive
contexts, as expected (Ingham 2011). It will be seen that aucun in this period con-
stituted the most commonly used indefinite in all three contexts, with 76% of the
total of possible occurrences, 73% of the total of non-assertive contexts, and 59%
of the total of negative contexts. Quelque also occupied all three contexts, but in
smaller percentages compared to aucun.
Table 1a. The frequency of ‘nul’, ‘aucun’ and ‘quelque’ in positive, non-assertive
and negative contexts in Period I
PERIOD I Positive contexts N/A contexts Negative contexts Total
Nul 0 12 59 71
0% 17% 83%
Aucun 171 77 92 340
50% 23% 27%
Quelque 54 17 6 77
70% 22% 8%
Evidence from a correspondence corpus for diachronic change in French indefinites 1450–1715
Table 2a (below) depicts the individual development of nul, aucun, and quelque
in Period II, in which all three items continued to evolve in the same direction as
in the previous period, shown in Table 2a. Nul mostly occupied negative contexts,
as before; aucun was acquiring more preponderantly negative associations, while
quelque now stood in positive contexts in 85% of its uses.
Table 2a. The frequency of ‘nul ’, ‘aucun’ and ‘quelque’ in positive, non-assertive
and negative contexts in Period II
PERIOD II Positive contexts N/A contexts Negative contexts Total
Nul 0 18 72 90
0% 20% 80%
Aucun 39 19 95 153
26% 12% 62%
Quelque 267 39 9 315
85% 12% 8%
By Period III in our data, the identity of each of nul, aucun and quelque has
been transformed, and the latter two have essentially completed the changes that
Richard Ingham & Amel Kallel
emerged so conspicuously in the 16th century data. Quelque is now the preferred
indefinite in both positive and non-assertive contexts, with 100% and 75% of the
total respectively. As regards its semantic identity, aucun now stands in negative
contexts 89% of the time, i.e. almost as often as nul. Its overwhelmingly negative
character was now decisively established (Table 3a):
Nul 0 3 32 35
0% 9% 91%
Aucun 0 14 110 124
0% 11% 89%
Quelque 394 43 27 464
85% 9% 6%
6. Discussion
The single main development to which the findings of this research attest is the
dramatic rise in the frequency of the indefinite quelque, which by the Classical
period accounted for nearly two-thirds of all data points. Its establishment in the
system of indefinites occurred by the latest in the middle third of the 16th century,
on the basis of the correspondence data surveyed. At this time it was the single
most frequent form not only in positive but also in non-assertive contexts. On the
other hand, it never succeeded in becoming a viable indefinite form in negative
clauses despite some initial gains here. Correspondingly, the share of aucun in
Evidence from a correspondence corpus for diachronic change in French indefinites 1450–1715
positive contexts slumped, to under 15% in the mid-16th century, and to nil in the
later 17th century.10
There is evidence in these correspondence data of a rather sudden shift
between the later 15th century and the 16th century as regards the use of indefi-
nite expressions. It appears to have followed the semantic bleaching of quelque into
an ordinary (non-emphatic) indefinite. The rise in uses of quelque with specific
indefinites only became possible when this earlier association was lost. The data
confirm the trajectory of quelque from a free choice to an irrealis, and then to a
realis indefinite, expected from the research of Combettes (2004). Up to 1480, it
was entirely restricted to irrealis instances, e.g.:
(19) Vous devez essayer de faire quelque eschec sur eulx.
Lettres de Louis XI 3, 187 (1467)
‘You should try to place them in some difficulty’
The first case where it was used for a specific (realis) referent appeared in the last
quarter of the 15th century:
(20) Et, pour ce qu’il y a quelque different, nous escripvons a nostre tres chier
et grant amy… Lettres de Louis XI, 9, 236 (1482)
‘And because there is some dispute, we write to our very dear great friend’
This is clearly a realis use of quelque, since the writer could not have denied the
existence of the dispute in question.
Quelque had already established itself very securely in non-assertive contexts:
uses in conditional clauses are already very common by Period I and indirect
negation uses are also found, e.g.:
(21) ( = 12a) Affin qu’ilz n’ayent cause de vouloir faire quelque rompture
en ce qui a esté appoincté Lettres de Louis XI 6, 13 (1475)
But the 15th century correspondence still shows aucun to have been the indefinite
overwhelmingly used in positive contexts, as in Old French. This included a pref-
erence for aucun over nul in non-assertive and in negative contexts, where nul had
been the preferred Old French form, and is exemplified in:
(22) Se vous avez aucune chose touchant le conté de Valentinoiz….
envoiez le moy. Louis XI 9, 312 (1482)
‘If you have anything concerning the county of V…. send it to me.’
. In period II as aucun was acquring a negative character, it appeared in plural form in
only a small minority of contexts, barely a dozen altogether. This does not support the asser-
tion of Brunot (1906: 431) that the early development of aucun into a negative item occurred
in plurals. We are grateful to an anonymous reviewer for raising this issue.
Richard Ingham & Amel Kallel
Although very occasional uses of les aucuns, and d’aucuns were noted (cf. Prévost &
Schnedecker 2003), in line with the development noted by these authors of aucun
as a post-determiner item, we would like to suggest that the key development
involved in the restructuring of the domain of indefinites was not a change in
the clausal syntax of negation, but a reorganization in the functions of individ-
ual indefinites. The process gained momentum without any indication as yet that
French had become a NC language in syntactic terms. In the mid-15th century
aucun clearly retained its non-assertive as well as negative uses, a fact incompat-
ible with a change to the status of an n-item required if French had become a NC
language. Likewise, nul continued to be used in the 16th century in non-assertive
contexts.
Let us return to the issue of the distribution at this time of indefinite forms
in terms of the Haspelmath (1997) semantic map. In the 16th century, aucun
remained relatively common in specific indefinite contexts, as also in negative
clauses. In other words, it continued as a productive form only at the periphery
of the territory it had occupied in the 15th century. However, quelque occupied
contiguous zones on the semantic map, nos. 3 and 5 at first, then nos. 2, 3 and 5.
