Ma gdalena
Charrytiska-Wój cik
The Syntax of Old English Experiencer Verbs
Abstract
The paper offers a new classification of the syntactic frames OE Experiencer verbs can appear
in.
The characteristicsof
individual construction types seen against the background of the variation exhibited by individual verbs found in them allow us to propose that the fifteen attested structures are in fact only variants ofthree basic types.The discussion reveals that OE clausal arguments are Case-marked in the same way as
NP arguments.
A
further observation is that OE lexicon is relatively insensitive to category distinctions but sensitive to Case.l. Introduction
This
paper presents an analysisof
the syntaxof Old English
Experiencer verbs based on an extensive data study. SectionI
offers a presentationof
the possible range
of
structuresin which
Experiencer verbs were foundin
Old English. Section 2 is devoted to the examination of the
morphosyntactic properties
of
these constructions and the presentationof
the distributional pattems
typical of
selected Experiencer verbs. Section 3 contains abrief
summary of the findings of the paper.2.
The syntactic frames of OE Experiencer verbsExperiencer verbs express a physical or mental experience
which
involvesa human
experiencerand optionally the
causeof the
experience. The argument representing the human affected by the experience is assigned the@-role
of
Experiencer,while the
argument representingthe
causeof
the experienceis
assignedthe
@-roleof
Theme. Consequently,the
lexical entryof
any Experiencer verb contains the information that the verb assignsthe @-role of Experiencer and possibly also that of Theme to its
argument(s).
IVhile
the inventoryof
the possible @-roles is commonto
all Experiencer verbs, the syntactic structuresin
which these arguments appeardiffer from
verbto
verb. The Experienceris
always expressedby
an NP;SKY Journal of Linguistics I 5 (2002), 3
I-60
t1 MAcDALENA CuenzvÑsre-WÓ;clr
yet the Theme can either be realised as an NP, PP, CP, or
it
can beleft
out.These structures
will
be referredto in this
paper as 2NPs frame, NP+PP frame, PROP frame, and 1NP frame respectively.In
whatfollows we will
discuss these frames
in
detail.For clarity of
exposition,in all
examplesin this
paperwe
underline Experiencers,italicise fhe
Themes, and use bold type for verbs.Let
us beginwith
2NPs frame, i.e. the onein which
both arguments are realised as NPs.This
frame may appearin four
basic syntactic shapes presented under (1a-d) below.(1) a. tlam wife þa word wel
licodon,the-datwife-dat
the-nom
words-nomwell pleased-plgilpcwide
Geates;boastful-speech-nom of-the-Gaut
'well did those words please the woman' the boastful speech of the Gaut' Beowulf 21 (HCEI)'
b. Ne wilnege ic dæs synfulla
dea&s,("')'
not
desire-lsgl-nom the-gen sinful
death-gen'l
do not desire the death ofthe sinful' The Benedictine Rule 4 (HCET) cd.
Him ofhreow
dæshe-dat
pitied-3sg the-gen 'He was sorry for the man' Homl. Th. t. 192,16 (B&T)mannes man-gen
for
dæmþingemen lYst
ælcesfor tha
thingmen-acc desires-3sg eachþe hi
lyst.that
them-acc desires-3sg'For that reason men desire all the goods they desire' Alfred's Boethius 88 (HCET)
þara gooda
the
goods-genIn (la)
the Experienaer damwífe
bearsDAT
Case, the Themeþa word
is*uik"d NOM
andit
controls verbal concord, as evidencedby the
pluralform of
the verb licodon. This type is referred to in the literature as Type I.In (1b), known
asType II, thã
Experienceric is NOM, the
Theme dæç synfuttan deadesis GEN
and the verb agreeswith
theNoM
Experiencer.I Al I examples from HCET arc quotedby the title and page number ofthe original text. ln the examples cited after secondary sources we stick to the abbreviating conventions used there'
Swr¡x
or OE EXPERTENCER VERBS JJ(1c) and (ld)
representtwo
subtypesof the
so-calledType
N.2In
both examples the verb is in the default 3SG form indicating lack of concord andthe
Theme argument bears genitive Case:dæ
mannesin (1c)
and ælcesþara
goodain (ld). By
contrast, the Experiencer isDAT in (lc): him,but
ACC
in(ld):
men.3The next frame to be
discussedis
characterisedby lack of
an expressed Theme,i.e. INP frame. This frame had
threebasic
syntactic realisations, illustrated by the examples below.(2) a.
Ne ic ne
scamigenot l-nom not
shame-lsg 'I do not feel shame'Ps. Spl. 30, 20 (BeT) b. Ðeah monnum swa ne þince
yet
men-datso
not seemed-3sg 'Yet it did not seem so to the men' Bt. 39, 8; Fox224,17 (B&T)c. Þa sceamode ealle his widerwinnan.
then shamed-3sg
all
his enemies-acc 'Then all his enemies were ashamed.'The Old English Version of the Gospels, Lk. 13,17
ln
(2a) the Experiencerucis in NOM
Case andit
controls verbal concord, as evidencedby the
1SGform of
the verb.(2b)
and(2c)
are subsumed together asType O by Allen
(1995) asthey
share some features, namely bothexhibit
lackof
aNOM
NP and the verbis
invariably 3SG though theonly
argument present bothin
(2b) andin
(2c) is plural. The Experiencer isDAT
in (2b): monnum andACC in(2c):
ealle hiswiturwinnan.
2 The terms we use to denote the types are due to Elmer (l 98 I ). Fischer and van der Leek (1983) call types l, II, a¡d N: Cause-subject, Experiencer-subject, and subjectless respectively. Our choice ofthe labelling Las been influenced by the fact that Elmer's terms, being older than Fischer and van der Leek's, are more widely used and, as pointed out by Allen (1995: 69), Elmer's terms .have the advantage of not prejudging the grammatical relations involved'.
