LINGUIST List 9.1331

Fri Sep 25 1998

Sum: Negative Complementizers

Editor for this issue: Martin Jacobsen <>


  1. Asya Pereltsvaig, Negative Complementizers

Message 1: Negative Complementizers

Date: Fri, 25 Sep 1998 10:48:25 -0400
From: Asya Pereltsvaig <>
Subject: Negative Complementizers

Dear Linguists,
On September 13 I posted the following query:

Does anyone know of languages that have negative complementizers? By
negative complementizers I mesan one of two things:

1) a complementizer that contains a negative element, like English
'unless' (= if not), 'lest' (=so that not) or

2) a complementizer that is selected by an "inherently negative" verb,
like 'enik' in Basque (according to Laka's (1990) dissertation) Any
information is greatly appreciated. I will post a summary.

In response I received about 30 replies. I greatly appreciate the help
of the following people (in no particular order):

Gisbert Fanselow
Ton van der Wouden
Lynn Santelmann
Ray Harlow
David B. Wharton
Stephane Goyette
Tim Stowell
Melanie Siegel
Dr. Antony Dubach Green
Torodd Kinn
Gyonggu Shin
Aldo Sevi
Julian Lloyd
Geoffrey Sampson
David Adger
Peter Svenonius
Daniel Collins
Chris Hogan
Nigel Duffield
Ross Clark
Jason Merchant
Mark Donohue
John Koontz
Rita Kreitman
Leslie Barrett

Several people mentioned LATIN.

Ton van der Wouden wrote:

Latin 'ne' comes to mind. You might want to take a look at our
analysis which was published in the SALT III Proceedings. Here is the

incollection {VdWouden-Zwarts:SALT3,
 title = {A Semantic Analysis of Negative Concord},
 author = {Ton van der Wouden and Frans Zwarts},
 year = 1993,
 booktitle = {SALT III: Proceedings of the third conference on
 semantics and linguistic theory},
 editor = {Utpal Lahiri and Adam Zachary Wyner},
 publisher = {Cornell University Department of Modern Languages and
 pages = {202-219}

Ray Harlow wrote:

I'm sure I am only one of very many who tell you that Latin has

complementisers in both your senses. In sense one, _ne_ 'lest, in
order that..not..' introduces subjunctive clauses with a variety of

negative final clauses, negative indirect commands, complements of
verbs of fearing (I am afraid, let it not happen = I am afraid it will
happen _timeo ne..._). Those indefinite pronouns in Latin which are
sensitive to polarity are in the "negative" form when following _ne_.
2. _quin_ introduces the complements of negated verbs of prevention,
among other things that it does. In general though, while itself
containing a negation, it occurs only after matrix clauses which are
negated or questioned. This is not exactly what you asked for, but is
arguably similar.

David B. Wharton wrote:

Latin has one, possibly two of these:

*ne* = (roughly "that...not"), to complementize negative commands,
statements of purpose, and sometimes just assertions.

Caesar mihi imperavit ne dicerem: "Caesar told me that I shouldn't

Miles fugit ne hostes occiderent: "the soldier ran away so that the
enemy wouldn't kill him."

Sometimes *ut* is used negatively after verbs of fearing:

vereor ut satis pecuniam habeam: "I fear I won't have enough money,"
but this is rare.

Later he added:

In reading your response, I see I made a grammatical error in one of
the sentences (not related to the complementizer) -- I'd hate to see
it go out on the Net uncorrected! The last example should read *satis

Stephane Goyette wrote:

I know of at least one language with a negative complementizer (In
sense 1 of your query), though somewhat lacking in native speakers:
Latin. The complementizer UT ("in order to") cannot be used if the
subordinate clause it introduces is negative: instead, one uses the
complementizer NE, which furthermore is inherently negative, i.e. a
subordinate clause introduced

by NE does not contain any other negative element (In fact, when
learning Latin in High school, we were taught that NE is basically a
contraction of UT and NON (the general negator), a description which
is untrue diachronically but not inapt, synchronically). To exemplify:

 "I see the/a man"

 "I don't see a/the man"

 "I came to see the man"

 "I came in order not to see the man"

 *VENI UT HOMINEM NON VIDEAM is quite ungrammatical. Since two
negatives cancel one another out in Latin, I suspect VENI NE HOMINEM
NON VIDEAM may have been possible -as a convuluted synonym of 3).