Its uses far outnumbered those of aucun in these positions, even though the latter
form remained possible there. The discontinuous nature of the two areas where
aucun remained a preponderant form in the 16th century seems to have played a
key role. On the basis that non-contiguous distribution of functions is impermis-
sible, aucun could not continue occupying only functions nos. 1, 2, 6, 7 and 8,
having given up the intermediate functions to quelque. Hence the semantic map
of French indefinites was gradually re-drawn, so that one of the peripheral zones
hitherto occupied by aucun, nos. 1–2, was eliminated. This was already subject to
competition from quelque, whereas zone 8 (direct negation) experienced no real
challenge from its competitor form.
Despite its ability to appear in all contexts, quelque was already gaining a posi-
tive identity in the late 15th century, when it was used as a non-specific indefinite
in positive contexts in the majority of its uses. However, aucun at this stage was
also identified more with positive than with other contexts, so on the basis of these
data, no acquisitional shift to a negative identity would have been warranted at this
stage. The two forms were in competition for the left-hand side of the H aspelmath
semantic map. The picture changed markedly in the mid-16th century. Now a
clear majority of the uses of aucun occurred in a negative context. Speaker varia-
tion, in addition, would no doubt have meant that in everyday use some children
Evidence from a correspondence corpus for diachronic change in French indefinites 1450–1715
acquiring language at this time would have heard it in negative contexts an even
higher proportion of the time than the correspondence data suggest. In this way,
we can suppose that a negative feature would have started to be attributed to aucun
among some speakers at least. As one generation succeeded another, the process
would have become self-reinforcing, leading to the situation observed here where
over 90% of occurrences of aucun by the later 17th century were negative. At this
point, it is highly likely that input evidence in everyday language would have mas-
sively favoured the attribution of a negative feature to aucun, the position reached
in contemporary French.
The correspondence data analysed here have thus allowed us to trace the tra-
jectory taken by the two key items in the changing landscape of French indefinites,
in their development towards their values in modern French. It has been possible
to do this in terms of the properties of the items themselves, rather than appealing
to changes in the sentence-level representation of negation in French. The prin-
cipal factor in the change is seen to have been the intrusion of quelque into the
system of indefinites: if this had not taken place, it is a moot point as to whether
aucun would have undergone the semantic shift that it did prior to the contempo-
rary period.
7. Conclusion
. This need not be associated with a UG theory of grammar acquisition: a language learner
bias against an item having discontinuous functions might well be seen as cognitively given,
an instance of the avoidance of a discontinuous construct that is already shown in children’s
prelinguistic cognition (Spelke 1990).
Evidence from a correspondence corpus for diachronic change in French indefinites 1450–1715
before recent centuries, given the nature of the textual record. In a single-genre
corpus, this problem is addressed, and the linguistic changes that emerge can be
taken as real, not artefacts of a confound between genre and time-period.
Nevertheless, some limitations of the study need to be noted. First, the sam-
pling is not claimed to be equally representative of every part of the time-span
1450–1715. It was done in a way that reflected the ready availability of data sources,
in particular the series of correspondence collections in the series first identified
as the basis of the corpus, publications of the Société de l’Histoire de France. The
sampling using these sources thus did not cover the whole time span 1450–1715
uninterruptedly, but contained certain gaps, notably between 1575 and 1650,
which remain to be filled. It is possible that important developments occurred
during this time span, which were not picked up by the sampling adopted. In prac-
tice we do not think that this issue invalidates the finding in the present research
of the directions of change in the distribution of particular items, though it is quite
possible that further research, notably into the pre-Classical period, may establish
an earlier timing for them. Secondly, the single genre corpus on which this study
draws clearly belongs to a very different register from that of everyday conversa-
tion, the medium in which language change is generally supposed to take place.
This limitation, however, affects all data sources from before the contemporary
period: the question is whether another written genre would offer a better approxi-
mation to the use of indefinite in everyday speech in the period concerned. Prose
comedies would be a promising further source of investigation, which the authors
are currently examining.
Primary sources
Lettres du Duc de Bourgogne au Roi d’Espagne Philippe V et à la Reine 1709–1712. Tome Premier.
Alfred Baudrillart et Léon Lecestre (eds.), 1912.
Lettres du Cardinal Mazarin à la Princesse Palatine 1651–1652. M. Ravenel (ed.) Paris: Crapelet,
1836.
References
Brunot, Ferdinand. 1906. Histoire de la langue française, Tome II: Le seizième siècle. Paris:
Armand Colin.
Combettes, Bernard. 2004. La grammaticalisation d’un déterminant indéfini: Quelque en Moyen
Français. Scolia 18: 9–40.
Delbrück, Berthold. 1910. Germanische Syntax: Zu den negativen Sätzen. Leipzig: Teubner.
Déprez, Viviane. 2011. Atoms of negation. In The Evolution of Negation: Beyond the Jespersen
Cycle, Pierre Larrivée & Richard Ingham (eds). Berlin: Mouton de Gruyter. DOI: 10.1515
/9783110238617.221
Eckardt, Regine. 2006. Meaning Change and Grammaticalisation. An Enquiry into Semantic
Reanalysis. Oxford: OUP.
Foulet, Lucien. 1930. Syntaxe de l’ancien français. Paris: Champion.
Hansen, Maj-Britt. 2011. Discussion of ‘Viviane Déprez: Atoms of negation. An outside-in
micro-parametric approach to negative concord’. In The Evolution of Negation. Beyond
the Jespersen Cycle, Pierre Larrivée & Richard Ingham (eds), 273–283. Berlin: De Gruyter
Mouton. DOI: 10.1515/9783110238617.273
Hansen, Maj-Britt. 2012. On the evolution of temporal n-words in Medieval French. Language
Sciences 34(1): 76–91. DOI: 10.1016/j.langsci.2011.06.003
Hansen, Maj-Britt. 2013. Negation in the history of French. In The History of Negation in the
Languages of Europe and the Mediterranean [Oxford Studies in Diachronic and Historical
Linguistics 5], David Willis, Christopher Lucas & Anne Breithbart (eds). Oxford: OUP.
