on the relation between double clausal negation and negative concord

58
On the relation between double clausal negation and negative concord 1 Lauren Van Alsenoy & Johan van der Auwera University of Antwerp 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. personne in (2)). (1) French Je ne le vois pas. I NEG him see NEG ‘I do not see him.’ 1 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).

Upload: antwerp

Post on 01-Mar-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

On the relation between double clausal negation and negative concord1

Lauren Van Alsenoy & Johan van der Auwera

University of Antwerp

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. personne in

(2)).

(1) French

Je ne le vois pas.

I NEG him see NEG

‘I do not see him.’

1 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).

2

(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, earlier, 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

2. Some obvious similarities and differences

In (1) double negation, henceforth abbreviated as ‘DN’, is illustrated with French and French

is probably the best studied case. But the phenomenon is widespread, see e.g. the illustration

from the Zeela dialect of the Bantu language Luba in (3).

3

(3) Luba, Zeela dialect (Kabanga Mukala p.c.)

kà-kùpííl-éè-pó mwáàná

NEG.1-hit-FIN-NEG 3.child

‘Father has not hit the child (at all).’

‘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 1917.2 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).

(4) NEG1 → NEG1 NEG2 → NEG 2

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.’

2 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).

4

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 e

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.’

5

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 exponence 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 …’)

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 negative (any more) is called ‘the

standard view’ by De Swart (2010: 177).3 Voices of authority often referred to in this context

are those of Damourette & 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 & Pichon, is to express discord (‘une discordance’) or, in the

clearer terms of Tesnière (1959: 111):

3 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

(contagion) by the negative particle ne, not that ne lost its negative force.

6

Le discordantiel ne forme pas à lui seul la négation. Il la prépare seulement. Et c’est le

forclusif qui la réalise.4

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.’

4 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.

7

(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.

A further similarity is that for both DN and NC the multiple exponence is not restricted to

a factor of two. Thus both have tripling variants. For NC this is not special at all and it has

long been realized. (11) is an English non-standard example.

(11) I didn’t say nothing to nobody.

‘I didn’t say anything to anybody.’

For DN, the recognition that the Jespersen scenario of weakening and strengthening 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 negative 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.

8

(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 never had much fun with anyone anywhere.’

For clausal negation, however, marking with three negations is fairly rare already, making 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 context the emphasis carried by the ‘even’

9

sense bleached and ne contaminated them to ‘nobody’ and ‘nothing’. Pas has a similar origin,

‘not even one step’, originally in the context of movement verbs, the ‘step’ meaning also

bleached and got contaminated, but in addition, the nominal negative sense gave way to a the

sense of clausal negation (‘not’).5 This evolution is sketched in (14).6

5 Not every indefinite often considered to be in the same paradigm as personne and rien

underwent the same evolution, however. See Hansen (2011).

6 Note that we are fully aware of the fact that rien and esp. 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).

(a) - Qui est venu ce soir?

who is come this evening

‘Who has come this evening?’

- Personne.

nobody

‘Nobody.’

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.

10

(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’

So at least in the case of French, there is a sense in which DN presupposes NC. As we will see

in the next section, there are two accounts that take facts such as those of French into a

typological forum.

3. A typological relation?

In 2010 Henriëtte de Swart suggests a direct connection between DN and NC. After a

typological analysis of the Jespersen cycle and NC, in general, and of their French versions, in

particular, she continues:

In fact, the analysis developed here suggests that a crucial condition for the

development of a discontinuous negation along the lines of French is for the language to

[...] display strict negative concord. It would be worth exploring this issue in more

detail, but currently I do not have all the cross-linguistic data needed to substantiate this

claim, so the connection is left for future work. (De Swart 2010: 184)

Let us look at this quote in some detail. First, a condition can be necessary or sufficient. De

Swart (2010) uses the term ‘crucial condition’. We take this to mean ‘necessary condition’,

11

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.7 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.8 Fourth, the link between DN and strict NC would only concern DN

‘along the lines of French’. This expression is vague: just how similar does the DN of a

7 The phrase left out in the quote says that a language must be a so-called ‘Type III language’.

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 standard

French has strict NC, we take it that the deleted phrase has a typing error.

8 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

discontinuous 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.

12

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.9

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 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’.

9 Bernini & 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’.

13

(16) NC is a necessary condition for PreVN10

A few remarks are in order. First, for clarity’s sake we added the word ‘necessary’ 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 concern 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 without 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

10 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.

14

admitted language types is attested and there are no attestations of the non-admitted type.11 Of

course, as remarked already, the languages are nearly all European, the only exceptions being

Tamazight Berber and Hebrew. To base a typology – and this is indeed what Zeijlstra (2004)

aims to be doing – on this kind of heavily biased sample is dangerous and at best only a

highly tentative ‘pilot typology’ (van der Auwera 2012).

11 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) I didn’t see anybody.

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.

