Date post: | 11-Nov-2023 |
Category: |
Documents |
Upload: | independent |
View: | 1 times |
Download: | 0 times |
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
THE SYNTAX OF WELSH NEGATION1
By Robert D. Borsley and Bob Morris JonesUniversity of Essex University of Wales, Aberystwyth
Abstract
Colloquial Welsh negation involves many complexities. Nega-
tive sentences require an n-word in subject or object position
or as a post-subject adverb. The picture is complicated by the
fact that the main negative adverb is homophonous with a
negative quantifier. Further, there are differences between
negative subjects and negative objects, and between different
negative adverbs. Non-finite negation often involves a nega-
tive verb, but it seems somewhat simpler than finite negation.
It allows, but does not require, an n-word. The main facts can
be accommodated fairly readily within Head-driven Phrase
Structure Grammar (HPSG). Within this framework non-
finite negation can be analysed in terms of a small number
of negative heads, finite negation, in terms of a number of
types of negative verb.
1. Introduction
Negation has been a major concern in syntactic theory since the
late 1980s. There has been extensive discussion of negation in
English and French, and other Germanic and Romance languages.
However, there has been little work on negation in the Celtic
Transactions of the Philological Society Volume 98:1 (2000) 15±47
# The Philological Society 2000. Published by Blackwell Publishers,108 Cowley Road, Oxford OX4 1JF and 350 Main Street, Malden, MA 02148, USA.
1 An earlier version of this paper was presented at the Negation: Syntax, Semanticsand Pragmatics Conference at the University of Salford, October 29±November 1,1998, and some of the material included here was presented at UMIST in February1999, at the Poznan Linguistics Meeting in April 1999, at the 6th Welsh SyntaxSeminar at Gregynog in June 1999, and at the HPSG Conference in Edinburgh inAugust 1999. The research on which the paper is based was largely carried out at theGrapes Hotel in Maentwrog, Gwynedd. We are grateful to Paul Rowlett and twoanonymous referees for a number of helpful comments. Any imperfections are ourresponsibility.
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
languages and very little work on negation in Welsh.2 Jones and
Thomas (1977, IX, 4) provide some discussion within an early
transformational framework, Awbery (1990) discusses the distinc-
tive properties of negation in Pembrokeshire Welsh within a similar
framework, and Rouveret (1994, 2.4.3) discusses the basic facts
within a version of Principles-and-Parameters theory (P&P). How-
ever, these works just scratch the surface of a complex array of data.
In this paper, we will look at all the main types of negative
construction in Welsh, identify their central properties, and develop
an analyis within Head-driven Phrase Structure Grammar (HPSG),
drawing on work on negation and related phenomena in English
and French in Kim (1995), Kim and Sag (1996), and Warner
(forthcoming).
There are considerable differences between formal literary
Welsh and colloquial Welsh. Some work has concentrated on the
former and some on the latter. We will concentrate on colloquial
Welsh, but we will also refer briefly to the literary language.3 Our
focus will be on the syntax of negation. Inevitably we will make
some reference to meaning, but we will not offer any semantic
analyses.
The paper is organised as follows. In section 1, we begin by
presenting some basic data. Then, in section 2, we look at the role of
a negative quantifier dim in negative sentences, and in section 3, we
consider the role of a homophonous negative adverb. In section 4,
we consider non-finite and imperative negation. Then, in section 5,
we outline an HPSG analysis of the data. Finally, in section 6, we
offer some concluding remarks.
2. Some basic data
In literary Welsh, negative sentences commonly contain only a pre-
verbal negative particle. Thus, corresponding to the affirmative
16 transactions of the philological society 98, 2000
2 Acquaviva (1996) provides an interesting discussion of the rather differentnegative facts in Irish.
3 In the examples from colloquial Welsh, some of the spellings are modified toreflect spoken forms (e.g. llyfre and welish, instead of literary llyfrau and welais).
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
examples in (1), both of which are possible in literary Welsh, we
have the negative example in (2).4
(1) a. Fe gaf sefyll yma.
PRT can stand here
`I can stand here.'
b. Caf sefyll yma.
can stand here
`I can stand here.'
(2) Ni chaf sefyll yma.
NEG can stand here
`I can't stand here.'
Gaf in (1a) shows what is known as soft mutation, while chaf in (2)
shows so-called aspirate mutation. Caf in (1b) is the basic form. All
three examples show that literary Welsh is a null-subject language.
Negative sentences in literary Welsh may also contain a post-verbal
negative element or n-word, as (3) illustrates:
(3) Nid oes neb yn yr ystafell.
NEG is no one in the room
`There is no one in the room.'
However, there is no need for negative sentences to contain an n-
word. The situation is different in colloquial Welsh.
In colloquial Welsh, the preverbal negative particle does
not normally appear.5 However, many verbs show the mutation
17borsley and jones ± the syntax of welsh negation
4 The traditional literary style favours the use of a preverbal negative particle, only,but grammars which focus on literary Welsh (e.g. Thorne 1993: 219 and Thomas1996: 540) note that negative sentences can also contain a postverbal negative marker.Thus (i) is an alternative to (2).
(i) Ni chaf ddim sefyll yma.NEG can NEG stand here`I can't stand here.'
5 However, subordinate clauses are often introduced by na(d). We have exampleslike the following:
(i) Dw i 'n gwbod [na fydd Mair yna]am I in know NEG will-be Mair there`I know Mair won't be there.'
In addition, some southern dialects have a preverbal negative marker na(g).
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
which is traditionally attributed to the negative particle, and some
forms of the copula show a remnant of the negative particle in the
form of an initial d. The following are representative negative
sentences:
(4) Does neb yn yr ystafell.
is no one in the room
`There is no one in the room.'
(5) Welish i neb.
saw-1SG I no one
`I saw no one.'
(6) Cha' i ddim sefyll yma.
can I NEG stand here
`I can't stand here.'
(7) Dydy Mair byth yn gweithio.
is Mair never in work
`Mair never works.'
(4) and (6) correspond to the literary Welsh examples in (3) and (2),
respectively. (5) and (6) have overt pronominal subjects, reflecting
the fact that colloquial Welsh is not a null-subject language. The
following are related affirmative sentences:
(8) Mae Gwyn yn yr ystafell.
is Gwyn in the room
`Gwyn is in the room.'