Haspelmath, Martin. 1997. Indefinite Pronouns. Oxford: OUP.
Ingham, Richard P. 2006a. On two negative concord dialects in early English. Language Varia-
tion and Change 18: 241–266. DOI: 10.1017/S0954394506060121
Ingham, Richard P. 2011. Grammar change in Anglo-Norman and continental French: The
replacement of non-affirmative indefinite nul by aucun. Diachronica 28(4): 441–467. DOI:
10.1075/dia.28.4.01ing
Jespersen, Otto. 1917. Negation in English and Other Languages. Copenhagen: A.F. Horst & Son.
Kallel, Amel. 2007. The loss of negative concord in Standard English: Internal factors. Language
Variation and Change 19: 27–49. DOI: 10.1017/S0954394507070019
Kallel, Amel. 2011. The Loss of Negative Concord in Standard English: A Case of Lexical Reanaly-
sis. Newcastle upon Tyne: Cambridge Scholars.
Ladusaw, William. 1992. Expressing negation. In SALT II Proceedings, C. Barker & D. Dowty
(eds), 237–260. Ithaca, NY: Cornell Linguistic Circle.
Larrivée, Pierre. 2004. L’Association négative: Depuis la syntaxe jusqu’à l’interprétation. Geneva:
Droz.
Lightfoot, David. 2006. How New Languages Emerge. Cambridge: CUP. DOI: 10.1017/CBO97805
11616204
Evidence from a correspondence corpus for diachronic change in French indefinites 1450–1715
Martineau, France & Déprez, Viviane. 2004. Pas rien/Pas aucun en Français classique: Variation
dialectale et historique. Langue Française 143: 33–47. DOI: 10.3917/lf.143.0033
Muller, Claude. 1991. La négation en français: Syntaxe, sémantique et éléments de comparaison
avec les autres langues romanes. Geneva: Droz.
Nurmi, Arja (ed.). 1998. Manual for the Corpus of Early English Correspondence Sampler.
Helsinki: Department of English, University of Helsinki.
Prévost, Sophie & Schnedecker, Catherine. 2003. Aucun(e)(s)/d’aucun(e)(s)/les aucun(e)(s):
Évolution du français médiéval au français moderne. Scolia 8: 39–73.
Roberts, Ian. 2000. Some remarks on the diachrony of French negation. DELTA 16: 201–219.
DOI: 10.1590/S0102-44502000000300008
Rowlett, Paul. 1998. Sentential Negation in French. Oxford: Oxford University Press.
Spelke, Elizabeth. 1990. Principles of object perception. Cognitive Science 14: 29–56. DOI: 10.1207
/s15516709cog1401_3
Willis, Daniel. 2011. Negative polarity and the quantifier cycle: Comparative diachronic per-
spectives from European languages. In The Evolution of Negation. Beyond the Jespersen
Cycle, Pierre Larrivée & Richard Ingham (eds), 285–323. Berlin: Mouton. DOI: 10.1515
/9783110238617.285
Zeijlstra, Hedde. 2008. Negative concord is syntactic agreement. LingBuzz 000645.
The continuity of the vernacular
The evolution of negative doubling in French*
Pierre Larrivée
Normandie Université, UCBN and CRISCO (EA4255)
Labov’s idea that the vernacular is the most stable variety of a language raises
questions especially where languages of wider communication are concerned.
Whether the vernacular practices of a language’s geographical varieties are
convergent synchronically and historically can be established by looking at
particular variables. One such variable is investigated in this paper on the
co-occurrence of a clausal negator with a n-word (e.g. I didn’t do nothing, i.e.
anything). The quantitative study of negative doubling in Quebec and France
historical and contemporary vernacular sources demonstrates overall stability for
France from the 14th century to the present. That negative doubling is ten-fold
more frequent in contemporary vernacular Quebec French than in contemporary
vernacular France French may be dues to difficulty in accessing the vernacular
in France French and in the presumed lesser impact of normative pressures in
Quebec. Tracking one variable over six centuries provides evidence that stability
characterises vernacular varieties.
1. Introduction
The purpose of this paper is to assess the idea from Labov (1966) that the vernacu-
lar is the most stable variety of a language. In the case of languages with a norma-
tive tradition, the vernacular defined as the spontaneous, non-monitored variety
of a language remains the least accessible variety, and especially in a historical per-
spective. Such a variety may be more immediately accessible in so-called periph-
eral varieties that may not be subject to normative pressures to the same degree as
the central variety is. While features of regional English, French and German are
* This work was presented at the Across the line of Speech and Writing Variation meeting in
November 2011 at Louvain-la-Neuve. It benefitted from the kind support of Anthony Lodge,
Richard Ingham and France Martineau, and from the extensive and thorough comments of
one indefatigable reviewer. The usual disclaimers apply.
Pierre Larrivée
often treated as merely dialectal, these may simply relate to vernacular practice,
as has been shown for e.g. North-of-England English (Anderwald 2005), Quebec
French (Larrivée 2004) and Bavarian German (Weiss 1998). The zero hypothesis
by virtue of the inertia principle (Longobardi 2001) would then be that these ver-
nacular features would equally be found in other geographical varieties includ-
ing the central one,1 and at convergent rates. Convergence of vernacular variables
across varieties of a language of wider communication would provide support for
Labov’s conception of the vernacular as a more stable variety.