15

PreVN NC

Possible

Czech, Polish, Russian, Serbo-Croatian; Catalan, French

Italian, Spanish, Portuguese, Romanian; Greek;

Hungarian; Tamazight Berber, Hebrew

+ +

Quebecois; Bavarian, Yiddish, NC English − +

German, Norwegian, Swedish, Standard English − −

Impossible Ø + −

Table 1. Languages studied by Zeijlstra (2004) and supporting the claim in (16)

4. Negative 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 Americas (see Appendix

A). These languages are taken from a representative 179-language 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 world12, 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 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 classified 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

12 Whole world coverage is the aim of Van Alsenoy (in prep.).

16

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 of 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 Austronesian and Austro-Asiatic.

17

DN NC

Possible

Af: Ewe / Am: Karok (2 languages) + +

Af: Kanuri, Kunama, Somali / Am: Huave, Chiapas Zoque, Páez

/ Au: Chamorro (7 languages)

− +

Khoekhoe, Ju’huan, Diola-Fogny, Yoruba, Igbo, Nupe, Ijo,

Bagirmi, Ngiti, Ma’di, So, Maasai, Dongolese Nubian, Murle,

Masalit, Koyraboro Senni, Masa, Iraqw, Maale, Pacoh, Pnar

(Khasi), Khmer, Nicobarese, Khmu’, Vietnamese, Seediq,

Kambera, Maori, Biak, Paiwan, Tagalog, Tukang Besi, Karo

Batak, Greenlandic, Plains Cree, Yuchi, Koasati, South East

Puebla Nahuatl, Comanche, Pima Bajo, Makah, Bella Coola,

Lilloeet, Kutenai, Klamath, Nez Perce, Seri, Wappo, Sochiapan

Sochiapan Chinantec, Chalcatongo Mixtec, Otomí, Chocho,

Purépecha, Misantla Totonaca, Mam, Epena Pedee, Tuyuca,

Andoke, Warao, Waorani, Yagua, Shipibo-Konibo, Jaqaru,

Nadëb, Apalaí, Mekens, tapieté, Canela-Krahô, Trumai, Kwazá,

Wari’, Paumarí, Movima, Mosetén, Chipaya, Pilagá, Mapuche,

Gününe Küne (78)

− −

Impossible Af: Gbeya-Bossangoa, Degema, Supyire, Kresh, Tera, Egyptian

Arabic / Am: Navajo, Haida, Wiyot, Kiowa, Tsimshian, Wintu,

Maricopa, Awa Pit, Araona, Imbabura Quechua (16 languages)

+ −

Table 2. Languages chosen to test the claim in (15)

18

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 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 & 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.

19

PreV N NC

Possible

Af: Ewe, Somali / A: Chiapas Zoque, Huave, Karok / Au:

Chamorro (6 languages)

+ +

Af : Kanuri, Kunama / Am: Páez (3 languages) − +

Af: Khoekhoe, Diola-Fogny, Igbo, Nupe, Ijo, Bagirmi, Ma’di,

Nubian, Masalit, Masa, Iraqw, Maale / Am: Eskimo

Greenlandic, Koasati, Wappo, Chocho, Epena Pedee, Tuyuca,

Warao, Waorani, Shipibo-Konibo, Apalaí, Mekens, Tapieté,

Canela-Krahô, Trumai, Kwazá, Mapuche / Au: Biak (29

languages)

− −

Impossible Af: Ju|’Hoan, Gbeya-Bossangoa, Yoruba, Degema, Supyire,

Kresh, Ngiti, So, Maasai, Murle, Koyraboro Senni, Tera,

Egyptian Arabic / Am: Navajo, Haida, Plains Cree, Wiyot,

Yuchi, Kiowa, Nahuatl, Comanche, Pima Bajo, Makah, Bella

Coola, Lillooet, Kutenai, Klamath, Nez Perce, Tsimshian,

Wintu, Seri, Maricopa, Sochiapan Chinantec, Mixtec, Otomí,

Purépecha, Misantla Totonaca, Mam, Awa Pit, Andoke, Yagua,

Imbabura Quechua, Jaqaru, Nadëb, Wari’, Paumarí, Araona,

Movima, Mosetén, Chipaya, Pilagá, Gününe Küne / Au: Pacoh,

Pnar, Khmer, Nicobarese (Car), Khmu’, Vietnamese, Seediq,

Kambera, Maori, Paiwan, Tagalog, Tukang Besi, Karo Batak

(65 languages)

+ −

Table 3. Languages chosen to test the claim in (16)

20

The conclusion is clear: more than half of the languages studied are counterexamples to

Zeijlstra’s claim that NC is a necessary condition for PreVN. So we should consider

Zeijlstra’s universal to be wrong.