3 According to Fischer and van der Leek ( I 983: 355 table I 9), the Np arguments in Type N can appear in two more structures, namely as ACC ACC and DAT ACC. It is worth noting thar the existence of the ACC ACC t,?e is not recognised either by van der Gaaf( I 904), Visser ( I 963-73), or Mitchell ( I 985). This is so because there is only one genuine example ofthis kind. For a discussion conceming the type see Allen ( I 995: 74ff).
The DAT ACC type, on the other hand, has been argued for in the literatuie. However, as demonstrated by Allen (1995: 79) on the basis ofher own examination ofall the attested examples ofthis kind, 'there is no necessity to assume that any ofthem is DAT ACC'. Therefore, we follow Allen in not recognising the existence ofthis type.
34 MAGDALENA CHnnzvÑsre-WÓ¡ctr
1rr
¡p+PP
frame the Theme is expressed by a PP. Again, as above,it
appears in three basic structures. Consider the examples below.
(3) a.
We witon þæt ...þu
ne recståe
Hte3um mennwe know
that
you-nom not care aboutany
man'We know that you do not care for anybody' Ags. Gosp. Mark xii. 14 (OED)
b.
Ac ic woldewitan
hwederde þuhte be
dambut
I
wanted to-know whether you-data thought-3sg aboutthat& dt hæJit hweder hY were
dept you
have-2sgwhether it was-subj.sg
whetherlæne de
æcetransitory or
eternal'But I would know whether you thought ofwhat you have, that it was temporary or eternal'
Shm. 176, 29 (B&7)
c.
Hie sculon, donnehie
ymbhwø fweod,
cyrran to hiera agnum they should then them-acc5 about what doubr3sgto-tum
to their owninngedonce intellect
'They ought to - when they have doubts about something - turn to their own intellect'
Past. 16: Wst 102, 4-8 (B&T)
In
(3a) the Experiencer isNOM: þu
andit
controls verbal concord.As in (2)
above, the ãxamples quoted under (b) and (c) share some features: lackol a NoM NP and lack of verbal
concord.In (3b) the
Experiencer is expressed by aDAT NP:1e, in
(3c) the Experiencer bears ACC Case: hie'thì
Theme is expressed by a PPin
all three examples: be ænqum menn in (3a), be dam de du hæfst in (3b), andymb hwætin(3c).
The last frame
is
characterisedby
clausal Themes.It
is referredto
as pROpó andit
can appearin five
different syntactic struçtures, presentedin
(4) below.(4) a.
Gifwe
scomiaþdã
we to uncudum monnum suelc sprecenif
we-nom shame-pl that we to unknownmen so
speak-subj'pl . п is the form of both DAT and ACC but it has to be interpreted as DAT s ince þyncan is not aftested with unambiguously ACC Experiencers only with unambiguously DAT ones', É1¿ co-uld thóoretically stand for eithei for NOM.PL or for Acc.PL but the form of the verb (3SG) implies that we are not dealing with NoM here as NOM Experiencers control verbal concord, which is absent here'
u The term ,pROp' a;d the labels used for individual PROP types are due to Allen (1995).
SYNTAX oF OE EXPERIENCER VERBS 35
'If we are ashamed that we speak in this way to strangers' Past. 10; 5w1.63.6 (B&T)
Rofne
randwigan restanlyste stout-acc warrior-acc to-rest desired-3sg'A stout warrior wanted to rest' Beo. Th. 3590; B. 17% (BAT)
þa wæron
ægderge
swiftrange
unwealtran,ge
eacthen
were both
quickerand
more steady and alsohieran þonne þa
odru.næron nawder ne
onhigher than the
other not-wereneither not
asFresisc
gescæpenene on Denisc, but
swa himFrisian shaped nor as Denish but so
them-datselfum duhte þæt hie
nytwyrùstebeon
meahten.selves-dat
seemed-3sg that
they most-usefulbe
might'They were both quicker and steadier and also higher that the others. They were shaped neithe¡ as the Frisian nor as the Danish. But as
it
seemed to them they might be most useful'Chronicle Ms A Early (O2) 90
(HCE\
d. Lareow,ne
olþingd hitde gificþus
wer geceose?master not displeases-3sg
it
you-dat7if I
thus man choose 'Master, doesn't it displease you if I thus choose a man?' Apollonius of Tyre 32 (HCET)e. þa ofþuhte pal Mariuse
bæm consule. Iuliuses eame, then regretted-3sg that Marius-datthe
consulJulius'
uncleþø
monda
gewinnolde
him betæan.that one that
war
not-would him entrust'Then it offended consul Marius, Julius' uncle, that he was not put in charge of the war'
Alfred's Orosius 23 (HCET)
In
(4a) the Experienceris NOM:
we,in (4b) it is ACC:
rofue randwigan.The
examples(4c), (4d), and (4e) all contain DAT
Experiencersi himselfum in (4c), de in (4d),
and,Mariuse þæm consule... in (4e).
Additionally,
(ad) and (4e) containhit
and!æt
respectively,while
no such elementis
presentin
(4c). (4a)is
an exampleof
the PersonalpROp
Type (PERSfor
short),(4b)
and(4c)
are classified together asType
S,(4d)
isb.
c.