The assessment of the idea of the stability of the vernacular is conducted
by looking at New and Old world varieties of French. The selected grammatical
feature is the co-occurrence of the main clausal negator pas ‘not’ (and competi-
tor point for earlier periods) with a n-word such as aucun ‘no(ne)’, rien ‘nothing’
and jamais ‘never’.2 Such negative doubling is selected because while it is associ-
ated with Quebec French, it is also found in European French vernacular variet-
ies, although intuitively at much lower rates. Assuming that the same underlying
grammar is at play,3 this raises the question of the potential divergence in the rates
of use, and whether a convergence could be observed historically. Convergence in
the rates of negative doubling in earlier periods of French and in one contempo-
rary vernacular variety would support the idea that there is continuity in the ver-
nacular varieties of languages of wider communication. One objective is therefore
. Clearly, the extent of the convergence between the vernacular in different regions is an
empirical question. In other words, precise studies are needed before it can be asserted that
there is one or several vernaculars in languages of wider communication such as French.
. Clausal negators in this context are post-verbal adverbs that immediately follow the conju-
gated verb; n-words convey negation (they occur in fragment answers, with constituent scope,
and can communicate double negation readings, Giannakidou 2007; Larrivée 2011), and they
refer to an ontological category such as animate, time, place and so on (Haspelmath 1997).
The joint use of clausal negator and n-word is called doubling, as opposed to the negative
spreading of jointly used n-words in the widely accepted terminology of den Besten (1986).
While strictly speaking use of a n-word with ne would correspond to doubling, given that ne
is a marker of normative practice since at least the 18th century when it starts being dropped
(Martineau 2011; for earlier French ne-drop, see Ingham 2011), it cannot meaningfully be used
to track the continuity of the vernacular.
. As pointed out by the reviewer, there is an assumption that pas and n-words are in a
process of change up until the 19th century (Martineau 2011, Martineau & Déprez 2004). If
this were the case, there would obviously be little use of conducting a historical comparison
of configuration that have different grammatical statuses. Fortunately, observations from non-
literary texts for the relevant items evidence a high preponderance of uses in negative contexts
as early as the 14th century (Larrivée 2011). These terms are therefore comparable within the
vernacular over the period considered here.
The continuity of the vernacular
The question of whether one should bother to study the vernacular is not ille-
gitimate, in view of the considerable difficulties of obtaining data of spontaneous
interactions that reveal the competence of speakers that would not be modulated
by normative practices and face-saving considerations. These challenges are well-
worth facing however in view of the benefits of having access to a systematic, pri-
mary and stable linguistic practice.
The vernacular is presumed to be a more systematic language variety. The sys-
tematic nature of the vernacular can be argued by pointing at unexpected produc-
tions in normative practices that are allegedly not found in the vernacular. The form
of relative pronouns in normative French is expected to reflect the function it has
in the dependent clause. Vernacular French however tends to resort essentially to
qui as the subject wh- and que for all complement wh-words (Blanche-Benveniste
2010). Complex markers such as dont ‘of-wh’ for complements introduced by de
are therefore rare (Blanche-Benveniste 2010), and given to productions that do
not correspond to syntactic expectations. A sequence such as “Dans cette ville,
il y a 10 règles dont toute personne doit respecter et savoir” “In this town, there
are 10 rules of which everybody must respect and know” (Berrendonner 1998: 91;
see also 2004) illustrates the notion of hypercorrection. Not documented for the
vernacular, such unexpected productions in normative practices are thought to
be due to the misapplication of normative rules because these are mastered less
systematically as they are acquired later than those of the vernacular.
The acquisitional status of the vernacular as the initial input of language learn-
ing points to its primary character. The vernacular is the first variety in which native
speakers are normally socialised. That is why a corpus of interactions between chil-
dren and adults that are unlikely to be very affected by normative c onsiderations
Pierre Larrivée
These illustrations make it clear that doubling cuts as much across languages as
within them. It does seem in fact that very few of the languages rejecting dou-
bling do not have some variety that allows it. The 1944 languages on the WALS
map for feature 115A (Negative Indefinite Pronouns and Predicate Negation)
comprises eleven languages disallowing predicate negation with n-words.
These include two Indo-European languages, Dutch and German, which do
have varieties with negative doubling such as West Flemish (a.o. Haegeman &
Lohndal 2010) and Bavarian ((6) above, Weiss 1998). English, Icelandic, Swed-
ish, French, Italian, Spanish, Portuguese are among the thirteen languages that
have a mixed behaviour. Mixed behaviour covers a number of scenarios, and
the story remains that apparently no one language variety is governed by a uni-
form rule. Divergences are common in the behaviour of items: the clausal nega-
tors of normative French do not pattern together, and while pas ‘not’ is not
thought to happily join with n-words, no such felicity limitations apply to plus
‘no more’, which has otherwise the same distribution as pas (placement after the
conjugated verb, impossibility to express a fragment answer on its own). The
same situation is shown to hold in French-based Creoles, where some items
require co-occurrence with a clausal negator that is allowed by others and that
yet others refuse (Déprez 2011). This argues in favour of a model of concord
based on the internal structure of items rather than a rule applying uniformly
to a language (variety).
The varying patterns of negative doubling across items, language varieties
and languages are reflected by the contrasted views expressed in studies on the
situation in French. Many formal syntactic studies claim that French categori-
cally excludes negative doubling configuration such as Il n’a pas rien dit ‘He ne
. Discounting the twelve languages with negative existential constructions, that seem or-
thogonal to our concerns.
Pierre Larrivée
has not said nothing’ (Biberauer & Robert 2011; Corblin & Tovena 2001; Déprez
2000: 261; Giannakidou 2000: 350; Muller 1984: 64–65; Rowlett 1998: 143ss;
de Swart 2010: 156). Many functional and descriptive publications signal its exis-
tence in spoken France French (Bauche 1946; Frei 1982[1929]: 59; Harris 1978;
Gadet 1992; Hansen 2011) and in different regional varieties (Quebec French
and Parisian French in particular; see Gaatone 1971: 174; Larrivée 2004: 146ss,
152; Muller 1991: 260–263, 279, 322–324; Rowlett 1998: 132, 166ff; de Swart
2010: 201ff; Zeijlstra 2009). The existence of negative doubling in the history
of French is provided quantitative documentation by Martineau and Déprez
(2004: 35–36, and references therein).