(9) Welish i Gwyn.
saw-1SG I Gwyn
`I saw Gwyn.'
(10) Ga' i sefyll yma.
can I stand here
`I can stand here.'
(11) Mae Mair yn gweithio.
is Mair in work
`Mair is working.'
18 transactions of the philological society 98, 2000
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
As these examples show, some of the negative sentences contain a
distinctive verb form. All contain an n-word. As we will see, all
negative sentences contain an n-word in colloquial Welsh.
Not surprisingly, perhaps, the n-words in (4)±(7) can appear in
other positions, as the following illustrate:
(12) Dw i ddim wedi gweld neb.
am I NEG after see no one
`I haven't seen anybody.'
(13) Dydy hi ddim wedi newid dim.
is she NEG after change NEG
`She has not changed at all.'
(14) Dydy Mair ddim yn gweithio byth.
is Mair NEG in work never
`Mair never works.'
In (12), neb appears as object of a non-finite verb. In (13), dim, the
basic, unmutated form of ddim, follows a non-finite verb. (We will
refer to this element as dim in subsequent discussion.) In (14), byth
appears after a non-finite verb. Notice that all these examples
contain two n-words, but have just a simple negative interpretation.
Thus, like many languages, colloquial Welsh has negative concord.
The final n-words in (12) and (13) can be replaced by non-negative
elements with no change in truth conditional content, as in the
following:
(15) Dw i ddim wedi gweld unrhyw un.
am I NEG after see anyone
`I haven't seen anybody.'
(16) Dydy hi ddim wedi newid o gwbl.
is she NEG after change at all
`She has not changed at all.'
(12) and (13) are essentially emphatic versions of (15) and (16).
Notice now that it is not possible to replace the n-words in (4) and
(5) by a non-negative element. Thus, the following examples are
ungrammatical:
19borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(17) *Does unrhyw un yn yr ystafell.
is anyone in the room
(18) *Welish i unrhyw un.
saw-1SG I anyone
Notice also that it is not possible to omit post-subject dim from (12)±
(14).
(19) *Dw i wedi gweld neb.
am I after see nobody
(20) *Dydy hi wedi newid dim.
is she after change NEG
(21) *Dydy Mair yn gweithio byth.
is Mair in work never
It seems, then, that negative sentences must contain an n-word in
close proximity to the verb, more precisely in subject or object
position, or as a post-subject adverb.6
3. Quantifier dim
A number of the examples that we considered in the last section
contain the negative adverb dim. An important complication in
Welsh negation is the existence of a homophonous negative quanti-
fier. It seems, however, that Welsh is not unique here. Rowlett
(1996, 1998) argues that French pas is both an adverb and a
quantifier. It looks, then, as if we have an interesting similarity
between Welsh and French here.
Quantifier dim appears in examples like the following, which we
assume have the structure indicated:7
20 transactions of the philological society 98, 2000
6 Other n-words are dim byd `nothing/anything' (literally `no world') and erioed,which, like byth, translates as `never/ever'. Byth is used in non-perfective contexts,while erioed is used in perfective contexts.
7 Jones and Thomas (1977: 324) propose that dim is a quantifier in examples like(22) and Jones and Thomas (1977: 363, fn.5) propose that it is a quantifier in exampleslike (25). Awbery (1990) suggests that there is a single dim which is somehow absorbedinto the following NP in examples like these. This approach is incompatible with mostcurrent frameworks, and it is not clear how it could account for the full range of data.
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(22) Does [dim dyn] yn yr ystafell.
is NEG man in the room
`There is no man in the room.'
(23) Does [dim o 'r dynion] yn yr ystafell.
is NEG of the men in the room
`None of the men is in the room.'
(24) Welish i [ddim dyn]
saw-1SG I NEG man
`I saw no man.'
(25) Welish i [ddim o 'r dynion]
saw-1SG I NEG of the men
`I saw none of the men.'
In other words, we assume that these examples are rather like (4)
and (5), except that they involve complex negative subjects and
objects.
On the analysis that we are proposing, (22)±(25) are rather like the
following examples containing the quantifiers rhai and peth, both of
which can be glossed as `some':
(26) a. Mae rhai dynion yn yr ystafell.
is some men in the room
`There are some men in the room.'
b. Mae peth siwgr yn y cwpan.
is some sugar in the cup
`There is sugar in the cup.'
(27) a. Mae rhai o 'r dynion yn yr ystafell.
is some of the men in the room
`Some of the men are in the room.'
b. Mae peth o 'r siwgr yn y cwpan.
is some of the sugar in the cup
`Some of the sugar is in the cup.'
(28) a. Welish i rai dynion.
saw-1SG I some men
`I saw some men.'
21borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
b. Welish i beth siwgr.
saw-1SG I some sugar
`I saw some sugar.'
(29) a. Welish i rai o 'r dynion.
saw-1SG I some of the men
`I saw some of the men.'
b. Welish i beth o 'r siwgr.
saw-1SG I some of the sugar
`I saw some of the sugar.'
The analysis provides us with a straightforward explanation for the
ungrammaticality of the following examples:
(30) *Does dim y dyn yn yr ystafell.
is NEG the man in the room
(31) *Does dim o ddynion yn yr ystafell.
is NEG of men in the room
(32) *Welish i ddim y dyn.
saw-1SG I NEG the man
(33) *Welish i ddim o ddynion.
saw-1SG I NEG of men
These examples are ungrammatical for exactly the same reason as
the following:
(34) a. *Mae rhai y dynion yn yr ystafell.
is some the men in the room
b. *Mae peth y siwgr yn y cwpan.
is some the sugar in the cup
(35) a. *Mae rhai o ddynion yn yr ystafell.
is some of men in the room
b. *Mae peth o siwgr yn y cwpan.
is some of sugar in the cup
(36) a. *Welish i rai y dynion.
saw-1SG I some the men
b. *Welish i beth y siwgr.
saw-1SG I some the sugar
22 transactions of the philological society 98, 2000
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(37) a. *Welish i rai o ddynion.
saw-1SG I some of men
b. *Welish i beth o siwgr.
saw-1SG I some of sugar
We assume that all three quantifiers take as complements either a
bare indefinite NP or a PP containing o and a definite NP.8 We will
see in the next section, however, that there is more to be said about
examples like (32).