7 Þe is morphologically ambiguous between DAT and Acc. However, the only case that oJþyncau assigned to the Experiencer was DAT so in this example y'e has to be interpreted as DAT.
referred
to
as Typehit, aîd
(4e) exemplihes TypeDEM
(a demonstrative pronounþæt is
used). These are thetraditionally
recognised PROP types(cf. Allen
(1995)). There are, however, three additional syntactic structuresin which
Experiencer verbswith
clausal Themes can be found.They
are presented below.(5) a. gifhi þæs
wilniadþæthim
heorayfel unwrecen sieif
they-nom thaT-gen desire-pl that themtheir
evil unpunished is be dæ:s gltes andefneby the
sin's
ProPof ion'Ifthey ask for it that they should not get theirjust deserts' Alfred's Boethius 123 (HCET)
b.
acþæ me þinc<l
dærþd
bio sio sode&
sio fulfremede bulthat-gen me-dat seems-3sg that thatis
the truth and the perfectgesætd &
mæg ælcum hirefolgera
sellan durhwunigendne welan 'happiness which may
each
her followersgive continuous
wealth.But
it
seems to mé that true and perfect happiness is of such kind that it continuously gives wealth to each of its followers'or: .For
if i
mistake not, true and perfect happiness is that which makes a man truly satisfied, powerful, u"nerãt"d, renowned, and happy'8Alfred's Boethius XXXIII 78 (HCEI)
c. Hine þæ
heardostlangode
hwanne heof
ãsse worlde moste' him-acc rhat-gen strongest longed-3sgwhen
he fromthis world
might 'He strongly desired to be allowed to leave this world'Blickl. Homl. 227,1 (B&T)
All
examplesin (5) contain a
clausal Themeand a
demonstrative pronoun bearingGEN
Case: þæs. The differences between (5a),(5b),
and (5c)lie
in:-
the Case of the Experiencer: in (5a)it
is NOM, in (5b)it
is DAT' in (5c) it is-
the ACC;verbal concord: in (5a)it
is controlled by the Experiencer and in (5b, c) the verb shows no concord.All
these existing syntactic pattems availablefor
Experiencer verbsin
OE have been summarised in TableI
below for ease of reference'36 Meco¡Lern CH,cRZYÑsre-WÓ:clr
8 Translated by Cooper (1902).
Srrr¡x
or OE EXPERTENCER VERBS 5tTable 1. The attested syntactic pattems available for OE Experiencer verbs
2. Analysis
As
can be seen, thereexist
15 different types but even a cursory glance atthe table reveals that the
constructionslisted there show
significantsimilarities
so,in
effect,it
may be possibleto
reduce the numberof
types they represent. Let us begin by comparing the propertiesof
2NP typeswith
thoseof INP
types.' Some fields in this column have been left empty as not all constructions presented in Table I have their individual names in the literature.
Frame NO Typ"n Experiencer Theme Verbal concord
Example
2NPs I
I
DAT-NP NOM-NP +Theme (la)ll
l
NOM-NP GEN-NP +Experiencer(lb)
lll N DAT/ACC-NP GEN-NP 3SG (1c, d)
lNP lv NOM-NP Ø *Experiencer (2a)
o
DAT/ACC-NP Ø 3SG (2b, c)NP+P P
vl NOM.NP PP +Experiencer (3a)
vll DAT-NP PP 3SG (3b)
vlll ACC-NP PP 3SG (3c)
PROP lx PERS NOM-NP CP *Experiencer (4a)
X s ACC/DAT.NP CP 3SG (4b, c)
xl hit DAT.NP hit CP 3SG (4d)
xll DEM DAT-NP
þø
CP 3SG (4e)xllt NOM-NP
þæ
CP *Experiencer (5a)xlv DAT-NP
þæ
CP 3SG (sb)xv ACC-NP
þa
CP 3SG (5c)38 MAcDALENA Cunnzvñsrc¡-Wólctr
2.1.
2NP types vs.lNP
typesComparing the properties
of
particular types representing 2NPs framewith
thoseof INP
frame, we immediately notice that oneof
the variantsof
thelatter, namely the one with the NOM
Experiencercontrolling
verbalconcord (cf. N" iv) is strikingly similar to Type II (cf. N" ii). The only
difference between the two types consists in the fact thatin
2NPs frame theTheme is
expressedby a GEN-NP, while in lNP frame it is left
unexpressed.
In effect, if the
expressionof the
Themewere
treated as optional, we could regard thetwo
constructions as variantsof
TypeII,
thus avoiding the needto list
them separatelyin
the lexicon. Instead, we could propose a common lexical entry for thetwo
constructions' The lexical entry would be as follows:Type II
@-roles:
Experiencer (Theme)syntax:
(GEN-NP)A
similar relationship holds in the caseof
TypeN
(N"Ìii)
Type O (N"v):
again, the difference between themis limited to
the Theme,while
the remaining features are shared by thetwo
types: the Experiencer isDAT
orACC,
the verbis
invariably 3SG, and noNP
bearsNOM
Case. Thus, wewill
propose afterAllen
(1995) that TypeO,
instantiatedby
examples (2b) and(2c¡,
shouldin fact
be treated as avariant of
TypeN, in which
the Theme has beenleft
unexpressed. The commonlexical entry for
the two types is as follows:Type N
@-roles Experiencer (Theme) syntax DAT/ACC-NP
(GEN-NP)
In effect, it is
unnecessaryto
recognisethe
existenceof INP
framewith its two
syntactic realisations aswe
are dealing herewith
variantsof
the relevant 2NP types, namely Type
II
and Type N.SvNrex on OE EXPERTENCER VERBS 39
2.2.
2NPsframe
vs. NP+PPframe
Comparing
2NP
typeswith
NP+PP typeswe
noricethat Type II shares
some features with
the first
type listed in
Table I
under Np+pp
frame, i.e.
the one in which the
Experienceris
expressedby a NOM Np
which controls verbal concord(N" vi).