These contradictory claims should be easy to apportion on the basis of
attested data in corpora of the vernacular. Such an assessment has not been
proposed to my knowledge, and the reasons for this state of affair might be that
corpora of vernacular French are surprisingly few and far between. The early cor-
pora of vernacular French (the semi-directed exchanges of the Montreal series
going back to Sankoff-Cedergren 1971; Montreal 1984 and Montreal 1995; the
similar format of the Ottawa-Hull corpus collected in a region where French is
in heavy contact with Canadian English and directed by Shana Poplack; and the
heterogeneous France French Corpus de Référence du français parlé also known
as Corpaix led in the late 1980’s by Claire Blanche-Benveniste) are not easily
accessible to researchers. The semi-directed discussions recorded from the late
1980’s, transcribed and made partially available by the Belgian team V alibel
seem slightly formal in tone. Another Belgian team, Elicop, makes available
classroom role-plays and the semi-directed discussions gathered in Orléans in
the 1960’s by British sociolinguists, the endeavour h aving been reproduced in
different corners of France in the 1980’s by Kate B eeching. Semi-directed nar-
rations for Paris French had to wait the 2000’s Corpus de français parlé parisien
(CFPP) initiative led by Sonia Branca, and 2012 for the OFROM corpus of
Swiss French narrations proposed by Mathieu Avanzi and Marie-José Béguelin.
Unmonitored spontaneous discussions in French are only made accessible in
the Corpus de français parlé au Québec (CFPQ) under the leadership of Gaétane
Dostie. Though still far from satisfactory, the situation nonetheless allows the
existence and extent of negative doubling in Quebec and France French to
be assessed.
Quebec French was always known to allow negative doubling widely (Léard
1995; Lemieux 1985), but what remains unknown is to what extent. It is impor-
tant to establish the relative weight of negative doubling so as to allow compari-
son between varieties. This is established by determining the number of relevant
occurrences of pas with clause-mate n-word aucun ‘no(ne)’, personne ‘nobody’,
rien ‘nothing’, jamais ‘never’ and nulle part ‘nowhere’, and calculate the percentage
The continuity of the vernacular
of these both against the total number of pas5 and the total number of n-words.
Relevant occurrences are exclusive of non-negative uses of the markers (expletive,
exceptive, interronegative uses), cases of n-words coordinated to a negative clause
as in the equivalent of He didn’t speak, or nothing, and non-clause-mate uses in
dependent infinitives and subordinates – as in I don’t want to say nothing of the
sort, I don’t think he said nothing). As a comparator, we further determined the
number of clause-mate n-words with clausal negator plus ‘no more’, which unlike
pas is not stigmatised in this configuration, to control for any normative effects.
The manual examination6 of each co-occurrence returned by the CFPQ search
engine provides the rates presented in Table 1.
Table 1. Rates of negative doubling in contemporary vernacular Quebec French
aucun personne rien jamais nulle Totals % of ND to
part total negators
. The reviewer’s opinion is that comparison of cases of negative doubling against the total
number of pas is not useful because not all contexts with pas allow n-words and therefore
negative doubling: for instance, no n-word could be used as a direct object of an intransi-
tive verb, and it is claimed referring to Martineau and Déprez (2004) that n-words are not
found as subjects of a clause negated by pas. Whether the latter is true is clearly an empirical
question; as to the former, adverbial phrases containing jamais or other n-words could be
found with intransitives, and the rate of intransitives is presumably going to be comparable
across corpora. Nonetheless, I have provided another measure of negative doubling against
the number of n-words.
. Made necessary by the polyfunctionality of many items, plus in particular for which the
negator reading ‘no more’ has the same spelling as the comparative ‘more’.
. The reviewer has proceeded to reexamine the proportions by looking at the corpus in
2013. I provide the calculations here: 56 occurrences of personne (15 ND with pas, at 26.78%),
281 occurrences of rien (19 ND with pas, 6.76%). These figure are within the range of those
provided in Table 1 for the 2010 version of the corpus used for this research.
Pierre Larrivée
of plus (21/219) and 5.63% of n-words8 (21/373). These rates could profitably be
compared with those in more normative Quebec French practices, but the absence
of any freely accessible corpus of highbrow written or electronic press however
makes this impossible as of yet.
An established corpus of normative France French is available through Fran-
text, a corpus of mostly literary texts extending back to the 17th century. A simple
search of negative doubling for the 20th century part of the corpus shows that it
does exist and represents a heterogeneous sets of conditions: leaving aside irrel-
evant structures such as the n-word being co-ordinated to a negative clause or
depending on an infinitive, clause-mate co-occurrence of a n-word with pas is
found in archaic sequences where the n-word appears to be an indefinite (2 in
Claudel, 1 in Farrère, 1 in Roubaud) and in literary representations of informal
speech (1 in Proust, 4 in Barbusse, 2 in Martin du Gard, 5 in Céline, 1 in Beauvoir,
1 in Jacques Roumain; and 9 in Quebecois author Germaine Guévremont). How
does the actual France French vernacular stand with respect to negative doubling?
Looking at co-occurrence of negators and n-words was replicated with the CFPP,
with results in Table 2.
. The aggregate number of n-words provides a better comparative given the important
disparities between personne and rien due to the low number of occurrences of the former.
The continuity of the vernacular
much higher rate than pas; these rates are at more than 100 to 1 in Paris versus
nearly 10 to 1 in Quebec. Such contrasts are confirmed by an examination of the
Beeching corpus, where negative doubling is at 0,15% against pas, 8,13% against
plus, and against the 225 n-words at 3,11% for pas doubling and 22,22% for plus
doubling. The existence of doubling with pas and plus and the disparity between
them is also supported by a non-quantitative examination of Corpaix, Elicop and
Valibel corpora.