Chris Hogan wrote:

Latin has a negative complementizer of the first type: 'ne'. Examples
are (drawn from Allen & Greenough: "Latin Grammar"):

1. ne qua eius adventus procul significatio fiat
 "that no sign of his arrival may be made at a distance"

2. huic ne ubi consisteret quidem contra te locum reliquisti
 "you have left him no ground even to make a stand against you"

It also appears that it might have negative complementizers in the
second class, as Allen & Greenough note:

"Verbs denoting an effort to hinder take either (1) a Subjunctive

with quominus or ne, or (2) the infinitive: as, ---

 ne facerem impedivit
 "prevented me from doing"

Verbs of fearing take the Subjunctive, with ne affirmative and ne non
or ut negative. Thus,---

 timeo ne Verres fecerit
 "I fear that Verres has done", etc.

Note.---In this use ne is commonly to be translated by "that" or

and ne non by "that not"" (Allen & Greenough 1888, p. 361)

Another often mentioned language is IRISH.

Tim Stowell wrote:

I believe that Modern Irish has a negative complementizer; I would
advise you to check old articles by McCloskey; also I think I might
have mentioned it in an old paper I wrote on Irish raising back in the
mid 80s in NLLT.

Dr. Antony Dubach Green wrote:

Irish has a bunch of negative complementizers. Complementizers are
usually called "verbal particles" or "preverbal particles" in the
descriptive literature, but I'm pretty sure they're C0.

n = negative particle with nonpast tenses
 N chaitheann s tobac.
 NEG.NONPAST consume.PRES he tobacco
 'He doesn't smoke.'

nor = negative particle with past tenses
 Nor chaith m tobac.
 NEG.PAST consume.PAST I tobacco
 'I didn't smoke.'

nach(1) = 'that' + negative with nonpast tenses
 Deir s nach dtgfaidh s an chloch.
 say.PRES he that-not.NONPAST lift.FUT he the stone
 'He says that he won't lift the stone.'

nr(1) = 'that' + negative with past tenses
 Dirt s nr thg s an chloch.
 say.PAST he that-not.PAST lift.PAST he the stone
 'He said the he didn't lift the stone.'

nach(2) = negative interrogative particle with nonpast tenses
 Nach dtuigeann t?
 Q-NEG.NONPAST understand.PRES you(sg)
 'Don't you understand?'

nr(2) = negative interrogative particle with past tenses
 Nr irigh s?
 Q-NEG.PAST get-up.PAST she
 'Didn't she get up?'

nach(3) = relative pronoun + negative with nonpast tenses
 an cailn nach n-itheann arn
 the girl REL-NEG.NONPAST eat.PRES bread
 'the girl who doesn't eat bread'

nr(3) = relative pronoun + negative with past tenses
 an leabhar nr cheannaigh an fear
 the book REL-NEG.PAST buy.PAST the man
 'the book which the man didn't buy'

mura = 'if...not; unless' with nonpast tenses
 Mura gceannaonn siad , c bhfaigheann siad ?
 if-not.NONPAST buy.PRES they it where get.PRES they it
 'If they don't buy it, where do they get it?'

murar = 'if...not; unless' with past tenses
 Murar chaill s , ghoid s .
 if-not.PAST lose.PAST he it steal.PAST he it
 'If he didn't lose it, he stole it.'

n = negative imperative
 N bris !
 NEG.IMP break.IMP it
 'Don't break it!'

nr(4) = negative wish with subjunctive mood
 Nr fhille s!
 NEG.SUBJ return.SUBJ he
 'May he not return!'