Note also that Table 3 shows preverbal negation to be more common than postverbal

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.

properties number of languages properties number of languages

+NC +DN 2 22% -NC +DN 16 20%

-DN 7 78% -DN 80 80%

Table 4. Correlation between the presence of negative concord and double negation

It says that of the 9 languages with NC 2 have DN and 7 don’t, whereas in the 96 languages

without NC 16 have DN and 80 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

21

shown in Table 5: knowing that a language has DN has no predictive value for answering

whether the language has NC not.

properties number of languages properties number of languages

+DN +NC 2 11% -DN +NC 7 9%

-NC 16 89% -NC 80 91%

Table 5. Correlation between the presence of double negation and negative concord

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 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.

properties number of languages properties number of languages

+NC +PreVN 6 67% -NC +PreVN 65 69%

-PreVN 3 33% -PreVN 29 31%

Table 6. Correlation between the presence of negative concord and preverbal negation

22

properties number of languages properties number of languages

+PreVN +NC 6 9% -PreVN +NC 3 10%

-NC 65 91% -NC 29 90%

Table 7. Correlation between the presence of preverbal negation and negative concord

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.

5. Languages with 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 exponents of clausal negation merged with the pronoun. We will start with

Ewe.

23

5.1. Ewe and contamination

Ewe is a DN language, with a preverbal negative prefix me- and a clause-final negator particle

o.

(18) Ewe (Ameka 1991: 64-65)

Kofí mé vá afí sia o

Kofi NEG come place this NEG

‘Kofi did not come here.’

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 whatever, whoever, whenever, etc. A positive, free

choice use is illustrated in (21).

(19) Ewe (Nada Gbegble, p.c.)

ame aɖe fe le abɔ la me

person one play at garden the in

‘Somebody played in the garden.’

24

(20) Ewe (Nada Gbegble, p.c.)

ame aɖeke me-fe le abɔ la me o.

person no NEG-play at garden the in NEG

‘Nobody played in the garden.’

(21) Ewe (Nada Gbegble, p.c.)

àmè kà ké-é lè …

person INT EMP-FOC is…

‘whoever it may be, …

Though aɖeke is originally an emphatic form of aɖe ‘one’, paraphrasable as ‘even one’ or as

stressed any in English, the use of ame aɖeke in (20) is restricted to negative sentences, like

French personne. However, it is not “as negative” as French personne, in that it cannot

function as an elliptic answer to questions. The correct negative answer to the question Who

played in the garden? is (22).

(22) Ewe (Nada Gbegble, p.c.)

me-nye ame aɖeke o

NEG-be person one NEG

‘Nobody.’ (lit. ‘It was nobody.)

Another parallel with French, is that more than one negative quantifier can occur with only

one semantic negation.

25

(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 combine 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 determiner: 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)

(24) Pseudo-English (based on Haspelmath 1997: 121)

You can go, where-even/ever it may be. >

You can go wher-ever.

Second, the scalar focus particles may combine directly with the generic or interrogative,

depending on what the derivational base of the indefinite is. Again a pseudo-English example

can be used for illustration as in (25):

26

(25) Pseudo-English

I never saw even-person.

In this case, the process of grammaticalization can be compared to the grammaticalization of

minimal-unit expressions, like French pas, personne and rien. In Ewe, the focus marker is

indeed used for non-specific free relatives, but as (21) shows, there is also an interrogative

element (ka). It seems safer therefore to assume that the negative indefinite resembles the

scenario of (25) rather than that of (24).

Adding a scalar focus marker seems to be a common means of derivation. Haspelmath

(1997: 157-158) mentions the use of scalar focus markers in negative indefinites in Selkup,

Nivkh, Japanese, Chechen, Hebrew and Lezgian. In our sample, we also see it in Kanuri

(Cyffer 2009: 86), Degema (Kari 2004: 131), Somali (Saeed 1999: 184) and optionally on the

indefinites in negative contexts in Shipibo Konibo (Valenzuela 2003: 370). It seems that in all

these cases we are dealing with the formation of negative indefinites by directly adding the

scalar marker to the base. None of these languages, however, is a DN language. Only Ewe is,

and what we thus see in an Ewe ‘nobody’ or ‘something’ sentence is a combination of DN

and NC.

Indefinites with scalar markers are not only used for negative indefinites. Just like

minimal-unit expressions like French pas and rien in French, scalar indefinites are also often

used in negative polarity contexts. We also found this in our sample, e.g. in Eskimo

Greenlandic (Bittner 1995: 76), Awa Pit (Curnow 1997: 314), Quechua (Cole 1982: 86) and

Seediq (Henningsson & Holmer 2008: 36). In Ewe, however, emphatic indefinites are

restricted to negative contexts and cannot occur in other negative polarity contexts.

27

5.2 Karok and negative absorption

Ewe aɖeke is originally positive but a negative context contaminates it and it becomes

negative. The form aɖeke itself is morphologically interesting too, as it is a univerbation of

ordinary aɖe ‘one, some’ and an emphatic marker. The latter is not independently negative.

This is different in Karok. Here we also see univerbation, but what gets added to the positive

marker is a directly negative marker. We call the univerbation of a positive pronoun or

determiner with a negative marker ‘negative absorption’, as defined in Haspelmath (1997:

205).