On the analysis that we are proposing, (22)±(25) contain quanti-
fier phrases headed by dim, either as subject or object of a finite
verb. Some evidence that this is indeed the right analysis comes from
the fact that we find the same strings in other NP positions. The
following show that they can appear as the object of a non-finite
verb:
(38) Dw i ddim wedi prynu dim llyfr.
am I NEG after buy NEG book
`I haven't bought any book.'
(39) Dw i ddim wedi prynu dim o 'r llyfre.
am I NEG after buy NEG of the books
`I haven't bought any of the books.'
The following show that they can also appear as object of a
preposition:
(40) Dydy Sioned ddim yn chwilio am ddim llyfre.
is Sioned NEG in look for NEG books
`Sioned is not looking for any books.'
(41) Dydy Sioned ddim yn chwilio am ddim o 'r llyfre.
is Sioned NEG in look for NEG of the books
`Sioned is not looking for any of the books.'
The following show that we also have these strings as the subject of a
non-finite clause:
23borsley and jones ± the syntax of welsh negation
8 There are some differences between the three quantifiers. Peth only allows an NPcomplement headed by a mass noun, and rhai only alows a plural NP complement.Dim allows any kind of NP complement.
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(42) Dydy Gwyn ddim yn disgwyl i ddim bechgyn fynd adre.
is Gwyn NEG in expect for NEG boys go home
`Gwyn didn't expect any boys to go home.'
(43) Dydy Gwyn ddim yn disgwyl i ddim o 'r bechgyn
is Gwyn NEG in expect for NEG of the boys
fynd adre.
go home
`Gwyn didn't expect any of the boys to go home.'
Examples like (38)±(43) provide strong evidence that dim NP and
dim o NP are constituents.
Some further support for this analysis comes from the fact that
dim cannot be separated from NP or o NP by an adverbial element.
Consider the following examples:
(44) a. Wela' i ddim pysgod bob tro.
will-see-1SG I NEG fish each turn
`I won't always see a fish.'
b. *Wela' i ddim bob tro bysgod.
will-see-1SG I NEG each turn fish
(45) a. Wela' i ddim o 'r pysgod bob tro.
will-see.1SG I NEG of the fish each turn
`I won't always see any of the fish.'
b. *Wela' i ddim bob tro o 'r bysgod.
will-see-1SG I NEG each turn of the fish
Here, in the (a) examples, the adverbial bob tro follows the NP, but
in the (b) examples it follows dim, and the result is ungrammati-
cality. If dim NP and dim o NP were both two separate constituents,
we would expect it to be possible for an adverbial element to split
them up. Where the subject is clearly followed by two separate
constituents, they can be separated by an adverbial, as the following
show:
(46) a. (Mi) roddith Mair arian i 'r capel bob tro.
PRT will-give Mair money to the chapel each turn
`Mair will always give money to the chapel.'
24 transactions of the philological society 98, 2000
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
b. (Mi) roddith Mair arian bob tro i 'r capel.
PRT will-give Mair money each turn to the chapel
On the other hand, an adverbial cannot separate the quantifiers rhai
and peth from their following NP or PP.
(47) a. Helpa' i rai achosion bob tro.
will-help-1SG I some causes each turn
`I'll help some causes every time.'
b. *Helpa' i rai bob tro achosion.
will-help-1SG I some each turn causes
(48) a. Helpa' i rai o 'r achosion bob tro.
will-help-1SG I some of the causes each turn
`I'll help some of the causes every time.'
b. *Helpa' i rai bob tro o 'r achosion.
will-help-1SG I some each turn of the causes
(49) a. Gyma' i beth siwgr bob tro.
will-take-1SG I some sugar each turn
`I'll take some sugar every time.'
b. *Gyma' i beth bob tro siwgr.
will-take-1SG I some each turn sugar
(50) a. Gyma' i beth o 'r siwgr bob tro.
will-take-1SG I some of the sugar each turn
`I will take some of the sugar every time.'
b. *Gyma' i beth bob tro o 'r siwgr.
will-take-1SG I some each turn of the sugar
Thus, we have further evidence that dim NP and dim o NP are
constituents.9
25borsley and jones ± the syntax of welsh negation
9 Rouveret (1994: 134) suggests that examples like the following provide evidencethat dim NP is a constituent:
(i) Dim Gwyn welish i.NEG Gwyn saw-1SG I`It wasn't Gwyn that I saw.'
One might suppose that dim Gwyn has been fronted, which would only be possible if itis a constituent. Notice, however, that we do not have this sequence in object position.
(ii) *Welish i ddim Gwyn.saw-1SG I NEG Gwyn`I didn't see Gwyn.'
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
It seems, then, that sentences with a negative subject or object
containing the quantifier dim are an important feature of negation
in colloquial Welsh. An obvious question to ask here is whether
there are any differences between subjects and objects containing
dim. In many dialects, the answer is yes. In many dialects, the object
of o can be a singular count NP within an object but not within a
subject. Thus, we have the following contrast:10
(51) Welodd Mair ddim o 'r ceffyl.
saw Mair NEG of the horse
`Mair didn't see the horse.'
(52) *Does dim o 'r ceffyl yn y cae.
is NEG of the horse in the field
`No horse is in the field.'
Obviously, a satisfactory analysis must accommodate this contrast.
4. Adverbial dim
Having looked at quantifier dim, we want now to look more closely
at adverbial dim.
Adverbial dim precedes certain predicative elements. Hence, one
might propose that it is a pre-modifier, left-adjoined to such
elements. There are, however, various kinds of evidence against
such an analysis. If adverbial dim were a pre-modifier of the
following constituent, one would expect dim VP to be possible
wherever VP is possible, but it is not. Consider the following:
26 transactions of the philological society 98, 2000
Only the following is possible:
(iii) Welish i ddim o Gwyn.saw-1SG I NEG of Gwyn`I didn't see Gwyn.'
We propose that dim is a negative complementiser in (i) and that only Gwyn has beenfronted. Some support for this analysis comes from the fact that literary Welshusually has not dim but nid in such examples.
(iv) Nid Gwyn welish iNEG Gwyn saw-1SG I`It wasn't Gwyn that I saw.'