David Adger wrote:

SCOTTISH GAELIC (and modern IRISH) have a negative complementisers in
your first sense:

nach - that not

(1) Tha mi a' smaoineachadh nach eil e an-sin
 Be I aspect think-VN that-not be he there
 "I think he's not there"

mur - if not

(2) Mur a bheil e an-sin, bidh mi mi-thoilichte
 If-not be he there, be-fut I unhappy
 "If he's not there, I'll be unhappy"

simple negation appears to be marked on the C position:

(3) Chan eil mi toilichte
 Not be I happy
 "I'm not happy"

assuming that the clause structure is CVSO. I think that Duffield in
his book on Irish argues that Neg is just below C and raises. You
might want to have a look at Paolo Acquaviva's paper in the Borsley
and Roberts 1996 volume on "the syntax of the Celtic languages" (CUP).

Notice that Laka's complementisers look more like polar elements, since
they're licensed by yes/no questions and in the if-clause of a
conditional. Josep Quer and I touch on this in a NELS paper a couple of
years ago (the NELS in Montreal, actually), and we have a longer paper
more or less finished which argues that -enik is just the complementiser

-en with an over polar determiner (something which Laka suggests in her

Peter Svenonius wrote:

Irish has "gan", a complementizer that expresses negation:
1. Ba mhaith liom gan e' a bheith a'balta e'iri'
 were good neg him to be able get.up
 'I would like him not to be able to get up'
2. Ba mhaith liom e' a bheith a'balta e'iri'
 were good him to be able get.up
 'I would like him to be able to get up'

(The apostrophes are meant to be accents; the examples are from my
1994 article in Studia Linguistica 48.2:142, but that article is about
selection, not about Irish or negative complementizers in general, so
you won't find much more information there.)

Nigel Duffield wrote:

Most of Chapter 2 of my Kluwer book "Particles and Projections in
Irish Syntax" is devoted to discussion of these particles. To cut 100
pages short, I conclude that the first 2 and the others that end in -r
are in T, the others are in C. See the book for further details and

Jason Merchant wrote:

I'm sure plenty of people have pointed you to Celtic, which has
various Cs inflected for tense and negation ('nach', 'nar').
McCloskey 1979 has a good survey. But Celtic may not be what you had
in mind.
 Lexical 'lest' occurs in Latin ('nisi') and ancient ('mipos')
and modern ('mipos', 'min') Greek. Some discussion of the latter can
be found in Giannakidou's diss (1997), the former in any grammar of
the ancient languages.

 Your 2nd question is more interesting: I don't know of any
other cases exactly like 'enik', but Adger and Quer had a NELS paper
(96 or 97) where they argued that certain kinds of embedded questions
really had polarity sensitive Cs; they're working on a new, bigger
version where they talk about Eng, Spanish, Catalan, and Basque, I
think. Quer may have some discussion in his diss (1998), but I'm not
sure. I'm a little hazy on the details, so it'd be best to email one
of them directly

A variety of other languages were mentioned as well.

Lynn Santelmann wrote:

I know that FINNISH has negative "agreement" on it's complementizers.
I don't have my Finnish grammar with me, so I can't come up with any
example, but hopefully, some native speakers of Finnish will respond
to your query and give you better examples than I ever could.

Ray Harlow said:

I may be the only one to pont out that MAORI (NZ's indigneous
Polynesian lg) has a negative cvomplementiser in your first sense
_kei_, which introduces clause with the sense, 'so that...not....'

Torodd Kinn wrote:

I don't know whether the term "complementizer" is appropriate, but New
Zealand Maori does have what is usually called a "verbal particle"

meaning "lest", "so that not". A warning: The difference between main
and subordinated clauses is rather unclear in Maori.