Karok is a DN language with as the main negative strategy a combination of a prefix pu-

and a suffix –(h)ara (Bright 1957: 137). Although doubling is the main strategy, the

postverbal element is absent under certain morpho-syntactic circumstances (whenever there is

a personal morpheme containing –ap on the verb or a ‘fourth-order class suffix’). It is also

absent when the verb contains the emphatic marker –xay. The prefixal negator pu- can have

different hosts: it freely attaches to ‘any word which stands before the predicate in a

predication’ (Bright 1957: 138). It can thus also attach to the indefinites fâˑt ‘what,

something’, ʔakáray or yíθθa ‘one’, and this way we get negative pronouns or determiners, at

least in some cases, and with our permissive attitude on NC status, we take the combination of

the negative pronoun or determiner and the negative predicate to constitute NC. Bright (1957:

140) lists the negative indefinites pu-ffa·t ‘nothing’, pu-ʔakára ‘nobody’, but he doesn’t

provide any examples. He does provide an example of a sentence starting with the negative

marker pu- and the indefinite determiner ‘one’.13

13 This emphatic element –xay is, remarkably, also the element that can replace the second

negator, as in (a).

28

(26) Karok (Bright 1957: 140)

pú-yíθθa-xay ká·n θa·nê-·ra

NEG-one-EMP lady lay-NEG

‘Not a single lady lay there.’

The negative absorption we see in Karok is a process found elsewhere. In Africa we only

found it in Egyptian Arabic (see below), in Austronesian only in Nicobarese14, but in our

(a) Karok (Bright 1957: 138)

pú-xay vúra-xay ʔamkú·f-xay

NEG-EMP just-EMP was-EMP

‘There was no smoke at all.’

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.

14 Not surprisingly, negative absorption doesn’t occur often in Austronesian or Austro-Asiatic

languages, since these are often isolating languages. Apart from that, negative indefinites 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).

29

sample it is common in the Americas: Chinantec, Chalcatongo Mixtec, Otomí, Chocho,

Comanche, Pima Bajo, Nahuatl, Misantla Totonac, Yuchi, and indeed Karok, all of them in

North America, and in South America also in Paumarí and Mekens (Galucio 2001: 93). Of

course, we know the phenomenon from European languages too, e.g. English and German,

and also, in a certain way, Russian and Spanish, which will be discussed below.

For the process of negative absorption, there are at least three parameters of variation. We

will characterize Karok in terms of these parameters and compare it to languages that have

different parameter settings.

A first parameter concerns the nature of the negative element that absorbs into the

indefinite. What combines with the indefinite in Karok is the preverbal part of a DN. This is

also the case in e.g. older West Germanic, too. In Old German (27), for instance, the negative

indefinites are composed of the positive indefinite, e.g. ieman, ioman ‘someone’ and n(i)-/nih-

/ne(h)-. The negative element ni/ne also serves as the preverbal part of the DN construction

ne/ni …nicht, illustrated in (28).

(27) Old High German (Jäger 2008: 206)

(a) ɨm-em wa yoʔsuk ya-o

NEG-have RP like ACC-3MSG

‘Nobody likes him.’ (lit. ‘there is not who likes him.’)

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 1979:186), the

American languages Wayampi (Jensen 1994:346), Lillooet (Davis 2005:20), Kutenai

(Boas&Goddard 1927), the Austronesian language Chamorro (Cooreman 1987:45).

30

nioman nimag zuueion herron thionen.

nobody NEG-can two lords serve.

‘Nobody can serve two lords.’

(28) Old High German (Jäger 2008: 61)

Ih nehábo nîeht in geméitun sô uîlo

I NEG-have not.at.all/NEG in vain so much

geuuêinot

cried.

‘I did not cry that much in vain.’

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 absorption 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.’

31

A variant on Karok negative absorption is found in Egyptian Arabic, which also has DN,

as illustrated in (30), but in this language both parts get absorbed – see (31).

(30) Egyptian Arabic

ma-saafir-t-š

NEG-traveled-1SG-NEG

‘I did not travel.’

(31) Egyptian Arabic (Haspelmath 1997: 207)

ma-ħaddi-š yiʕraf yi-ʔra xåt̥t̥-i

NEG-one-NEG can read writing-1SG

‘Nobody can read my writing.’15

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 absorption. The

15 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.

32

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 independent of (and sometimes

formally unrelated to) the verbal negator which expresses sentence negation’. Still we decided

to treat them together with other cases of negative 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 addition, 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 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 explaining why most languages prefer

preverbal to postverbal negation.

An interesting variation is Navajo. Navajo has double negation. The postverbal part seems

to be the oldest part (Van Gelderen 2008: 220), its verb is clause-final and the preverbal part

33

has a variable position. But when there is an indefinite, the preverbal part is attracted to the

indefinite16. 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.’

16 “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).

34

The third parameter of variation in negative absorption concerns the relation with NC. In

Egyptian Arabic there is no NC17, 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 the preverbal part of a DN is attracted to the

indefinite. We assume that the NC constellation in (27) was predated by a constellation in

which negation was only marked on the indefinite, which is very sporadically found in Old

High German, as shown in (35).

(35) Old High German (Jäger 2008:207)

Inthemo noh nu níoman ingisezzit uuas

in-which still nu nobody put was

‘in which nobody had been put yet’

The constellation in (35) is undesirable, however, so Haspelmath (1997: 203) argues: the

absorbed negation semantically still has sentence scope, i.e., it does not create a negative

word in a positive sentence, as with un- in the negative word undesirable in the positive

sentence (36).

(36) This outcome is undesirable.

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 probably also happened

17 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).

35

in Russian, which had non-strict NC, which was (slightly) preferred 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 independent 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 colloquial French, it now allows clause scope pas, thus

reinstating NC.

(37) Québec French (Haspelmath 1997: 205)

Le samedi soir, au mois de juillet, il y a pas

the satuday night in.the month of july there is NEG

personne en ville à Québec.

nobody in city in Québec

‘On Saturday nights in July, there is nobody in Quebec city.’

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. Haspelmath (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 Mesoamerican languages too. This is confirmed

36

by our sample. As already suggested when we listed the languages in our sample that have

negative absorption, it seems quite common in the Americas, as e.g. in Chalcatongo Mixtec.

In (38), the negator is on the verb, but in (39) it is on the indefinite.

(38) Chalcatongo Mixtec (MacCaulay 1996: 120)

tu-nixížaa-ró

NEG-were.located-2

‘You weren’t there.’

(39) Chalcatongo Mixtec (MacCaulay 1996: 124)

tú-ndéú nikii=Ø

NEG-who came=3

‘Nobody came.’

Another illustration is Mekens, which is particularly interesting, for it shows that the

absorption need not be prefixal, but can be affixal as well. In Mekens there are two clausal

negators, among which –bõ/-õ, depending on the phonological form to which they attach

(Galucio 2001:93). It can attach to verbs as in (40) but also to indefinites, as in (41).

(40) Mekens (Galucio 2001: 94)

e-aso-bõ ẽt

2SG-bath-NEG you

‘Do not bathe.’

37

(41) Mekens (Galucio 2001: 93)

arob-õ ki-iko ke te te ose

thing-NEG 1PL-food that truly FOC we

‘Nothing is our food, we are like that.’

Apart from the fact that the pattern of marking clausal negation on the indefinite only

seems more frequent than previously assumed, it is also found in languages in which it was

hypothesized to be impossible. According to Haspelmath (1997: 206), negative absorption is

predicted not to occur in verb-initial languages. But this is wrong. Tepetotutla Chinantec18 is

verb initial. Negation is normally a verbal prefix.

(42) Tepetotutla Chinantec

caL-ʔnioL teLH zóL ʔiHkuï̜ʔM

NEG-want Esther to.go Oaxaca

‘Esther doesn’t want to go to Oaxaca.’

(Westley 1991: 71)

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.

18 Strictly speaking, this variety is not part of the sample. The latter contains Sochiapan

Chinantec.

38

(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.

6. Languages with NC but no DN

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 interference 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 negative 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).

39

(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.’

In Somali, the marker ba on the indefinite may correspond to English ever.

(46) Somali (Saeed 1997: 127)

méel-ta-aad tagtó-ba ...

place-the-you go-ever

‘Wherever you go …’

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.19

19 Somali can also derive negative indefinites by adding a negative nominal suffix. We do not

understand this process.

40

(47) Somali (Saeed 1997: 186)

waxbá má aan sɪ́n

thing-NEG NEG I gave

‘I didn’t give him anything.’

It seems that in Kunama, the negative indefiniteness marker is also related to an emphatic

marker. In Kunama, ‘nobody’ and ‘nothing’ are expressed by (dáat)éllá-ná ‘nothing’ and kéll-

ána ‘nobody’ respectively (Bender 1996: 22-23). The ending resembles the emphatic marker

ma, which can be used on positive as well as negative verbs (Bender 1996: 36). That the use

of the negative indefinites requires an additional verbal negation is illustrated in (48).

(48) Kunama (Bender 1996: 23)

ellana na-nti-mme

nothing I-saw-NEG

‘I didn’t see 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.

(a) Somali (Saeed 1997: 186)

qóf-na má arkɪ́n

person-NEG NEG saw

‘I/you/he etc. didn’t see anyone.’

41

(49) Páez (Jung 2008: 196)

kɪ ̃x 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 indefinites are

clearly at least partly borrowed from Spanish. These languages are Zoque, 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

as the little boy NEG speak NEG answer

ñi-kwej.

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

42

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 Chamorro

resembles Spanish.

43

(54) Chamorro (Chung 2009: 103)

Ni unu sin᷉a tasokni nu esti na chinätsaga

NEG one can blame OBL this L hardship.

‘We can blame no one for this hardship.’

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

44

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

1 first person

2 second person

3 third person

DECL declarative

DET determiner

EMP emphatic

FIN finite

FOC focus marker

GEN genitive

INDEF indefinite

INEG incompletive negation

INT interrogative

MASC masculine

NEG negative marker

NEG.DET negative determiner

PL plural

POS possessive

PRES present

PT particle

Q question marker

R realis

S subject

SG singular

SS same subject

UNM unmarked

V verb

45

References

Ameka, Felix K. 1991. Ewe : its grammatical constructions and illocutionary devices. PhD

dissertation, Australian National University.