The examination of the rates of negative doubling in two varieties of contem-
porary vernacular French confirms the following facts:
–– Negative doubling with pas exists in vernacular France and Quebec French9
–– Negative doubling with plus is constant across contemporary vernacular vari-
eties, at about 10% of uses of plus or about 5% of n-word occurrences
–– These rates are much higher than those with pas
–– Negative doubling with pas is at least 10 times more frequent in Quebec
French than in France French (10.42 times wrt n-words – 7.5% vs. 0.72%,
15,57 times wrt the clausal negator – 1.09% vs. 0.07%)
The question that arises concerns the sources of the contemporary divergence
between vernacular France and Quebec French negative doubling with main
negator pas, and what the situation was historically. The expectation at this stage is
that the historical situation would have been comparable to either contemporary
France or Quebec French if there is stability in the vernacular, and that some iden-
tifiable factor would explain the divergence. How an answer is sought for these
questions is described in the next section.
While negative doubling with plus is convergent across contemporary French ver-
naculars, a divergence is found with pas negative doubling: it is well established
. The reviewer estimates that « a rate of under 1% suggests that the phenomenon is not part
of the grammatical system of the speakers », but does not cite any psycholinguistic evidence
for why this threshold should be significant, or explain why the phenomenon should then
exist at all if it is not part of the speakers’ competence. A threshold figure appears arbitrary
in as much as many productive syntactic phenomena are rare: very low frequency probably
characterises long-distance wh dependencies for instance (Who do you think bought roses?),
which are incontrovertibly part of the grammatical competence of speakers.
Pierre Larrivée
Table 3. Rates of negative doubling of rien with pas as compared to all sentential
scope uses of rien
16th 17th 18th 19th First half 20th
literary 0.3%* 2.8% 3.5% 2.7% 3.7%
informal 0.3%* 1% 1.7% 0 5.4%
The figures however appear curious in more ways than one. Variability may
be due to language change, to corpus limitations or to the selected criteria for
negative doubling. Occurrences of rien in a subordinate or a dependent infinitive
were included, which could explain the disparities with our own numbers in the
contemporary period. Yet, sequences such as Je ne veux pas qu’il dise rien ‘I don’t
want that he says nothing’ and Je ne veux pas dire rien de tel ‘I don’t want to say
nothing of the sort’ arguably involve something other than the direct doubling
in Je ne veux pas rien. Also, non-clause-mate uses with dependent infinitives
are more prominent in 17th and 18th centuries according to our non-quantified
observations. The corpus used comprises a large number of literary sources and
a smaller proportion of vernacular texts in which French and Quebec sources are
aggregated. There are grammatical divergences between the contemporary and
the historical varieties to be noted. The main post-verbal negator is not uniquely
represented by pas, which undergoes important competition from marker point
(see Larrivée in press for quantitative information on vernacular 16th and 17th
century texts). Both still have some amount of polarity uses in literary texts that
make their status somewhat ambiguous. Likewise, the negative status of n-words
is not always clear, since in literary material, rien ‘nothing’ still has a notable
proportion of polarity uses as shown by Martineau and Déprez (2004). This is
not the case of nul ‘no(ne)’ (Ingham 2011), which therefore is an ideal marker to
assess negative doubling in the history of French, to be measured against both
The continuity of the vernacular
Table 4. Negative doubling with nul in the Dictionnaire de Moyen Français10
1350–1360 1400–1410 1450–1460 1500–1510 1550–1560
The figures seem slightly more proportional than those in Table 3. Leaving
the 1550–1560 period aside, the tendencies of negative doubling at 0.29% of pas/
point (ranging from 0.11% to 0.42%) or at 0.55% of nul (with a peak at 1.79% in
1500–1510), and at 1.75% of plus (ranging from 1.13% to 2.98%) and 0.72% of nul
(from 0.22 to 1.179%), with a rise at the end of the period suggest that we are deal-
ing with a plausible set of data in view of contemporary figures. However, we again
find ourselves hostage to number of texts and occurrences. There is a slump in the
1400–1410 period, with an unexpected smaller number of plus doubling. Another
slump occurs in the 16th century. The later situation may be due to a mix of fewer
texts (the number of pas and point halve, and an even more important decrease is
noted for plus), a concentration on one item (nul) and a change in the paradigm of
. Numbers in the first cell each side of the slash/reflect numbers of doubling with pas and
point respectively.
Pierre Larrivée
n-words (nul intuitively seeming to disappear entirely from the vernacular at the
time, Ingham 2011).
The historical study of the vernacular is faced with the notorious difficul-
ties of finding data that reflect it. These difficulties come from the sociolin-
guistic fact that everyday language is rarely seen as worthy of being recorded
by the specialised and expensive technology of writing. The vernacular is thus
represented by “bad data” as discussed by Labov (1994: 11), and Nevalainen
and R aumolin-Brunberg (2003) among many others. Such bad data can be
approached with “good” methods that seek to uncover traces of the vernacular,
as is the case with work on prescriptive traditions (e.g. Wendy Ayres-Bennett
2004 for French). “Better” data can be looked for (Lodge 2009: 212) in the-
atrical parodies of the speech of the lower-classes (as unearthed by Anthony
Lodge himself; see Birk 2004), letters from semi-literate writers (as studied by
e.g. France Martineau), and texts from peripheral communities (as explored
by Raymond Mougeon, see Mougeon et al. 2010, for instance). The latter are
useful because peripheral communities may be more removed from norma-
tive pressures, and provide texts that give a better idea of everyday language:
Christopher Lucas finds in his 2009 study of changes in the grammar of Ara-
bic negation that the most revealing Middle Arabic texts come from Medieval
Spain, and are composed by Christians and Jews, “presumably because they had
less incentive to adhere to the classical standard” (Lucas 2009: 58). However, for
the purpose of this article, the New World data are rather late, and no publicly
available corpora exists yet for historical vernacular Quebec French (although
the letters from semi-literate writers assembled in the Corpus de français fami-
lier ancien project led by France Martineau are to become available). Old World
data therefore had to be used, and the plays, pamphlets and private diaries
put together for Martineau and Mougeon (2003; partly used for Martineau &
Déprez 2004) was referred to. The earlier plays were investigated from the ver-
sions (starred below) edited and made available on the Oxford Test Archive
website by Anthony Lodge, who has kindly given me access to his edition of two
private diaries; to this I added a set of letters of citizens to revolutionary France
administration edited by Sonia Branca-Rosoff and Nathalie Schneider. In these
I looked for all three negators with all n-words in order to measure the rates of
negative doubling, with text-by-text results listed below:11
. The numbers on each side of the ‘/’ indicate the number of instance of negative doubling
against the occurrences of the relevant clausal negator.