My source: Bruce Biggs: Let's Learn Maori. A Guide to the Study of the
Maori Language. Reed Education, Wellingtion. Revised edition: 1973.
P. 85: "The particle *kei* immediately preposed to a stative or a
universal indicates a warning or caveat. Frequently a clause beginning
with *kei* will follow a negative imperative beginning with *kaua*."
My remarks: A stative is an adjective/verb type word which cannot be
passivized. A universal is a verb type word which can be passivized.
Both types can also be used as nouns. I think *kaua* is a stative
meaning "not". Example, p. 85: "*Kaua e piki raakau kei taka koe!*
Don't climb trees lest you fall!" I think the literal translation is
"not (imperative-like) + verbal particle (non-past or subordination) +
climb + tree + lest + fall + you". I think it is correct to say that
*e piki raakau* is a nominal complement clause, subordinate to the
negative *kaua*. So the sentence begins rather like: "Not! that (you)
climb trees " Whether *kei taka koe* is subordinate, I can't tell.
Another example, p. 85: "*Kia aata koorero taatou kei rongo mai aku
hoa.* Let us talk quietly so that my companions do not hear." I try a
literal translation: "verbal particle (with desiderative meaning) + be
quiet + talk + we (more than two, including you) + lest + hear +
hither + my (plural possession) + companion". In *aata koorero* the
head is *aata*, so the meaning is something like "be quiet talk-wise".
Biggs provides more examples.

*************************************************** wrote
BIBLICAL HEBREW pen is very much like English
Yep, it's right there in the story of Adam and Eve
who are warned not to eat the apple "pen t'mutun"
("Lest ye die").

Melanie Siegel wrote:

maybe you mean something like shika in JAPANESE?:

ni-juu-hachi-nichi wa <P> gogo shika ai-te-i-nai N-desu kedomo

(28th TOPIC afternoon SHIKA not-be-free COPULA SENTENCE-PARTICLE)

(On the 28th, there is only the afternoon available.)

SHIKA combined with a negative Verb means ONLY.

Gyonggu Shin wrote:

KOREAN has two of what you wanted to get.

1. Negative words

moru - da
don't know declarative ending
ex: na-nun no-rul mor - un - da
 I-NOM You-ACC don't know tense declarative

mal - da
don't do delclarative

2. Negative complementizer

ga-ji mal-ara
go-Neg complementizer don't - imperative ending

ga-ji an-ass-da
go-Neg complementizer don't-past-declarative ending

******** ******************************************
Aldo Sevi wrote:

"lule", "ilmale" and "ela" in [Modern] HEBREW seems to be such things,
but I guess the term complementizer is not the right term, maybe

ilmale she-eyn ha-nashim yexolot le-hidamot le-mal?axim, hayiti medame
ota le-mal?ax elokim (Shay Agnon, the beginning of "Tehila")

ilmale that women can't resemble angels, I would resemble her to an
angel of god suggested glosses: "if women could resemble angels,
 "because women can't resemble....., I

ela (one of the uses of English 'but', seems also to need negation)

ani lo , ela balshan 'I'm not a philosopher, but a linguist' *ani
filosof, ela balshan ani filosof, aval (gam) balshan 'I'm a
philosopher, but (also) a linguist *ani filosof ela gam balshan

It seems "lule" or "luley" is morphologically composed from "lu"
(counterfactual if) and "lo" (not), , and it means the same as
"ilmale". Both are used only in a high register of Hebrew. (invented)

lule azarta li, lo hayiti matsliax 'if you hadn't helped me, I wouln't

************************************************ Julian Lloyd wrote:

XHOSA, (Guthrie S41), widely spoken in South Africa, seems to have the
complementiser in a negative form, coexistant of course with positive
forms! Hleze (lest...) is a combination of -hla (to descend, happen)
and the perfect form of -za (to go) and it occupies the normal
position in S' > Comp S.

Another negative complementiser structure, negates the regular comp,
derived from the infinitive form ukuba (to be)by adding the old
locative form, -phandle (outside), thus: ...ngaphandle kokuba
uyasebenza. (...unless he/she works.)

This latter is described in Du Plessis and Visser (1992:218).

Swahili uses isiwe (lest...) as a conjunction, but more common would
be a subjunctive construct; angalia usife (take care lest you die)
u=2Pers Sing, -si-=NEG Formative, -fe = Subj. Form of ukufa (to die)

One could presume that most Bantu languages would follow similar

Peter Svenonius wrote:

NORWEGIAN has "hvis ikke", literally, 'if not'
3. Han kommer hvis ikke det regner.
 he comes if not it rains
 'He'll come if it doesn't rain'

Note that "ikke" is not ordinarily clause-initial:
4. Han haaper at det ikke regner.
 he hopes that it not rains
 'He hopes that it won't rain'
5. * Han haaper at ikke det regner.
 he hopes that not it rains

This suggests that "hvis ikke" may be a complementizer even though
it's written as two words. Norwegian also has the expression "med
mindre", literally 'with smaller', also meaning 'unless', but it's
considered formal or literary. 