Bamgbose, Ayo. 1966. A grammar of Yoruba. Cambridge: Cambridge University Press.

Bender, Lionel. 1996. Kunama. München: Lincom.

Bernini, Giuliano & Paolo Ramat. 1996. Negative sentences in the languages of Europe. A

typological approach. Berlin: Mouton de Gruyter.

Bitnner, Maria. 1995. Quantification in Eskimo. In Quantification in Natural Languages,

Bach, Emmon W., Eloise Jelinek, Angelika Kratzer & Barbara H. Partee (eds), 59-80.

Dordrecht.

Breál, Michel. 1987. 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.

Bright, William. 1957. The Karok language. Berkeley: University of California Press.

Broadwell, George A. 2001. Optimal order and pied-piping in San Dionicio Zapotec. In

Formal and Empirical Issues in Optimality Theoretic Syntax, Peter Sells (ed), 197–123.

Stanford: CSLI.

Brown, Richard D. 1994. Kresh. In Typological Studies in Negation, Peter Kahrel & René van

den Bergh (eds), 163-189. Amsterdam: John Benjamins,

Carlson, Robert. 1994. A grammar of Supyire. Berlin: Mouton de Gruyter.

Chapman, Shirley and Derbyshire, Desmond C. 1991. Paumari.. In: Handbook of Amazonian

Languages 3, Desmond C. Derbyshire & Pullum, Geoffrey K. (eds), 161-352. Berlin:

Mouton de Gruyter.

46

Charney, Jean Ormsbee. 1993. A Grammar of Comanche. London/Lincoln: University of

Nebraska Press.

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: MIT Press,

Cole, Peter. 1982. Imbabura Quechua. Amsterdam: North-Holland Publishing

Company.

Cooreman, Ann M. 1987. Transitivity and discourse continuity in Chamorro narratives.

Berlin: Mouton De Gruyter.

Curnow, Timoth Jowan. 1997. A grammar of Awa Pit (Cuaiquer): An indigenous language of

south-western Columbia. PhD dissertation, Australian National University.

Cyffer, Norbert. 2009. Negative Patterns in Kanuri. In Negative Patterns in West African

languages and Beyond, Norbert Cyffer, Erwin Ebermann & Georg Ziegelmeyer (eds),

71-92. Amsterdam: John Benjamins.

Damourette, Jacques & Édouard Pichon. 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.

De Swart, Henriëtte 2010. Expression and interpretation of negation. An OT typology.

Dordrecht: Springer.

Donohue, Mark. 1999. A grammar of Tukang Besi. Berlin: Mouton de Gruyter.

Dryer, Matthew. 1998. Univerals of negative position. In Studies in syntactic typology,

Michael Hammond, Edith Moravcsik & Jessica Wirth (eds), 93-124. Amsterdam:

Benjamins.

47

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).

Egli, Hans. 1990. Paiwangrammatik. Wiesbaden: Otto Harrassowitz.

Early, Robert. 1994a. Lewo. In Typological studies in negation , Peter Kahrel & René van den

Berg (eds), 65-92. Amsterdam: John Benjamins.

Early, Robert. 1994b. A grammar of Lewo, Vanuatu. PhD dissertation, Australian National

University.

Faarlund, Jan Terje. 2012. A Grammar of Chiapas Zoque. Oxford: Oxford University Press.

Galucio, Ana Vilacy. 2001. The Morphosyntax of Mekens (Tupi). PhD dissertation, University

of Chicago.

Gelo, Daniel J. 1995. Comanche Vocabulary. Trilingual edition. Compiled by Manuel García

Rejón. Austin: University of Texas Press.

Hale, Ken & Paul Platero. 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: Oxford University Press.

Hansen, Maj-Britt Moseegaard. 2011. On the evolution of temporal n-words in Medieval

French. Language Sciences 34: 76-91.

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).

Hekking, Ewald. 1995. El otomí de Santiago Mexquititlán: Desplazamiento lingüístico,

préstamos y cambios gramaticales. Amsterdam: Instituut voor Functioneel Onderzoek

van Taal en Taalgebruik (IFOTT).

48

Henningsson, Lars-Ake & Arthur Holmer. 2008. The distribution of quantifiers in Seediq.

Lund University Working Papers 53: 23-41.

Hernández, Roberto Escalante & Zarina Estrada Fernández. 1993. Textos y gramática del

Pima Bajo. Sonora: Departamento de Letra y Linguistica, Universidad de Mexico.

Horn, Laurence 1989. A natural history of negation. Chicago: University of Chicago Press

Israel, Michael. 2011. The Grammar of Polarity. Cambridge: Cambridge University Press.

Jäger, Agnes 2008. History of German negation. 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 Europa.

Kari, Ethelbert Emmanuel. 2004. A reference grammar of Degema. Köln: Rüdiger Köppe

Verlag.

Kim, Yuni. 2008. Topics in Phonology and Morphology of San Francisco del Mar Huave.

PhD dissertation, University of Berkeley, California.

Koshy, Anish. 2009. Indefinite pronouns in Pnar. Mon-Khmer Studies 38: 41-56 .

Linn, Mary Sarah. 2001. A Grammar of Euchee (Yuchi). PhD dissertation, University of

Kansas.

Lucas, Christopher. 2009. The development of negation in Arabic and Afro-Asiatic. PhD

dissertation, University of Cambridge.

Macaulay, Monica. 1996. A Grammar of Chalcatongo Mixtec. University of California Press.

MacKay, Carolyn Joyce. 1999. A Grammar of Misantla Totanac. Salt Lake City: University

of Utah Press.

MacSwan, Jeff. 1999. A Minimalist Approach to Intrasentential Code Switching. New York:

Garland.

49

Man, Edward Horace. 1889. A dictionary of the central Nicobarese language (English-

Nicobarese and Nicobarese-English). London: W.H. Allan.

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. Amsterdam, Philadelphia : John Benjamins.

Sakel, Jeanette. 2002. A Grammar of Mosetén. Berlin : Mouton de Gruyter.

Samarin, William J. 1966. The Gbeya Language. Grammar, Texts and Vocabularies.

Berkeley & Los Angeles: University of California Press.

Schachter, Paul & Fé Otanes. 1983. Tagalog reference grammar. University of California

Press.

Smeets, Ineke. 1986. A grammar of Mapuche. Berlin : Mouton de Gruyter.

Tesnière, Lucien. 1959. Éléments de syntaxe structure. Paris : Klincksieck.

Valenzuela, Pilar. 2003. Transitivity in Shipibo-Konibo grammar: a typologically oriented

study. PhD dissertation, University of Oregon.

Van Alsenoy, Lauren. 2011. Negative expressions involving indefinites in African languages,

Studies van de BKL – Travaux du CBL 6.

<http://webh01.ua.ac.be/linguist/SBKL/Vol6.htm>

Van Alsenoy, Lauren. In prep. A typology of indefinites. PhD dissertation, University of

Antwerp.

50

van der Auwera, Johan. 2009. The Jespersen cycles. In Cyclical change, Elly Van Gelderen

(d), 35-71. Amsterdam: John Benjamins.

--- 2010. On the diachrony of negation. In The expression of negation, Laurence R. Horn (ed),

73-101. Berlin: Mouton de Gruyter.

--- 2012. From contrastive linguistics to linguistic typology, Languages in Contrast 12: 69-86.

van der Auwera, Johan, Frens Vossen & Maud Devos. 2013. Le cycle de Jespersen à trois ou

quatre négations, La linguistique de la contradiction, Jacques François, Pierre Larrivée,

Dominique Legallois & Franck Neveu (eds), 19-30 .Bern: Lang.

van der Auwera, Johan & Annemie Neuckerman. 2004. Jespersen’s cycle and the interaction

of predicate and quantifier negation in Flemish, In Typology meets dialectology. Dialect

grammar from a cross-linguistic perspective, Bernd Kortmann (ed), 454-478. Berlin:

Mouton de Gruyter.

Vandeweghe, Willy. 2009. Negatievermenigvuldiging in het West-Vlaams. Studies van de

BKL 4 – Travaux du CBL 4. <http://webh01.ua.ac.be/linguist/SBKL/Vol4.htm>

Van Gelderen, Elly 2008. Negative Cycles. Linguistic Typology 12: 195-243.

Van Olmen, Daniel, 2010. Typology meets usage: The case of the prohibitive infinite in

Dutch. Folia Linguistica 44: 471–508.

Veerman-Leichsenring, Annette. 2000. Gramática del Chocho. Leiden: Centre of Non-

Western Studies (CNWS).

Vennemann, Theo. 1974. Topics, subjects and word order: From SXV to SVX via TVX, In

International Conference on Historical Linguistics I. John M. Anderson & Charles Jones

(eds), 339–376. Amsterdam: North-Holland

Vossen, Frens & van der Auwera, Johan. This volume. The Jespersen cycle seen from

Austronesian.

51

Westley, David O. 1991. Tepetotutla Chinantec Syntax. Arlington, Texas: The Summer

Institute of Linguistics and University of Texas.

Williamson, Kay. 1965. A Grammar of the Kolokuma Dialect of I̩jo. Cambridge: Cambridge

University Press.

Zeijlstra, Hedde 2004. Sentential negation and negative concord. PhD dissertation, Utrecht

University.

Appendix

A. The sample

The languages between square brackets are the ones for which insufficient data have been

gathered.

Table 7. Languages in the RS, families, genera, and sources

Family Genus Language Sources

Africa

Khoisan Central Khoisan Khoekhoe Hagman 1977

Böhm 1985

Northern

Khoisan

Ju∣’huan Snyman 1970

Dickens 2005

Niger- Adamawa- Gbeya Bossangoa Samarin 1966

52

Congo Ubangian

Northern

Atlantic

Diola-Fogny Sapir 1965

David J.Sapir, p.c.

Defoid Yoruba Ashiwaju 1968

Rowlands 1965

Bamgbose 1966

Edoid Degema Kari 2003

Kari 2004

Igboid Igbo Green & Igwe 1963

Nupoid Nupe Crowther 1864

Kwa Ewe Ameka 1996

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 and Tompo

ole Mpaayei 1955

Nubian Nubian (Dongolese) Armbruster 1960

von Massenbach

1963

Surmic Murle Lyth 1971

53

Kunama Kunama Bender 1996

Maban Masalit Leffel 2011

Saharan Kanuri Hutchison 1976

Cyffer 1998

Songhay Koyraboro Senni Heath 1999

Afro-

Asiatic

Biu-Mandara Tera Newman 1970

Masa Masa Caitucoli 1986

Eastern Cushitic Somali Saeed 1999

Southern

Cushitic

Iraqw Mous 1992

Omotic Maale Amha 2001

Semitic Arabic (Egyptian) Lucas 2009

Southeast Asia and Oceania

Austro-

Asiatic

Bahnaric Pacoh Alves 2006

Khasian Pnar (Khasi) Koshy 2009

Khmer Khmer Haiman 2010

Huffman 1970

Nicobarese Nicobarese (Car) Allan 1889

Das 1977

Braine 1970

Palaun-

Khmuic

Khmu’ Osborne 2009

Premsrirat 1987

Viet-Muong Vietnamese Nguyen & Nguyen

1997

Tran 2009

Austronesian Atayalic Seediq Asal 1953

54

Holmer 1996

Henningsson &

Holmer 2008

Central

Malayo-

Polynesian

Kambera Klamer 1998

Klamer 1994

Oceanic Maori Bauer 1997

South

Halmahera

West New

Guinea

Biak van den Heuvel 2006

Paiwanic Paiwan Egli 1990

Chamorro Chamorro Topping and Dungca

1973

Cooreman 1987

Meso-

Philippine

Tagalog Schachter & Otanes

1983

Sulawesi Tukang Besi Donohue 1999

Sundic Batak (Karo) Neumann 1922

Woollams 1996

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

55

Iroquoian Northern Iroquian [Oneida] Abbott 2000

Yuchi Linn 2000

Muskogean Muskogean Koasati Kimball et al. 1991

Kiowa Tamoan Kiowa Watkins 1984

Uto-Aztecan

Aztecan Nahuatl (South East

Puebla)

MacSwan 1999

Numic Comanche Charney 1993

Tepiman Pima Bajo Estrada Fernandez 1996

Wakashan Southern Wakashan Makah Davidson 2002

Salishan Bella Coola Davis and Saunders 1992

Davis and Saunders 1997

Interior Salish Lilloeet Davis & Caldecott 2007

Davis 2008

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 (Southeastern) ] Moshinsky 1974

Seri Moser and Marlett 2005

Yuman Maricopa Gordon 1986

Karok Bright 1957

Wappo-Yukian Wappo Thompson et al. 2006

Oto-

Manguean

Chinantecan Chinantec (Sochiapan) Foris 2000

Mixtecan Mixtec (Chalcatongo) Macaulay 1996

56

Otomian Otomí (Mezquital)

Hekking 1995

Popolocan Chocho Veerman 2000

Tarascan Purépecha Foster 1969

Chamoreau 2000

Totonacan Totonaca (Misantla)

MacKay 1999

Mixe-Zoque Zoque (Chiapas) Faarlund 2012

Huavean Huave Stairs and de Hollenbach 1981

Kim 2008

Mayan Mam Collins 1994

South America

Chibchan Aruak [Ika] Frank 1990

Paya [Pech] Holt 1999

[Rama]

Choco Choco Epena Pedee Harms 1994

Páezan Páez Jung 1989

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

57

Peba-Yaguan Yagua Payne and Payne 1990

Cahuapanan [Jebero] Bendor-Samuel 1961

Panoan Shipibo-Konibo Valenzuela 2003

Quechuan Quechua (Imbabura) Cole 1982

Aymaran Jaqaru Hardman 1966

Vaupés-Japurá Nadëb Weir 1994

Arawakan [Baré] Aikhenvald 1995

Cariban Apalaí Koehn and Koehn 1986

Tupian Tupari Mekens Galucio 2001

Tupi-Guarani Tapiete González 2005

Macro-Ge [Bororo] Huestis 1963

Crowell 1979

Ge-Kaingang Canela-Krahô Popjes and Popjes 1986

Trumai Guirardello 1999

Kwazá Van der Voort 2004

Chapacura-Wanhan Wari’ Everett and Kern 1997

Mura [Pirahã] Everett 1986

Arauan Paumarí Chapman and Derbyshire 1991

Katukinan [Canamarí] Groth 1988

Tacanan Araona Pitman 1980

Movima Haude 2006

Mosetenan Mosetén Sakel 2004

Uru-Chipaya Chipaya Cerrón-Palomino 2006

Guaicuruan Pilagá Vidal 2001

Araucanian Araucanian Mapudungun Smeets 1989

Zúñiga 2000

Chon Puelche Gününe Küne Casamiquela 1983

B. Austronesian languages with DN (based on Vossen & van der Auwera , this volume)

58

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