Theonly
difference betweenType II
andN" vi is that in Type II,
aswe
have already established,the
Theme mayeither
be realised as anNP or it
can beleft
unexpressed,while
here the Theme is expressedby
a PP. Thus,it
seems that we shouldin
fact classiS, Novi
as avariant of
TypeII. This
means that TypeII
has three variants,which
are differentiatedonly
by the Theme:it
can be expressedby
anNp,
PP,or it
can beleft
out. The commonlexical
entryfor
themodified
type would be thefollowing:
Type II
@-roles: Experiencer (Theme)
syntax:
(GEN-NP/(PP)An
analysis of the morphosyntactic properties of the remainingtwo
NP+PP types listed in TableI
(cf. N"vii
andviii)
invites comparisonwith
TypeN:
in all three constructions the Experiencer isDAT
or ACC, the verb is invariably 3SG, and there is no NOM-NP. As in the case of TypeII
and its variants, the only difference between the structures consists in the Theme. Therefore,it
is natural to conclude that here again we are dealingwith
variants of one basic type, namely Type N. The modified lexical entry for TypeN
would then be as follows:Type N
@-roles: Experiencer (Theme) DAT/ACC.NP (GEN-NPy(PP) syntax
The
optionality
postulatedin
the lexical entriesof
thetwo
types, i.e. TypeII
and TypeN,
allows us to projectall
the relevant subtypes:if
the Theme is expressedit
can be represented eitherby
an NP or a PP,giving
2NPs orNP+PP frame
respectively.Alternatively, the
Themecan be left out, in
which case the resulting syntactic constructionwill
beINP
frame.In
conclusion,the
proposed readjustmentsallow us to
reduce the40 MAcDALENA CunnzvÑsrn-WÓ¡clr
number of
typesthat
needto be listed in the lexicon under
individual entriesof
Experiencer verbs as the eight various syntactic types discussed sofar,
i.e. N"si-viii in
fact represent only variantsofthe
th¡ee basic types, referredto
as TypeII,
TypeN,
and TypeI. In
contrastto
TypesII
andN, which can
appearin all
three variants, TypeI
does not have the optionof
leaving out the Theme or expressingit
by meansof
a prepositional phrase.Its lexical entry is thus the
following:
Type I
@-roles:
Experiencer Themesyntax:
DAT-NPTable2 below presents a summary of the types discussed so far' Tnre Exoerience¡ Theme Verbal concord
DAT-NP NOM.NP +Theme
il
NOM-NP (GEN-NPY(PP) +ExperiencerN DAT/ACC-NP IGEN-NPYßP) 3SG
Table 2.
2.3.
2NPsframe
vs. PROPframe
In
sections2.1
and2.2 we
discussedthe
constructionswhich were
sostrikingly similar
that postulating a commonlexical
entryfor
the relevant types wãsonly
natural. Here our task isto
see whether the seven differentryþ"r ."pt"t"nting
PROP frame can also be reduced to variants of the basicãñf ryp"r. Due to the considerable structural differences among
the
attesrcá variants of 2NPs frame and PROP frame, we resorted to a different
procedure when
comparing individual
constructions, namely apart
from
ànalysing the morphosyntactic properties characteristic of
these structures,
*e
madã a textuai studyof
the variationof
15 selected Experiencer verbs (see (6) below) to examine their distributional pattems'(6)
gehreowangelician hreowan langian lician
to rue, repent, grieve, pity for something to please, delight
to cause/feel pity regret for something to cause/feel longing, desire, discontent, or pain to please
SyNrax or OE EXPERTENCER VERBS
to cause/feel pleasure or desire for something to displease
to cause/feel griefor pity for something to displease, to be displeasing
to cause/feel regret or sorrov/ about something to cause/feel shame about something to cause/feel doubt about something to cause/feel doubt about something to seem/think, to appear
to desire, to ask for something
4t
lystan mislician oJhreowan oflícian oJþyncan sceamian tweogan tweonian þyncan wílnian
This study is
basedprimarily on the Old English part of the
HelsinkiCorpus of English Texts. Additional
sourcesare Bosworth and Toller
(B&I)
togetherwith the
supplement (^Bfs),Oxford English
Dictionary (OED), Visser (1963-73), andMitchell
(1985). Occasionally we resorted tothe
data quotedby
Wahlén(1925), Elmer (1981), Ogura (1986), Allen (1995,
1996) andto the Old Englßh
Versionof the
Gospels(ed. R. M.
Liuzza), Gregory's Pastoral Care (ed.
H.
Sweet), The Paris Psalter (ed. T.Jebson), and Meters
of
Boethius (ed.T.
Jebson).Finally,
some examples have beenkindly
provided by Professor Cynthia Allen.2.3.1. Type PERS
A
comparisonof the formal
propertiesof 2NP types with the
existing PROP types revealsa similarity
betweenType II
andType
PERS. Bothtypes have a NOM Experiencer controlling verbal concord and
the difference betweenthe two is limited to the
Theme.Type II allows
the Theme to beleft
out or expressed by a GEN-NP or a PP.In
Type PERS the Themeis
expressedby
a clause.This
suggests that here againwe
cantalk
about a variantof
TypeII.
The question is what readjustmentin
the lexicalentry we
needto
proposeto
accountfor this
particular realisationof
the Theme.It
seems natural to postulate yet another categoryfor
the Theme, as in thefollowing:
Type
II
@-roles:
Experiencer (Theme)syntax:
(GEN-NPy(PP/(CP)Let us
notethat our
studyof the
distributional pattemstypical of
verbs appearingin
TypesII
and PERS reveals an interesting dependence, namely42
(7)'o u'
MAcDALENA CHnnzvÑsre-WÓ;ctr
only verbs which appear in Type
II
can also be found in Type PERS, which means that there is no verb which appears in Type PERS but is not foundin
TypeII. This
observation hastwo
important consequences.First of all, it offers further
supportfor the claim that Types II and PERS
representvariants
of
one construction. Secondly,this distributional
pattern reveals the dependenceof
Type PERS upon TypeII.
Therefore,it is
incorrect to includè aCp by
the sideof
anNP
and a PP as a possible categoryof
the Themein
the lexical entry asit
fails to capture this dependence.In
order toformalise the
relationshipthat
obtains betweenTypes
PERSand II we
propose
to
ignore the categoryof
the Theme in the lexical entry altogether.We claim
that the Case information aloneis
suff,rcientto
projectboth
an NP Theme which bears genitive Case and a CP Theme. This entails that the Themeis
Case-marked regardlessof
its category, i.e. both theNP
and theCP are
assignedGEN
Case.In other words, our position is that
the categorial statusof
the Theme is irrelevant and as long as the lexical entry includes the information about the available Caseit will
project the relevant structures as the Case can be absorbed by any Case-absorbing category,i'e'
NP or CP. Under this hypothesis the dependenceof
Type PERS upon TypeII follows
naturally, asit
isonly
on the basisof
2NPs frame that speakerscan
acquirethe
Case availablefor the
Theme.Our
objectivenow is
to providasupport for the claim that the CP Theme in Type PERS doesin
fact bear GEN-Case.In
orderto
do that wewill
resortto
afamiliar
featureof
OE, namely anticipation.
2.3.1.1. Anticipation
Mirchell
(1935: $1445-6) observes that a pronounin
the appropriate case, gender, and number may anticipate a nounwith or without
qualifiers. The ãuthor quotes thefollowing
examples to illustrate the point:And
he
[Malchus] andwYrde and he-nomM-nom
answered 'And he, M, answered'LS 34. 682 (Mitchell 1985: $ 1445)
b. þa da
hi
awocon,[se ealdor
and his profost]"'
when they-nom
woke
the-nom govemor-nom and his officer-nomro For ease ofexposition the square brackets are used in this section in all examples ofanticipation to the anticipates, while the anticipators are in bold type.
mark
SYNTAX oF OE EXPERIENCER VERBS 43
'When the govemor and his officer woke up' ÆcHom ä. t72.17 (Mitchell1985: 91445)
As the
sentences axe grammaticalit is
clear thatno
principles have beenviolated, i.e. the O-Criterion
and CaseFilter
are satisfied.It is
therefore obvious that both elements of the pairs he-
[MatchusJ andhi -
[se ealdor and hisproþstJ
are Case- and @-marked.Both
elementsof
thetwo
pairs carry the @-role of Agent and both are nominative. Clearly, the only wãyin which
both elementsof
eachpair
can receive these properties is through a chain. Thetwo
available features, namely the external @-role and Case are sharedby
both membersof
the chain, i.e. the anticipator and the anticipate are coindexed and share the features via the indices.Clauses
may also be
anticipatedby a
personalpronoun: hit or
a demonstrative one:þættt
bearing the Case appropriatefor
the grammaticalfunction of
the subordinate clauses.This
typeof
anticipationis
illustrated in (8) below.(8) a. grette Geata leod,gode
þancodewisfæstwordum
greeted Geat's man God-dat thanked wisse
words-dat
þæs
[tle hire se willa gelampthat-gen
pt her the wish fulfilledþæt heo onænigne
eorl gelyfde Snena
frofrel.that she on
some
warrior counted-subj.sg wicked-deeds relief'She greeted the man of the Geats, thanked God with wise words for the fact that he¡ wish had been fulfilled, that she could count on some warrior for relief from from wicked deeds'
Beowulf2l (HCET)
He
him þæt
ondrede [þæt he sceolde innan atyddrian]he him-dat(refl) that-acc
fears
that he should inside grow-weak 'He fears that he will become weaker inside'Gr. D. 59, 26 (BTs)
Andgyfhit geweorde,
þætmanmidtyhtlan&
mid uncræftum andif
it-nom happen-subj.sg that man with charge and with ill-practice sacerd belecge],...priest accuse-subj.sg
'And
if it
should happen that a man accuses a priest of charge and ofill
practice'
Laws (Eleventh Century) Q Cnut) 284 (HCEI) b.
c.
"
Þrb, which is also used in this function is rare, so we will limit our discussion to the first two.44 MAGDALENA CHenzvÑsre-WÓ¡clr
d.
butanþæt
geweorde, [þæt he þanonætberste & swa deope
unless that-nom happens that he thence escape-subj.sg and
so
eamestlyfridsocne gesece, þæt
se cyningchim þwh
dæt feoresl,
peace-refugeseek-subj.sghat theking
him-datthroughthatlife grantgeunne
'Unless
it
happens that he may escape and seek a refugeof
peace so earnestly that the king may grant him his life because ofthat'Laws (Eleventh Century)
(
Cnut) 280 (HCEI)In
(8a) and (Sb) the bracketed clauses are anticipatedby þæs and þæt respectively.In (8c)
and(8d)
the embedded clauses areanticipatedby hit
and þæt.t2
In
thelight
of what has been said about the relationship between the relevant elementsin (7)
above,it would
be unreasonableto
deny the existenceof the
samekind of
relationship between membersof the
pairs presentedin
(8), as the mechanismofanticipation
should not be influencedby the
categorial statusof
the elementsinvolved. In
effect,we
conclude thatthe
anticipators and the anticipatesin (8) form
a chain and share the Case and @-rolevia
indicesof
the chain, i.e. both the pronominal NPs and the CPs axe Case- and @-marked.r3Let
usnow
compare the propertiesof
the chainsin (8) with
the properties exhibitedby
ordinary NPs appearingwith
the same verbs in the same functions. Consider (9) below.(e) 4.. Apollonius
hire þ-t
þancodeApollonius her-dat that-gen thanked 'Apollonius thanked her for that' Apollonius of Tyre24 (HCET)
12 For a suggestion conceming the status and structural position ofthe anticipated clause see Cardinaletti ( 1990). Wòiking with German data Cardinaletti proposes to treat the embedded clause unaccompanied by a p.onoin ur un ñgur"nt, while the clause in construction with ¿r is shown to display syntactic properties
þical ofan adjui'ct. This account resembles O'Neil's ( 1977) treatment ofOE relative clauses, which are also analysed as adjuncts.