The continuity of the vernacular
Table 5. Negative doubling occurrences in 16th–20th vernacular material per text
pas point plus Notes
The corpus does not seem to be very revealing. The number of texts and
occurrences is such that only the 1743–1842 period provides significant figures.
In that period, we find negative doubling with pas and point at 0.89% of the total
Pierre Larrivée
number of these negators, or at 0.27% of n-words, and with plus at about 13% of
that clausal negator and 5.59% of n-words. The figures present a rise from the ear-
liest period, and a peak as compared to the contemporary period for pas and point,
and a higher rate than contemporary figures for plus.
Earlier periods that do not find quantitative support in the corpus can be
shown to attest the existence of negative doubling through various indications.
A 1558 play by Bonaventure Des Périers independently known for his use of the
vernacular provides three straight cases of pas doubling with rien, one of which is
cited below:
(7) – Voire mais, vostre femme est toute faschée ; que luy avez-vous faict?
– J’aury pas rien faict, ma dam ; (B. des Périers, 1558, Nouvelles récréations)
– True, but your wife is very upset, what did you do to her?
– I didn’t do nothing to her, my lady
(8) a. les Roys n’ont pas aucun droit sur les biens de leurs sujets,
(NA. LE POLITIQVE VNIVERSEL)
Kings haven’t no right on the possessions of their subjects
b. Ce n’est point aucun interest qui m’a poussé à respondre
It is not no self-interest that led me to answer
The quantitative and qualitative data reflecting the vernacular from the 17th to the
19th century demonstrate the historical existence of negative doubling and allow
some patterns to be identified. The evidence that the rate of negative doubling is
at 0.3% of pas and point uses and 0.9% of n-words for the 1743–1842 period and
the non-quantified indications from 16th and 17th century suggest that the rate is
a stable one when compared to the earlier and the contemporary French situation.
. Because the website returns duplicate examples, this number is an estimation, based on
the fact that there are 18 different examples among the first 25 occurrences (7 repetitions
therefore), with a ratio of 72% applied to the raw number of 280 examples.
The continuity of the vernacular
At over 10% of negator uses and 5% of n-words for the 1643–1842 period, plus
doubling is anticipating the contemporary disparity with pas.
The trajectory of negative doubling is discussed in the next section that articu-
lates the methodological opportunities and challenges of the study.
Clearly, the stability of pas/point doubling from the earliest period to contempo-
rary France French lends support to Labov’s hypothesis on the stability of the ver-
nacular. This hypothesis is not supported by the increase of negative doubling in
Quebec French, nor by the low rate of plus doubling in the earliest period. The
rise in plus doubling may have several causes. One is the disappearance of other
similar negative items with a temporal dimension such as onc and mais that are
survived by plus alone (and jamais, see Hansen 2012). The higher rate of negative
doubling in contemporary Quebec French may be explained by the fact that it is
Pierre Larrivée
–– Is the vernacular variety more systematic, primary, and more stable than for-
mal styles?
–– And if so, is there a continuity of (grammatical) configurations across ver-
nacular varieties?
The answer is that negative doubling in France French is indeed largely stable with
pas (and point) since the Middle French period: rates of doubling are roughly at
1 occurrence per 333 uses of clausal negators, or 1 case per 150 n-words. Diver-
gence is observed with Quebec French that has fairly exactly 10 times more pas-
doubling than contemporary France French. This is speculated to relate to lack
of vernacular data for the French contemporary and historical varieties, and to
normative pressures. The comparative plus doubling is also relatively stable in all
varieties since the Renaissance period, at roughly 1 plus in 10 used with a n-word,
and 1 n-word in 20 used with plus.
The observed stability provides support for the hypothesis by Labov that the
vernacular is a more systematic style of language. The difficulties of research on
language change in the vernacular have been pointed out. The changes in gram-
matical paradigms force creative enquiries, and the issues in bringing together
sufficient numbers of sources bring one to look for various qualitative indicators.
However, comparing rates of configurations found in peripheral varieties (Axel &
Weiss 2010) against a non-stigmatised equivalent in sources that may reflect
unmonitored free expression offers opportunities to further the understanding of
language variation and change mechanisms.
Pierre Larrivée
A number of questions are raised for future research. Why does negative dou-
bling exist in French and other languages? Is it an optional choice, or does it asso-
ciate with some form of emphasis or activation, as marked negative configurations
so often do (Larrivée 2010 and references therein)? How is it learned? Is it an
occasional failure to adhere to a normative suppression, or is it an actively pro-
duced combination? It seems that Quebec and France French may illustrate differ-
ent scenarios here. More generally, the question is raised of the relations between
linguistic practices: if as Ariel (2007) suggest languages cannot afford the luxury
of having different grammars for different styles, what then is the nature of the
relationship between them? We hope to contribute to such questions in future
research work.
References
Available resources
CFPP2000. Branca-Rosoff, Sonia, Fleury, Serge, Lefeuvre, Florence & Pires, Mat. Discours sur
la ville. Corpus de Français Parlé Parisien des années 2000 (CFPP2000). 〈http://cfpp2000.
univ-paris3.fr/〉
CFPQ. Dostie, Gaétane. Corpus de français parlé au Québec. 〈http://pages.usherbrooke.ca/cfpq/〉
Dictionnaire de Moyen Français.
Frantext. 〈http://www.frantext.fr/〉
OFROM. 〈http://www11.unine.ch/〉
WALS. Matthew Dryer & Martin Haspelmath (eds). The World Atlas of Language Structures
Online. 2013. 〈http://wals.info/〉
Studies
Anderwald, Lieselotte. 2005. Negative concord in British English dialects. In Aspects of English
Negation, Yoko Iyeiri (ed.), 113–37. Amsterdam: John Benjamins.
Ariel, Mira. 2007. A grammar in every register? The case of definite descriptions. In The Grammar–
Pragmatics Interface, Nancy Hedberg & Ron Zacharski (eds), 265–292. Amsterdam: John
Benjamins. DOI: 10.1075/pbns.155.16ari
Axel, Katrin & Weiss, Helmut. 2010. What changed where? A plea for the re-evaluation of dialec-
tal evidence. In Continuity and Change in Grammar, Anne Breitbarth, Christopher Lucas,
Sheila Watts & David Willis (eds), 13–34. Amsterdam: John Benjamins. DOI: 10.1075/
la.159.01axe
Ayres-Bennett, Wendy. 2004. Sociolinguistic Variation in XVIIth Century France. Cambridge:
CUP. DOI: 10.1017/CBO9780511486630
Bauche, Henri. 1946[1920]. Le langage populaire. Paris: Payot.
Berrendonner, Alain. 2004. Grammaire de l’écrit vs grammaire de l’oral: Le jeu des composantes
micro- et macro-syntaxiques. In Interactions orales en contexte didactique: mieux (se) com-
prendre pour mieux (se) parler et pour mieux (s’)apprendre. Alain Rabatel (ed.), 249–264.
Lyon: Presses Universitaires de Lyon.
The continuity of the vernacular
Labov, William. 1966. The Social Stratification of English in New York City. Washington DC:
Center for Applied Linguistics.
Larrivée, Pierre. In press. Reanalysis of negatives as polarity markers? The last 400 years of
decline of the French preverbal negative clitic. Lingua.
Larrivée, Pierre. 2011. Is there a Jespersen cycle? In The Evolution of Negation. Beyond the Jes-
persen Cycle, Pierre Larrivée & Richard Ingham (eds.), 1–23. Berlin: Mouton de Gruyter.
DOI: 10.1515/9783110238617.1
Larrivée, Pierre. 2010. The pragmatic motifs of the Jespersen cycle. Default, activation and the his-
tory of negation in French. Lingua 120(9): 2240–2258. DOI: 10.1016/j.lingua.2010.03.001
Larrivée, Pierre. 2009. Les Français, les Québécois, et la langue de l’autre. Paris: L’Harmattan.
Larrivée, Pierre. 2004. L’Association négative. Depuis la syntaxe jusqu’à l’interprétation. Geneva:
Droz.
Léard, Jean-Marcel. 1995. Grammaire québécoise d’aujourd’hui: Comprendre les québécismes.
Montreal: Guérin Universitaire.
Lemieux, Monique. 1985. Pas rien. In Les tendances dynamiques du français parlé à Montréal,
Tome 1, Henrietta Cedergren & Monique Lemieux (eds), 91–140. Montréal: Office de la
langue française.
Lodge, Anthony. 2009. La sociolinguistique historique et le problème des données. In Socio-
linguistique historique du domaine gallo-roman, Dorothée Aquino-Weber, Sara Cotelli &
Andres Kristol (eds), 199–219. Berne: Peter Lang. DOI: 10.1017/CBO9780511486685
Longobardi, G. 2001. Formal syntax, diachronic minimalism, and etymology: The history of
French chez. Linguistic Inquiry 32: 275–302. DOI: 10.1162/00243890152001771
Lucas, Christopher. 2009. The Development of Negation in Arabic and Afro-Asiatic. Ph.D. dis-
sertation, University of Cambridge.
Martineau, France. 2011. Ne-absence in declarative and yes/no interrogative contexts: Some pat-
terns of change. In The Evolution of Negation: Beyond the Jespersen Cycle, Pierre Larrivée &
Richard Ingham (eds), 179–208. Berlin: de Gruyter.
Martineau, France & Déprez, Viviane. 2004. Pas Aucun/Pas Rien en français classique: Variation
dialectale et historique. Langue Française 143: 33–47. DOI: 10.1353/lan.2003.0090
Martineau, France & Mongeon, Raymond. 2003. Sociolinguistic research on the origins of ne
deletion in European and Quebec French. Language 79(1): 118–152.
Martinon, Philippe. 1927. Comment on parle en français. Paris: Larousse.
Mazière, Francine. 2011. Usage des négations dans le discours de stabilisation du français au
17e siècle. In La négation en discours, actes du colloque des facultés des Lettres et Sciences
Humaines de Sousse et de Kairouan, Tunisie, 3, 4 et 5 avril 2008, Jacqueline Bacha, Ammar
Azouzi & Khaled Saddem (eds), 15–33. Sousse: Université de Sousse.
Meisel, Jürgen M., Elsig, Martin & Bonnesen, Matthias. 2011. Delayed grammatical acquisition
in first language development: Subject-verb inversion and subject clitics in French inter-
rogatives. Linguistic Approaches to Bilingualism 1(4): 347–390. DOI: 10.1075/lab.1.4.01mei
Mougeon, Raymond, Nadasdi, Terry & Rehner, Katherine. 2010. The Sociolinguistic Competence
of Immersion Students. Bristol: Multilingual Matters.
Muller, Claude. 1991. La négation en français: Syntaxe, sémantique et éléments de comparaison
avec les autres langues romanes. Genève: Droz.
Muller, Claude. 1984. L’association négative. Langue Française 62: 59–94. DOI: 10.3406/lfr.1984
.5191
The continuity of the vernacular