10 Examples like (52) are grammatical in some southern dialects, but not in theWelsh of North and Mid-Wales.
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(53) *Dydw i ddim yn disgwyl i Mair ddim mynd i Aberystwyth.
am I NEG in expect to Mair NEG go to Aberystwyth
`I don't expect Mair to go to Aberystwyth.
(54) *Dydy Gwyn ddim yn degyg o ddim mynd i Aberystwyth.
is Gwyn NEG in likely of NEG go to Aberystwyth
`Gwyn isn't likely to go to Aberystwyth.'
(55) *Dydy Gwyn ddim yn awyddus i ddim mynd i Aberystwyth.
is Gwyn NEG in eager to NEG go to Aberystwyth
`Gwyn isn't eager to go to Aberystwyth.'
In these examples, we have adverbial dim before a VP which is
something other than the complement of a finite verb. In (53) it is
the predicate of a non-finite clause, and in (54) and (55) it follows
elements which Borsley (1986) analyses as prepositional comple-
mentisers. Such examples contrast with earlier ones showing that
quantifier dim can appear in any NP position. Their ungrammati-
cality suggests quite strongly that adverbial dim is not a pre-modifier.
Further evidence that adverbial dim is not a pre-modifier comes
from the fact that it can be separated by another adverbial from the
following material. The following illustrate:
(56) Cheisiodd Gwyn ddim bob tro ateb y cwestiwn.
tried Gwyn NEG each turn answer the question.
`Gwyn didn't try to answer the question every time.'
(57) Dydy 'r ceffyl ddim bob tro yn y cae.
is the horse NEG each turn in the field
`The horse isn't always in the field.'
(58) Dydy Gwyn ddim bob tro wedi mynd i Aberystwyth.
is Gwyn NEG each turn after go to Aberystwyth
`Gwyn hasn't always gone to Aberystwyth.'
As we saw earlier, quantifier dim cannot be separated from the
following NP or PP. Thus, it seems that adverbial dim, unlike
quantifier dim, does not form a constituent with the following
material.
A final piece of evidence that adverbial dim is not a pre-modifier
comes from ellipsis. It is possible to omit the material that follows
27borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
adverbial dim, but it is not possible to omit the complement of
quantifier dim.
(59) Welodd Sioned y car, ond nesh i ddim (ei weld o)
saw Sioned the car but did-1SG I NEG 3SGM see he
`Sioned saw the car, but I didn't see it.'
(60) Welodd Sioned y car, ond welish i ddim *(ohono fo).
saw Sioned the car but saw-1SG I NEG of-3SGM he
`Sioned saw the car, but I didn't see it.'
In (59), the material following adverbial dim can be omitted, but this
is not possible with the complement of quantifier dim in (60). It is
worth noting here that Kim and Sag (1996) point out that the
material following English not can be omitted where not immedi-
ately follows an auxiliary. They cite examples like the following:
(61) Although you want to have another cookie, you may not.
They conclude from this and other data that this not is not a pre-
modifier of the following material. It seems that adverbial dim is
similar.
It seems, then, that there is considerable evidence that adverbial
dim is not a pre-modifier, but an extra constituent of sentences in
which it occurs. Hence, we assume that an example like (6) has the
following schematic structure:
(62) Cha' i [Adv ddim] [VP sefyll yma]
We assume in fact that dim is an extra complement. Essentially the
same analysis is proposed for certain uses of English not in Kim
(1995), Kim and Sag (1996) and Warner (forthcoming), and for
certain uses of French pas in Kim (1995) and Kim and Sag (1996).
Some independent evidence that this is the right analysis for
adverbial dim comes from the fact that it is mutated. Borsley
(forthcoming) argues that a number of instances of mutation in
Welsh are the result of a rule which assigns mutation to a
complement immediately preceded by a phrasal sister. This rule
will account for the mutation of adverbial dim, provided it is a
complement.
We now look briefly at the semantically more complex negative
28 transactions of the philological society 98, 2000
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
adverb byth. The ungrammaticality of the following suggests that it
occupies the same position as adverbial dim:
(63) *Dw i byth ddim yn yfed cwrw.
am I ever NEG in drink beer
`I never drink beer.'
This suggests that it, too, must be an extra constituent of the
sentences in which it occurs, and, hence, that (7) has the following
schematic structure:
(64) Dydy Mair [Adv byth] [AspP yn gweithio]
Although byth is rather like adverbial dim, there are some important
differences. First, neb is not possible as a subject before adverbial
dim, but it is before byth.
(65) *Fydd neb ddim yma.
will-be no one NEG here
`No one will be here.'
(66) Fydd neb byth yma.
will-be no one never here
`No one will ever be here.'
Second, adverbial dim cannot be followed by an object NP, but byth
can.
(67) *Wela' i ddim Gwen eto.
will-see-1SG I NEG Gwen still
`I won't see Gwen again.'
(68) Wela' i byth Gwen eto.
will-see-1SG I never Gwen still
`I will never see Gwen again.'
Thus, adverbial dim is subject to two restrictions which do not apply
to byth. Clearly, this is an important matter.
There is one further matter that we should mention here. The
incompatibility of adverbial dim and a following object NP means
that negative sentences containing a transitive verb will commonly
have a negative object. Thus, (24), repeated here, is a typical
example:
29borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(24) Welish i ddim dyn.
saw-1SG I NEG man
`I saw no man.'
However, such sentences can also have a negative subject, as in (69):
(69) Welodd neb lyfr.
saw no one book
`No one saw any book.'
They can also have both, as in (70):
(70) Welodd neb ddim llyfr.
saw no one NEG book
`No one saw any book.'
Notice also that, unlike adverbial dim, quantifier dim can be
preceded by a negative subject.
5. Non-finite negation
We turn now to non-finite negation, including imperative negation.
We will see that there is a sense in which this is simpler than finite
negation.
We look first at simple non-finite negation. Here, corresponding
to the affirmative examples in (71a) and (72a), we have the negative
examples in (71b) and (72b).
(71) a. (Mi) geisiodd Gwyn ateb y cwestiwn.
PRT tried Gwyn answer the question
`Gwyn tried to answer the question.'
b. (Mi) geisiodd Gwyn beidio (ag) ateb y cwestiwn.
PRT tried Gwyn NEG with answer the question
`Gwyn tried not to answer the question.'
(72) a. Dw i 'n disgwyl i Mair fynd i Aberystwyth.
am I in expect to Mair go to Aberystwyth
`I expect Mair to go to Aberystwyth.'
b. Dw i 'n disgwyl i Mair beidio (aÃ) mynd i Aberystwyth.
am I in expect to Mair NEG with go to Aberystwyth
`I expect Mair not to go to Aberystwyth.'
30 transactions of the philological society 98, 2000
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
The examples in (71) contain a superficially subjectless non-finite
complement, and the examples in (72) contain a non-finite comple-
ment with an overt subject. The (b) examples contain a negative
element, which we will refer to as peidio, since this is its basic form.
This is standardly analysed as a non-finite verb, and we see no
reason to question this analysis. Peidio is optionally followed by the
preposition aà `with' (ag before a vowel).
The use of peidio to express negation within a non-finite con-
stituent leads to some interesting pairs of sentences. Consider, for
example, the following:
(73) a. Cha i ddim sefyll yma.
can I NEG stand here
`I can't stand here.'
b. (Mi) ga' i beidio sefyll yma.
PRT can I NEG stand here
`I can not stand here.'
In (73a), which appeared earlier as (6), the modal is within the scope
of negation. In (73b), the negation is within the scope of the modal.
We can now consider negative imperatives. Here, corresponding
to the affirmative imperatives in (74a) and (75a), we have the
negative imperatives in (74b) and (75b).11
(74) a. Ateba y cwestiwn.
answer-SG the question.
`Answer the question.'
b. Paid (ag) ateb y cwestiwn.
NEG-SG with answer the question.
`Don't answer the question.'
31borsley and jones ± the syntax of welsh negation
11 Very literary Welsh has negative imperatives of the following form:
(i) Nac ateba y cwestiwn.NEG answer-SG the question`Don't answer the question.'
(ii) Nac atebwch y cwestiwn.NEG answer-PL the question`Don't answer the question.'
But even in literary Welsh, paid and peidiwch are used.
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(75) a. Atebwch y cwestiwn.
answer-PL the question
`Answer the question.'
b. Peidiwch (ag) ateb y cwestiwn.
NEG-PL with answer the question.
`Don't answer the question.'
The examples in (74) are singular imperatives, while those in (75) are
plural. The negative versions involve inflected forms of peidio and
provide important evidence that this is a verbal element.
There is one further type of non-finite negation that we should
mention here, illustrated by the following:
(76) Mae Gwyn heb fynd i Aberystwyth.
is Gwyn without go to Aberystwyth
`Gwyn has not gone to Aberystwyth.'
Heb is homophonous with a preposition ± hence the gloss ± but we
assume that it is a member of a class of aspect markers in this use.
(76) has essentially the same meaning as the following, where wedi is
another aspect marker:
(77) Dydy Gwyn ddim wedi mynd i Aberystwyth.
is Gwyn NEG after go to Aberystwyth
`Gwyn has not gone to Aberystwyth.'
Unlike (77), (76) contains the basic form of the copula, which
appears in (8) and (11).12 Hence, while we regard (77) as a negative
sentence, we regard (76) as an affirmative sentence containing a
non-finite negative constituent.
An important fact about the various non-finite negative contexts
is that they may contain n-words. Thus, we have examples like the
following:
32 transactions of the philological society 98, 2000
12 Some speakers allow sentences like the following:
(i) Dydy Gwyn heb fynd i Aberystwyth.is Gwyn without go to Aberystwyth`Gwyn has not gone to Aberystwyth.'
However, a traditional colloquial style favours (77).
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(78) (Mi) geisiodd Gwyn beidio (aÃ) gweld neb.
PRT tried Gwyn NEG with see no one
`Gwyn tried not to see anyone.'
(79) (Mi) geisiodd Gwyn beidio (aÃ) newid dim.
PRT tried Gwyn NEG with change NEG
`Gwyn tried not to change (at all).'
(80) (Mi) geisiodd Gwyn beidio (aÃ) darllen dim llyfr.
PRT tried Gwyn NEG with read NEG book
`Gwyn tried not to read any book.'
(81) Paid/ peidiwch (aÃ) gweld neb.
NEG-SG NEG-PL with see no one
`Don't see anyone.'
(82) Paid/ peidiwch (aÃ) newid dim.
NEG-SG NEG-PL with change NEG
`Don't change (at all).'
(83) Paid/ peidiwch (aÃ) darllen dim llyfr.
NEG-SG NEG-PL with read NEG book
`Don't read any book.'
(84) Mae Gwyn heb weld neb.
is Gwyn without see no one
`Gwyn has not seen anyone.'
(85) Mae Gwyn heb newid dim.
is Gwyn without change NEG
`Gwyn has not changed (at all).'
(86) Mae Gwyn heb ddarllen dim llyfr.
is Gwyn without read NEG book
`Gwyn has not read any book.'
As we might expect, all these examples are ungrammatical if the
negative elements are not present.
(87) *(Mi) geisiodd Gwyn weld neb.
PRT tried Gwyn see no one
(88) *(Mi) geisiodd Gwyn newid dim.
PRT tried Gwyn change NEG
33borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(89) *Newid/ newidwch dim.
change-SG change-PL NEG
(90) *Wel/ welwch neb.
see-SG see-PL no one
(91) *(Mi) geisiodd Gwyn ddarllen dim llyfr.
PRT tried Gwyn read NEG book
(92) *Darllen/ darllenwch dim llyfr.
read-SG read-PL NEG book
(93) *Mae Gwyn wedi gweld neb.
is Gwyn after see no one
(94) *Mae Gwyn wedi newid dim.
is Gwyn after change NEG
(95) *Mae Gwyn wedi ddarllen dim llyfr.
is Gwyn after read NEG book
Thus, non-finite negation allows an n-word, but it does not
require one. In this it is like finite negation in many languages.
Consider, for example, Polish, which has examples like the following:
(96) Nie widziaøem nikogo.
NEG saw-1SG nobody
`I didn't see anyone.'
(97) Nie widziaøem Jana.
NEG saw-1SG Jan
`I didn't see Jan.'
(96) shows that finite negation in Polish allows an n-word. (97)
shows that it does not require one. We saw in section 2 that finite
negation in Welsh not only allows, but also requires, an n-word.
Thus, it is rather more complex than non-finite negation in Welsh
and finite negation in many other languages.
6. An hpsg analysis
We will now develop a detailed analysis of the data that we have
presented. As we indicated earlier, Rouveret (1994) sketches an
34 transactions of the philological society 98, 2000
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
analysis of some aspects of Welsh negation within the P&P frame-
work. It is not clear to us how the full range of facts could be
accommodated within P&P. However, we will not pursue this
question. Instead, we will develop an analysis within the Head-
driven Phrase Structure Grammar (HPSG) framework of Pollard
and Sag (1994). We will try to show that the framework permits a
fairly straightforward account of the data.
HPSG is a monostratal theory, in which the syntactic structure of
a sentence is a single, relatively simple, constituent structure. It
is head-driven in the sense that heads contain information
about the non-heads that they combine with. The information is
encoded in certain `valence' features, whose value is a list of synsem
objects, combinations of syntactic and semantic information. The
main ones are the COMPS feature, which generally indicates what
complements a head takes, and the SUBJ feature, which generally
indicates what subject a head requires. For HPSG, a simple
transitive verb like chased will have the following category:
(98) verbSUBJ hNPiCOMPS hNPi
" #
Such categories interact with a small number of immediate dom-
inance and linear precedence rules and general conditions on local
structures. In the present context, the most important immediate
dominance rule is the Head-Complement Rule, which we can
formulate as follows:
(99) Head-Complement Rule
[COMPS h i] ! [COMPS L], [ ]*
HEAD COMPS
`L' here stands for an arbitrary list and `[ ]' for an arbitrary list
member. The rule can be paraphrased as follows:
(100) A sign with the feature specification COMPS h i can have a
head daughter with the feature specification COMPS L and
any number of complement daughters.
Nothing in the rule ensures that we have the right kind of head and
the correct number and kind of complements. The former is ensured
35borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
by the Head Feature Principle, the latter by the Valence Principle.
We can formulate these principles as follows:
(101) Head Feature Principle
The HEAD value of a headed phrase is identical to the
HEAD value of the head daughter.
(102) Valence Principle
In a headed phrase, for any valence feature F, the F value of
the head is the concatenation of the phrase's F value with the
list of SYNSEM values of the F daughters.
Another important immediate dominance rule is the Subject-
Predicate Rule, which we can formulate as follows:
(103) Subject-Predicate Rule
[SUBJ h i] ! [SUBJ h[ ]i, COMPS h i], [ ]
HEAD SUBJ
This can be paraphrased as follows:
(104) A sign with the feature specification SUBJ h i can have a
head daughter with the feature specifications SUBJ h[ ]i and
COMPS h i and a subject daughter.
This too interacts with the Head Feature Principle and the Valence
Principle. Given these rules and principles and appropriate linear
precedence rules, we will have structures like the following:
(105)
[1]NP
[2]NP
the dog chased the cat
S
VP[SUBJ ⟨[1]⟩ ]
VSUBJ ⟨[1]⟩COMPS ⟨[2]⟩
Here, and subsequently, we use `XP' as an abbreviation for
X[COMPS h i] and `S' as an abbreviation for V[SUBJ h i,
36 transactions of the philological society 98, 2000
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
COMPS h i]. Following standard HPSG practice, we use boxed
integers or `tags' to indicate that the same object appears in more
than one position in the structure.
We can now return to Welsh. We will begin with non-finite
negation, which, as we noted in the last section, is fairly straightfor-
ward. All that we need are appropriate lexical categories.
With peidio, paid and peidiwch, one complication is that they are
optionally followed by the preposition aà `with'. HPSG recognises a
class of markers, which are the realisation of a feature MARKING
and appear in structures of the following form:
(106) X[MARKING [1]]
[MARKING [1]] X
Assuming this approach, we can propose the following category for
peidio:
(107) verb [nonfin]NEG +SUBJ ⟨[1]⟩COMPS ⟨ ⟩
nonfin(MARKING a)SUBJ ⟨[1]⟩
VP
This tells us that peidio is a non-finite verb, that it is negative, that it
takes as a subject whatever its complement requires as subject, and
that it takes as a complement a non-finite VP, which optionally
includes aà as a marker. Given this category, we will have the
following structure for the complement in (71b):
37borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
38 transactions of the philological society 98, 2000
(108)verb [nonfin]NEG +SUBJ ⟨[1]⟩COMPS ⟨⟩
verb [nonfin]NEG +SUBJ ⟨[1]⟩ COMPS ⟨[2]⟩
[2] VPnonfinMARKING aSUBJ ⟨[1]⟩
VPnonfinSUBJ ⟨[1]⟩
[MARKING a]
ateb y cwestiwnagbeidio
One point to note about peidio is that the phrases it heads are
not possible in all positions in which non-finite VPs appear. For
example, they cannot appear as the complement of an aspect
marker, as the following illustrate:
(109) *Mae Gwyn wedi peidio (aÃ) mynd i Aberystwyth.
is Gwyn after NEG with go to Aberystwyth
`Gwyn has not gone to Aberystwyth.'
We can account for this by stipulating that certain positions only
allow a [NEG ÿ] VP.
We can turn now to the imperative forms paid and peidiwch. We
will assume that imperatives have a phonologically empty post-
verbal subject, and following Borsley (1989) we will assume that
Welsh postverbal subjects are realisations of an extra member of the
COMPS list. Given these assumptions, we can propose that they
involve categories of the following form:
(110) verb [imperative]NEG +SUBJ ⟨ ⟩ COMPS ⟨[1]NP
[NULL +]’ VP ⟩ nonfin(MARKING a)SUBJ [1]
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
This will give the following structure for (74b):
(111) verb [imperative]NEG +SUBJ COMPS ⟨ ⟩
verb [imperative]NEG +SUBJ COMPS ⟨[1] , [2]⟩
paid e
[1] NP[NULL +]
[2] VPnonfinMARKING aSUBJ ⟨[1]⟩
ag ateb y cwestiwn
⟨ ⟩
⟨ ⟩
The negative aspect marker heb is more complex semantically
than peidio, but is simpler syntactically, since it is not optionally
followed by aà . All we need here is the following category:
(112) aspNEG +SUBJ ⟨[1]⟩COMPS VP ⟩
nonfinSUBJ ⟨[1]⟩
⟨
This will give the following structure for the complement in (76):
(113) aspNEG +SUBJ ⟨[1]⟩ COMPS ⟨ ⟩
aspNEG +SUBJ ⟨[1]⟩COMPS ⟨[2]⟩
[2] VPnonfinSUBJ ⟨[1]⟩
fynd i Aberystwythheb
Thus, it is quite easy to accommodate the various forms of non-
finite negation within an HPSG analysis.
39borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
We noted in the last section that non-finite negation licenses n-
words. We will assume that n-words are licensed by a higher
semantic negation. A detailed HPSG account of the licensing of
n-words in Polish and Italian is developed in PrzepioÂrkowski and
KupsÂc (1999). We assume that essentially the same approach can be
applied here.
One further point that we should note here is that non-finite
negation seems fairly straightforward from a semantic point of view.
We can assume that semantic negation resides in the various forms
of peidio and heb. Of course, heb also has an additional aspectual
meaning.
We turn now to finite negation. Here, as we have emphasised, an
n-word is not just possible, but necessary. We propose that this is
because Welsh has a number of types of negative verb, all of which
require an n-word in close proximity. We will identify n-words with
a feature NC (NEGATIVE CONCORD), marking them as [NC +].
We do not necessarily assume that negative verbs have negation as
part of their meaning. This might be the right assumption, but the
fact that negative verbs are often formally identical to their
affirmative counterparts perhaps casts some doubt on it. The
alternative is to assume that negation resides in an n-word. How-
ever, this would entail that n-words are systematically ambiguous
with both negative and non-negative meanings.13
We will look first at examples with a negative subject or object. As
we have seen, there are simple examples, where the subject or object
is neb, and complex examples, where the subject or object contains
quantifier dim. As we have also seen, there is a difference between
subjects and objects in the latter case. The object of the preposition o
cannot be singular in a subject, but it can in an object. This means
that we must distinguish subjects from objects. We will do this with
the feature REST(ricted), marking subjects as [REST +] and objects
as [REST ÿ]. We will assume that neb is unspecified for REST. This
will allow it to appear in either subject or object position. We will
discuss how exactly the contrast between subjects and objects can be
accommodated shortly.
40 transactions of the philological society 98, 2000
13 An analysis of French negation in which n-words are systematically ambiguousis developed in Richter and Sailer (1999).
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
Assuming the REST feature, we can propose the following
category for does in (4) and (22):
(114) verb [fin]NEG +COMPS ⟨ NP , PP ⟩
NC +REST +
This will give the following tree for (4):
(115) verb [fin]NEG +COMPS ⟨ ⟩
verb [fin]NEG +COMPS ⟨[1] , [2]⟩
[1] NP NC + REST +
[2] PP
yn yr ystafellnebDoes
We can propose the following category for welish in (5) and (24):
(116) verb [fin]NEG +COMPS ⟨NP, NP ⟩
NC +REST _
This will give us the following tree for (5):
(117)
[2]NP NC + REST _
verb [fin]NEG +COMPS ⟨ ⟩
verb [fin]NEG +COMPS ⟨[1] , [2]⟩
[1]NP
welish i neb
Thus, it is quite easy to accommodate examples with a simple
negative subject or object.
41borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
42 transactions of the philological society 98, 2000
We now consider examples with a complex negative subject or
object. We will assume that quantifiers are a subset of nouns and
propose the following category for quantifier dim, where it takes a
PP complement:
(118)
PFORM oREST [1]
noun [quant]NC +REST [1]COMPS ⟨ PP ⟩
This allows quantifier dim to be either [REST +] or [REST ÿ], but
requires the PP complement to have the same value for REST. This
means that o will be either [REST +] or [REST ÿ]. We assume that
the complement of o cannot be a singular count NP if it is [REST +],
but is unrestricted if it is [RESTÿ], but we will not formalise this idea.
Given these assumptions, we will have structures like the following:
(119)
[1] PP PFORM o REST +
noun [quant]NC +REST +COMPS ⟨ ⟩
noun [quant]NC +REST +COMPS ⟨[1]⟩
o ’r dyniondim
(120)
[1] PP PFORM o REST _
noun [quant]NC +REST _
COMPS ⟨ ⟩
noun [quant]NC +REST _
COMPS ⟨[1]⟩
o ’r ceffyldim
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(119) is what we need for the subject in (23), while (120) is what
is needed for the object in (51). Dim o'r dynion will also have a
[REST ÿ] analysis. This is needed for the object in (25).
We can now consider examples containing a postverbal negative
adverb. We suggested earlier that these elements are extra comple-
ments. As we have seen, there are differences between dim and byth.
Hence, we must distinguish between them. We will do this with a
feature COMPLEX, marking dim as [COMPLEX ÿ], and byth as
[COMPLEX +]. This reflects the fact that dim is semantically simple
whereas neb is semantically complex. Assuming this, we can propose
the following category for cha' in (6):
(121)
[NC _ ]
verb [fin]NEG +COMPS ⟨ NP , Adv , VP⟩
NC +COMPLEX _
This will give us the following tree for (6):
(122)
[2] Adv NC + COMPLEX _
[3] VP
verb [fin]NEG +COMPS ⟨ ⟩
verb [fin]NEG +COMPS ⟨[1] , [2] , [3]⟩
[1] NP [NC _]
sefyll ymaddimiCha’
One point to note about (121) is that it requires a [NC ÿ] subject.
This will prevent a negative subject with adverbial dim, ruling out
examples like (65). We can assume the following category for dydy
in (7):
(123) verb [fin]NEG +COMPS ⟨NP, Adv , AspP⟩
NC +COMPLEX +
This will give the following tree for (7).
43borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
(124) verb [fin]NEG +COMPS ⟨ ⟩
verb [fin]NEG +COMPS ⟨[1] , [2] , [3]⟩
[1] NP [2] Adv NC + COMPLEX +
[3] AspP
yn gweithiobythMairDydy
Notice that (123) imposes no restrictions on its subject. This means
that a negative subject will be possible with byth as in (66).
We have now provided examples of the four types of negative
verb that we need to accommodate the full range of negative finite
clauses. There is one important restriction that we should emphasise
here. As we noted in section 3, adverbial dim cannot be followed by
an NP. This means that only intransitive verbs will have negative
counterparts of the form in (121), whereas all verbs with an object
will have negative counterparts of the form in (116), and all verbs
will have negative counterparts of the form in (114) and (123). The
restriction of categories like (121) to intransitive verbs is somewhat
like the situation in English, as analysed by Kim (1995), Kim and
Sag (1996) and Warner (forthcoming). In English, we have data like
the following:
(125) a. Sandy is not coming.
b. *Sandy came not.
Kim and Sag and Warner propose that this is because only auxiliaries
have negative counterparts which allow not as an extra complement.
Welsh has a similar but not quite so restrictive constraint.
An obvious question to ask about the analysis we have outlined is:
where do the negative verbs come from? There are two possibilities
here. One is to derive negative verbs from affirmative verbs through
lexical rules. This is the approach which is developed in Kim (1995)
and Kim and Sag (1996) for English and French. Within this
approach, we would need four lexical rules, one for each of the
four types of negative verb that we have proposed.
An alternative approach is possible given the standard HPSG
assumption that lexical entries are the result of combining or
44 transactions of the philological society 98, 2000
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
unifying the information associated with various types in a number
of hierarchical classifications of lexical entries. Such an approach
has been developed in connection with English negation in Warner
(forthcoming). Within this approach, we could propose the follow-
ing hierarchy of negative verb types for Welsh:
(126)
argument negation adverbial negation
finite negation
object subject simple complex
Each of these types will have specific information associated with it,
and each will inherit information from the superordinate type.
Categories like (114) will be the result of combining the information
associated with finite negation, argument negation and subject.
Similarly, categories like (116) will be the result of combining the
information associated with finite negation, argument negation and
object. Space considerations preclude a detailed development of this
approach. However, it seems to us that it is preferable to a lexical
rule approach. One advantage is that it captures the similarities
between the various types of negative verb, which are missed by a
lexical rule approach.
6. Conclusions
In this paper, we have looked at the main types of negative
construction in Welsh, both finite and non-finite, and developed a
detailed HPSG analysis. Of course, we do not suppose that we have
said the last word about Welsh negation. We assume that future
research will produce more comprehensive and more sophisticated
analyses. It may be that it will show that some of the facts that we
have accounted for with quite specific stipulations, e.g. the contrast
between negative subject and negative objects and the contrast
between adverbial dim and byth, are a consequence of more general
principles. We have chosen, however, to provide an account of all
the main facts and not just some of them, leaving the rest to be
accounted for by principles which may or may not emerge from
45borsley and jones ± the syntax of welsh negation
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
future research. Whatever future work may show, we think we have
laid some firm foundations here.14
Robert D. Borsley Bob Morris Jones
Department of Language Department of Education,
and Linguistics, University of Wales,
University of Essex, Aberystwyth,
Wivenhoe Park, Ceredigion S23 2AX
Colchester CO4 3SQ. E-mail: [email protected]
E-mail: [email protected]
References
Acquaviva, Paolo, 1996. `Negation in Irish and the representation of monotonedecreasing quantifiers', in Robert D. Borsley and Ian G. Roberts (eds.), The Syntaxof the Celtic Languages: A Comparative Perspective. Oxford: Blackwell, 284±313.
Awbery, Gwenllian M., 1990. `Dialect Syntax: a Neglected Resource for Welsh,' inR. Hendrick (ed.), Syntax and Semantics 23: The Syntax of the Modern CelticLanguages. San Diego: Academic Press, 1±25.
Borsley, Robert D., 1986. `Prepositional complementizers in Welsh', Journal ofLinguistics 20, 277±302.
Borsley, Robert D., 1989. `An HPSG approach to Welsh', Journal of Linguistics 25,333±354.
Borsley, Robert D., forthcoming. `Mutation and constituent structure in Welsh',Lingua.
Jones, Bob M. and Thomas, Alan R. 1977. The Welsh Language: Studies in its Syntaxand Semantics. Cardiff: University of Wales Press.
Kim, Jong-Bok, 1995. The Grammar of Negation: a Lexicalist Constraint-BasedPerspective, Ph.D. dissertation, Stanford University.
Kim, Jong-Bok and Ivan A. Sag, 1996. `French and English negation: a lexicalistalternative to head-movement', unpublished paper, Stanford University.
Pollard, Carl and Ivan A. Sag, 1994. Head-driven Phrase Structure Grammar.Chicago: University of Chicago Press.
PrzepioÂrkowski, Adam and KupsÂcÂ, Anna, 1999. `Eventuality negation and negativeconcord in Polish and Italian', in Robert D. Borsley and Adam PrzepioÂrkowski(eds.), Slavic in HPSG. Stanford: CSLI Publications.
46 transactions of the philological society 98, 2000
14 One thing future research will need to do is consider the realisation of negationin various dialects, e.g. Pembrokeshire Welsh. According to Awbery (1990), thesequence dim y N, which does not occur in most varieties of colloquial Welsh, occursboth immediately following the finite verb and immediately following the subject inPembrokeshire Welsh. Awbery does not provide the kind of data that would enable usto determine whether dim in these sequences is the quantifier or the adverb. If it is theformer, the differences between Pembrokeshire Welsh and other colloquial varietieswill be quite minor. If it is the adverb, they will be more substantial.
d:/transfer/trps/98-1/borsley.3d ± 29/3/0 ± 12:38 ± disk/np
Richter, Frank and Sailer, Manfred, 1999. `A lexicalist collocation analysis ofsentential negation and negative concord in French', unpublished paper, Universityof TuÈbingen.
Rouveret, Alain, 1994. Syntaxe du gallois: principes geÂneÂraux et typologie. Paris:Editions du CNRS.
Rowlett, Paul, 1996. Negative Configurations in French, Working Papers inLanguage and Linguistics 11, European Studies Research Institute, University ofSalford.
Rowlett, Paul, 1998. Sentential Negation in French. Oxford: Oxford UniversityPress.
Thomas, Peter W., 1996. Gramadeg y Gymraeg. Caerdydd: Gwasg Prifysgol Cymru.Thorne, David A., 1993. A Comprehensive Welsh Grammar. Gramadeg Cymraeg
Cynhwysfawr. Oxford: Basil Blackwell.Warner, Anthony R., forthcoming. `English auxiliaries without lexical rules', to
appear in Robert D. Borsley (ed.), The Nature and Function of Syntactic Categories.New York: Academic Press.
47borsley and jones ± the syntax of welsh negation