,r The fact that h¡t and þæt anticipating clausal arguments are @-marked is not uncontroversial. See, for example, Visser (1963-t3) and Mitchell (1985), who consider hit and þæt anticipators ofclausal Themes ¿t the same time classiling tirem as formal subjects devoid ofany meaning. Bolinger ( I 979) and Vikner ( I 995) argue against assigning-the status ofexpletìves to fhe correspondin_g MnE pronouns.in parallel examples.
Si;ilari, cardinitettiitlgo) argues thàt German ¿s cannot be analysed as an expletive when it co-occurs with an åmbedded clause ând shows that es has the status ofân argument. Dutch åel, as anaìysed by Bennis (1986), conesponds to German ¿s in this respect. While these studies do not deal with historical data, an ìna"pána"nt .*à.ination canied out by Naya (t
ll:¡
dealing with ålr and 1ær anticipating subject clarses in OE conoborates the claim that these pionoun. ur. not .xpletive. See section 2.3.2 for the details ofNaya's investigation.-SYNTAX oF OE EXPERIENCER VERBS 45
b.
Ic ondrædeme
godI fear
me-dat(refl) God-acc'I fea¡ God' Gen.42,18 (BA7)
c.
Gewurdon manigewundor
on manegum landum happened-plmany
wonders-nomin many
lands 'Many wonders happened in many lands'Ors.5, 10; Bos. 108, 16(BAT)
I-n (9a) the
NP
objectþæ
bears genitive Case andis
assigned the intemal@-role
of
Themeby
themain
verbþancian. Note
thatthe chain in
(8a) exhibits exactly the same features.As for
(9b), theintemal
argtmtentgod
bears accusative Caseand the role of
Themeprovided by the
predicate ondredan. The chainin
(8b) has the same features.In
(9c) theNp
subject bears nominative Case and the extemal @-roleof
Theme. The chainsin
theparallel
examplesinvolving the
same verb geweorþan, quoted under (8c, d), are suppliedwith
the same properties.In
conclusion, the comparisonof the properties exhibited by the
chainsin (8) with the
featuresof
the c_orresponding NP argumentsin
(9) reveals that the inventoryof
Cases and@-roles
in a given
verb-argument relationshipis not influenced by
the categoryof the
argumentin OE.
Consequently,the
unanticipated clausal arguments quoted under (10) below are expected to be Case- and @-markedwith
the same properties as their respective equivalents presented under (8) and (9) above.(10)
a. Ic dancige<le,
dæt ic ne eom na swilce odre mannumI thank
you-datthatlnotam notlike
othermen 'I thank you that I am not like other men'Hml. Th. ä.428,19 (B&7)
He
him
ondrædan sceal dæt he unmedome sie he him-da(refl)fear
shall that he unworthy is 'He ought to fear that he is unworthy'Past. 73,
2l
(BTs)&.
æfre negeweorde, þrt
Christen man gewifigeand ever not happen-subj.sg that Ch¡istian man marry-subj.sg in VI manna
sibfæce
on his agenum cynne, ...in 6
men's degree-of-relationshipin
hisown
kinn'And it should never happen that a Christian man marry within six degrees ofconsanguinity'
b.
c,
46 MAcDALENA CHARZYÑsKA-WÓJcIK
Laws (Eleventh Century)
flI
Æ þelreQ 250 (HCET)Working
on what has been established above, we conclude that the clausalurgutn"nt in (lOa) is genitive, the
onein (lOb) is
accusativeand (l0c)
contains a sentential subject in nominative.¡a
In
sum, the discussion conceming anticipationallows
usto
concludethat
argument CPs possess the same features astheir NP
equivalents thus indicating that the categorial statusofthe
argument doesnot
influence the propertiesit
receives.This in tum
corroborates the hypothesis thatin
OldEnglish the
categoryof
the argument neednot
be includedin the
lexical entryofa
predicate.Let us now retum to Type
PERS.As has
been remarkedat
the beginningofthis
section, anticipationis
an optional device. Consequently,*"
"*p"õt that Type
PERS should alsooptionally allow an
anticipatorrs and,if we
are correctin
claiming that theCP in Type
PERS bears GENCase, the pronominal anticipator is also
expectedto be GEN.
This suppositionis
supportedby
the existenceof
exampleslike
the one quoted under (5a) above, repeated here as(l l).
(11) gifhi þæ wilniad þahím
heorayfelunwrecen sieif
they-nomthat-gen desire-pl thatthemtheir evil unpunished is beda
gltes andefnelbythe sin's proportion
'Ifthey ask for it that they should not get theirjust deserts' Alfred's Boethius 123 (HCEI)
The
existenceof GEN
anticipatorsin Type
PERSnot only
supports the claim that the clausal Theme is indeed case-marked in this construction butra See Charzyiska-Wójcik (2001) for a detailed discussion ofCase-marking ofclauses in Old English and Rostila (in piess) for a ãiscussion ofCase-marking ofclauses in general based on data from German, English, Finnish and Swedish.
rr Both anonymous reviewers suggest that anticipators are always present but they are not always phonologically realised. One of thè ieviewers points out that under this hypothesis we would not need to
äs.ur" ihat verbs Case-mark CPs as Case would always be assigned to nominal arguments, i.e. in constructions with clausal arguments Case would be always assignedtothe anticipator, either overt orcovert.
Theotherreviewerremarkstñatthe Case-markingofsubordinate clauses would follow naturally iftheclauses were always in apposition with an anticipator because elements in apposition Case-agree with their apposites.
Therevieweremphasisesthatsuch an asiumption accords withthemost strikingcharacteristic ofOld English, namely its paratactic style.
'ihis hypothesis is a very interestingalternative to the view thatanticipators are optionâl.Note, however, that the Casé-marking ofcl'auses will iollow under either hypothesis: through a chain with a pronominal anticipator, or via dñect Case-marking by the main verb. Therefore, we will not investigate here the differences between the t\.vo proposals.
SrMex op OE EXPERTENCER VERBS 47
it also allows us to further
reducethe
numberof
typesthat
needto
be recognised:while (5a) was listed in Table I as a
separate construction under Noxiii, it is now
clear thatit
should be interpreted asType
pERSwith
an anticipator, hence a variantof rype II. It
is important to add at thispoint
that textual data support the above conclusion: exampleslike
the one quoted above under(salll) exhibit
the same dependence uponType II
asType PERS, i.e. they are never attested
with
verbs which do not appearin
TypeII
and the dependenceworks only
oneway.
rùy'e cannow
formulate the revised lexical entry for TypeII:
Type
II
@-roles: Experiencer (Theme) (GENy(PP)¡ó syntax:
The major
assetof this
proposal consistsin the fact that it
captures the dependenceof
PROP types uponType II at the
sametime
revealing an interesting principle that seemsto
operatein
theoE
lexicon: the categorial statusofthe
argument need not be included in the lexicon atall;
selectionalrestrictions alone will prohibit the
appearanceof illegitimate
structures (such as the oneswith
clausal Experiencers).r7The
proposedlexical entry for Type II allows us to project five
syntactic structures listed separately in Tablel:
-
No ii, i.e. a 2NP type if both ¿¡rguments are realised as NPs;-
N"iy, i.e. a INP typeif
the Theme argument is not expressed; this is possible as the @-role of Theme and GEN/PP are marked as optioni;-
Novi, i.e. an NP+PP type if the Theme is realised as a pp;-
No ix, i.e. Type PERS if the Theme is realised as a clause;-
Noxiii, i.e. NOM-EXP+Iæ+CP, if the clausal Theme is anticipated by a16 An anonl.rnous reviewer suggests that treating PP as a Case-absorbing category would simplif,i the representations even more: to case alone, absorbed by Np, cp, or pp but phonologically realised only on Np.
However, as remarked in the review, there are no PP anticipators in OE, which could support this view. Note, moreover, that prepositions are Case-assigners and as such cannot receive Case due to Case Resistânce Principle. Finally, observe that the actual choice ofthe preposition is an idiosyncratic property ofindividual verbs and therefore has to be specified in the lexicon. Consequently, the proposed reduction in the lexical eÌtry ofExperiencer verbs is motivated only for NP and CP.
¡? See Charzyliska-Wójcik(2001) foramore detailed analysis ofthe variation between NP and CP arguments in various structures (with and without anticipators), supporting the claim that the category ofthe aigument need not be subcategorised for and that the major principle responsible for projecting the stnrcture ofãlauses in OE is the Case information included in the lexicon.
48 Mrconr.eNR CHnnzvÑsre-Wó¡clr
pronoun.'8
In
sum, the comparisonof
the relevant syntactic structures supportedby a
detailed studyof variation
allowed usto
concludethat Type II is
a basic syntactic pattem for all the variants mentioned above.2.3.2. Types
åiland DEM
We
will follow Allen
(1995)in
analysing thesetwo
constructions together asthey
arevirtually
identical: bothexhibit
aDAT
Experiencer, a clausal Theme, and a pronoun. Theonly
difference between them consistsin
the fact that one has a personal pronoun:hit,
while in the other the pronoun is a demonstrative one: þæt. The fact that the Experienceris
exclusivelyDAT invites
comparisonwith Type I. For the
convenienceof the
reader the relevant portion of Table 1 is repeated below.Frame No Type Experiencer Theme Verbal
concord
Example
2NPs I I DAT-NP NOM-NP +Theme (la)
PROP xl hit DAT-NP hit CP 3SG (4d)
xll DEM DAT-NP þd CP 3SG (4e)
At first
glance, the Case of the Experiencer is theonly
feature thatall
three types havein
common. However, important information about the correct interpretationof
Typeshit andDEM
comes from variation facts.Our
study revealedthat Types hit
andDEM
appearexclusively with those
verbswhich
appearin Type I but
the appearanceof
averb in Type I
does notautomatically involve its occurence in Type hit or DEM.
Thisdistributional pattem suggests that Types
hit
aîdDEM
represent variantsof Type I in which the
Themeis
expressedby a
clause rather than an NP.What remains to be discussed now are the
two
featwes that differentiate the typesin
question, i.e. the pronominal elementshit
and þæt and the concord parameters.Let us begin with the
statusof hit
andþæt. Morphologically
these pronouns are ambiguous betweenNOM
andACC. All existing
accounts, seefor
example Visser (1963-73),Mitchell
(1985),Allen
(1995), and Naya(1995), interpret these pronouns as nominative. Visser (1963-73)
and18 Note that th" p.oposed lexical entry actually predicts the existence ofthis type.
SyNrex on OE ExpenlpNcen Venss 49
Mitchell
(1985) claim thathit
and þæt are expletives functioning as formal subjects anticipating the clausal arguments.Allen
(1995) also treats these pronouns asformal
subjects. On the other hand, Naya ( 1995) argues thathit
and
læt
anticipating subject clausesin
OE are not expletive. Naya,s study reveals that when used as anticipators,hit
andþæt
are not interchangeable (Naya (1995: 3a)). The author shows thathit añ
þætdiffer
in the degreeof
referentiality, namely anticipatory åü is slightly less referential
than anticipatoryþæt (which can be
shownto carry
stress and/or emphasis).Therefore,
if
thetwo
pronouns can be shownto differ with
regardto
the degreeofreferentiality,
they cannot reasonably be claimedto
be devoidof
meaning.
This
agreeswith what we
have establishedin
section 2.3.1.1, namely that anticipators share the featuresof
the anticipates so they carnot reasonably be claimed to be devoidof
@-role. Another important argumentagainst analysing hit and þæt as formal subjects comes from
the comparisonof
Typeshit
andDEM with
Type PERS.As
has already been noted, Type PERS optionally contains avery
similar element, namelylæ,
which, in
parallelto
Typeshit
andDEM,
anticipates the postverbal clausal Theme.Therefore,
while it
could theoretically be claimed that, viewedfrom
a diachronic perspective, the presenceof hit
and þætin
Typeshit
andDEM is
dueto the growing
needin the
languageto
equip every clausewith
a subject, one can proposeno
suchmotivation for the
presenceof þæs in
Type PERS, so thisline of
reasoning is clearly fallacious. Consequently,it
can be said that the
claim
thalhit
and þæt areformal
subjectsin
Typeshit
andDEM
respectivelyis
circular and theory-intemal sinceit
presupposes thatOE
hadformal
subjects at the sametime
substantiatingthis claim
by ascribing the statusof
formal subjects to the elementsin
question. Hence,it
seems incorrect
to
treathit
and, þæt asformal
subjects.reIf
these pronouns are notformal
subjects, what are they? As we have already remarked, Type PERSoptionally
contains a pronoun (1æs) functioning as an anticipatorof
a clausal Theme.
It
seems naturalto infer
thathit
andþø in
Typeshit
andDEM
havethe
samefunction
as þæsin Type
PERS.This
conclusion is strengthenedby the fact that hit
andþæt
bearthe
Case expectedof
the anticipatorin
Typeshit
and,DEM,
i.e. nominative since, as the data study has revealed, these types are based on TypeI, which
assignsNOM to
the Theme. Establishingthe
Case andfunction of hit
andþæt
has important consequencesfor
the interpretationof
the verbal concord.First of all,
there More arguments against this view have already been presented in ft. 13.
50 MAGDALENA CnanzvÑsr¡-Vy'Ó¡cx
fact that the pronouns are nominative rather than accusative means that the
verb
agreeswith the
nominative pronoun ratherthan exhibiting lack of
con"orã. Furthermore,
the fact
thathit wÁ
þæt bear theO-role of
Theme indicates that the verb agreeswith
the Themejust
asin
the caseof
TypeI.
Let us then
summarisethe
morphosyntacticproperties of the
threeconstructions in a table.
Twe Experiencer Theme Verbal concord
I
DAT-NP
NOM-NP
+Theme
hit l¡rf-NOM NOM-CP
DEM þ,f/"-NoM NOM-CP
Table 3.
As we can
see,the
differences between TypesI, hit
andDEM
areonly superficial: all three types exhibit a DAT
Experiencer,NOM
Theme (eipressedby
anNP or
a CP anticipatedby
a pronoun) and a verbwhich
agreeswith
the Theme. Working on the hndings presentedin
section 2.3.1, namely that there is no needto
subcategorisefor
the actual categoryofthe
Theme,we
can proposea modihed lexical entry for Type I, which will
project both TypeI
and Typeshit andDB}ll:
Type I
@-roles: Experiencer Theme DAT syntax:
The lexical entry does not
speciff
the category of the Theme only the Caseavailable for it,
henceit
accountsfor the distributional
dependenceof
Types
hit
andDEM
upon TypeI.
Note, however, that since anticipation isoitìonal,
the lexical entry proposedfor
TypeI
predicts the existenceofyet
another variant
of
TypeI,
namely a structure in which the clausal Theme is not anticipated, i.e.:DAT-NP
Experiençe¡+
CP Theme: a pattemformally identical with a
subtypeof Type s (cf. N" x in Table l) in which
the Experiencer bearsDAT rather
thanACC
Case.This is, at
leastat first
glance, not a desirable effect asit
entails a splitwithin
Type Sfor
which welould
need independent support. Wewill
postpone the discussionof
this problemtill
we have analysed Type S in detail.SYNTAX oF OE EXPERIENCER VERBS 5l
2.3.3. Type S
The properties
of
Type S, i.e.DAT or ACC
Experiencer, lackof
aNOM NP
andof
verbal concordinvite
comparisonwith rype N, which
exhibits the same characteristics. The sole difference between thetwo
constructions liesin
the Theme: realised as an NP, PP or Øin
TypeN,
and as a clausein Type S. This
structural likeness suggeststhat
here againwe
are dealingwith
a variantof
the basic TypeN.
However,in
contrastwith
thepROp
types discussed so far, textual data do not corroborate this hypothesis:with
the exceptionof wilnian'to
desire'all
the verbs listedin
(6j-above appearin Type S, while
only gehreowan, Itreowan,langian,
lystan, o/hreowan, oJþyncan, sceemian, tweogan, tweonian,þyncan are found in Type
N.Consequently,
the claim that Type N is basic for Type S
carurot be sustained.Additionally,
a detailed studyof
the variation exhibited by verbs appearingin
Type S(cf.
Table 4 below) shows that no other2Np
type can be shown as underlying for Type S.Verbs in Type S Verbs in Type N Verbs in Type
I
Verbs in Type IIgehreowan gehreowan gehreowan
eelician gelician
hreowan hreowan hreowan
langian langian
lician lician
lystan lystan lystan
mislician mislician
ofhreowan ofh¡eowan oflneowan ofhreowan
oflician oflician
ofbyncan ofbyncan ofbyncan
sceamlan sceamlan sceatnlân
tweogan tweosan tweogan
tweonian tweonian tweonian
byncan bvncan bvncan
Table 4. The occurrence ofverbs ofType S in 2Np Types