6. (formal) Han kommer med mindre det regner.
 he comes with smaller it rains
 'He'll come if it doesn't rain'

Neither "med" nor "mindre" can function as a complementizer in

Daniel Collins wrote:

There is a variety of negative subordinating conjunctions in SLAVIC
languages, e.g., Old Russian nezhe(li) 'than', where ne is a negative
morpheme; similarly neg"li/nek"li 'than,' Czech nez^, etc.

Ross Clark mentioned TONGAN:

Vave mai na'a ke to:mui.
hurry hither lest 2s late
"Hurry up or you'll be late"

Na'a is often glossed as "lest", and marks clauses referring to
possibilities to be avoided. Samoan and some other Polynesian
languages have similar particles.
Mark Donohue wrote:

TUKANG BESI (Austronesian, Indonesia) has the complementiser _bara_,
which in a main clause means 'don't', as in

Bara nu-manga='e
don't 2sg.r-eat=3obj
'DOn't eat it!'

but as a complementiser means 'lest':

No-wuju-'e [bara no-wila pe'esa-no]COMP.
3R-persuade-3OBJ lest 3R-go own-3POSS
'They persuaded him not to go on his own.'

Ku-ma'eka [bara na-w[um]ande-mo]COMP.
1SG-fear lest 3I-rain.SI-PF
'I'm worried that it might be about to rain.'

This same distribution is found in INdonesian with 'jangan' 'don't'
and jangan sampai 'lest'

John Koontz mentioned OMAHA-PONCA (Dhegiha Siouan):

\op e'gidhe e'=giz^aN =hnaN =tte
\tr beware you do thus regularly lest

 Beware lest you keep doing that!

\rf Dorsey 1890:64.8

The construction is e'gidhe ... =tte 'beware lest ...', in which =tte

the future or irrealis enclitic, and e'gidhe is 'finally; as

Here's another example, as part of the context of a construction
involving =kki=z^i 'if NEG' in which the NEG =z^i doesn't seem to have
force. I've been puzzling over these (not very hard, I admit) off and
on for several


\op "E'gidhe dha?aNhe=tte!"
\tr Beware you flee lest (warns the young buffalo bull)

\op "AN'kk=az^i ha, ttigaN' =ha!"
\tr AUX NEG ASSERT grandfather VOC,

\op "a'=xt=aN a?aN'he=tta=daN,"
\tr How on earth I flee FUT CONTINGENT

\op a'=bi =ama MiN'kkasi=akha.
\tr he said QUOTE Coyote the

\op Adha'=b= e=gaN e=di'=thaN ie'naxidha agi'=bi= ama.
\tr He went away having thence to charge him he was coming QUOTE

\op E'=di akhi'=bi= kki=z^i, MiN'kkasi zha'=ha=bi= ama.<===
\tr There he arrived back when, Coyote he gored him QUOTE

\op MaNs^i'=aha adha' =idhe=dha=bi=kki=z^i, <===
\tr Sky direction he went he sent him when,

\op gat?(e)' =ihe gagha'=bi =ama.
\tr he killed by the fall he lay he made him QUOTE.

\rf jod 1890:105.2-5

Let me know if this isn't clear, or if you have a clue on why 'if-not'
= 'if, when'? I imagine it has something to do with unexpectedness,
but I suspect that is a relatively unsophisticated assessment of some
sort of irrealis or counterfactual.

Leslie Barrett has also mentioned ROMANSCH.

Thanks again to everybody who responded to my query.
Best wishes,
Asya Pereltsvaig
Department of Linguistics - McGill University
1001 Sherbrooke St. West
Montreal, Quebec, H3A 1G5, CANADA

Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue