Some Notes on Left-Dislocation in the Homilies of Wulfstan

Abstract

In this paper, I show how pragmatics and syntax are interconnected in Old English by examining the left-dislocation system in Wulfstan’s homilies. Syntactically, this article argues that left-dislocation fits in nicely with the system found in other Old English texts, despite certain superficial structural differences. The unexpected high number of demonstrative resumptives is accounted for by the accumulation of formulaic structures in one homily. Pragmatically, LFD performs a number of discourse functions. The main function is a generalizing one, as LFD introduces new topics in the discourse. This case study also shows that other functions traditionally linked with LFD can be found in Wulfstan’s texts. For instance, demonstrative resumptives show some degree of topic shifting and can be accompanied by the contrastive function. By contrast, personal pronoun resumptives can mark topic continuity with specific referents. Since the corpus data are necessarily limited because only one file from the YCOE is examined, some claims are not verifiable. A good example is the assumption found in the literature that personal pronoun resumptives in LFD tend to land low in the clause.

1. Introduction

This article examines Old English (OE) left-dislocated (LFDed) structures, illustrated in (1b) and (2b). Their Present-Day English counterparts are shown in (1a) and (2a). In these constructions, the resumptives in main clauses, typically demonstratives or personal pronouns, are anaphorically linked to the dislocated nominals. In the literature on Germanic languages, LFDed structures with demonstratives are labelled contrastive left-dislocation (CLD), whereas personal pronouns are found in hanging topic left-dislocation (HTLD); see, for instance, Grohmann (2003) and Sturgeon (2008) and the references cited therein. Taylor (2014, pp. 417–18) assumes that the same dichotomy can be found in OE LFDed structures as well. Furthermore, demonstratives in the former type occur in a high position in the clause, i.e., above the subject. Thus, þam in (1b) lands above the pronominal subject he. By contrast, personal pronouns in the latter type normally appear lower in the clause, i.e., below the subject. This is shown in (2b), where him surfaces lower than the subject hy (for more information on the structural positions in OE, see, for example, Taylor 2014, §8.2.3, and references cited therein).1 Interestingly, personal pronouns and independent demonstrative pronouns perform various pragmatic functions in the clause. Los and van Kemenade (2018) observe that in OE, demonstratives are considered as topic-switching devices, whereas personal pronouns mark topic continuity (for a similar system in modern Germanic languages, see, for instance, Comrie 2000; Bosch et al. 2003; Bosch and Umbach 2007; van Kampen and Pinto 2008). They report that the topic-shifting function of weak demonstratives is primarily linked with the so-called Spec, CP position, which is roughly equivalent to the position above the subject. When personal pronouns appear in the clause initial position, they express topic continuity. The question is whether similar pragmatic functions can be observed in OE LFDed structures, since OE is indeed very Germanic and Present-Day English LFDed structures are linked with certain pragmatic functions like the licensing of topic continuity or contrast (Traugott 2007). More specifically, the question is whether different structural positions of demonstrative and personal pronoun resumptives (high and low positions) are reflected in different pragmatic functions that LFDed structures perform.

(1a)

The people who earn millions and pay next to no tax, those are the targets.

(Birner and Ward 2002, p. 1413, quoted in Los and van Kemenade 2018, p. 148)

(1b)

þa

þonne

þe

his

leasungum

gelyfað

&

him

to

those

then

that

his

lies

believe

and

him

to

gebugað

þam

he

byrhð

her

for

worulde

turn-back

those

he

protect

here

for

world

‘those however who believe in his lies and turn back to him, them he protects here before the world’   (cowulf,WHom_4:40.123)2

(2a)

John, I never really liked him

(2b)

se

ðe

to

þam

gesælig

bið

þæt

he

to

ðam

mærðum

he

that

to

that

fortunate

will-be

that

he

to

the

glories

&

to

þam

myrhðum

cymð,

ne

ateoriað

hy

him

æfre,

and

to

the

joys

will-come

not

will-end

they

him

ever

‘he who will be so fortunate that he will come to those glories and to those joys, they will never come to an end to him’ (cowulf,WHom_7:155.497)

A major aim of this paper is to check how Wulfstan used LFDed structures against the backdrop of the findings on the syntax of LFD and pragmatics of demonstratives and personal pronouns in recent studies. For example, we know that personal pronouns are always more common than demonstratives in OE LFDed constructions. Traugott calls demonstratives in LFD merely ‘an alternative’ (Traugott 2007, p. 420). Bartnik (2024, p. 39) examined nine texts with the highest number of LFDed structures in YCOE and personal pronoun resumptives are more common in all of them. Sometimes, the numbers are rather disproportionate. For example, Ælfric’s Catholic Homilies I and II contain 178 and 103 personal pronoun resumptives, respectively, and only 46 and 42 demonstrative resumptives, respectively. Similarly, the ratio of personal pronoun and demonstrative resumptives in Cura Pastoralis is 193: 38. However, Wulfstan’s homilies show the opposite trend: out of 70 examples of LFD, 36 contain demonstratives and only 34 contain personal pronouns. Traugott (2007) includes Wulfstan’s homilies in her investigation, but she does not distinguish between personal pronouns and demonstrative resumptives in her calculations. Therefore, the question is whether the relatively high number of demonstratives is linked with their pragmatic function, Wulfstan’s rhetorical homiletic style, or something else. Cole (2017, p. 398) observes that Wulfstan unambiguously marks discourse-new referents by using se-pronouns. If LFDed structures introduce new topics, perhaps this fact would account for the demonstrative preference.

A second aim is to show that corpora like Taylor et al.’s (2003) York–Toronto–Helsinki Parsed Corpus of Old English Prose (YCOE) are excellent research tools, but even this corpus has some limitations because some texts might be incomplete or only include selected copies of a manuscript in the corpus. For instance, in order to determine the topic-shifting or topic-continuity functions of pronominal elements in LFD, we need to take into account the preceding or following context. Specifically, topic-shifting involves non-topical referents (mostly non-subjects) in the preceding context that later become topics (subjects). Topic continuity entails subsequent mention(s) of the preposed material in the clauses immediately following LFDed constructions. Traugott (2007, p. 426) adds a restriction whereby topic continuity should be coded only when the subsequent mention is the subject (not object) of a main (not subordinate) clause. If a truncated version of a manuscript is used as a basis for a corpus text, then the context might change and, consequently, the pragmatic function of a given element might be different. Sometimes, different copies of the same manuscript use different pronominal forms. Then, the question arises whether or not we actually deal with different types of LFDed structures. I will illustrate these points below.

The article is organized as follows. Section 2 outlines the methodology and the source of the data for this study. The structural distribution and functional interpretation of demonstratives in LFDed constructions is discussed in Section 3.1. Section 3.2 presents an analogical situation with personal pronouns in LFDed structures. Section 4 draws together some conclusions.

2. Corpus and Methodology

The corpus material for the present study has been extracted from YCOE. The cowulf.o34 file contains 28,768 words; the text is not a translation and it is of the West-Saxon dialect. Allen (2022, p. 117) notes that the YCOE file is based on the copies of the manuscripts that contain ‘homilies from considerably later than Wulfstan’s death in 1023′, which makes it a Late Old English compilation. Cowulf.o34 contains excerpts from the following homilies: Ib, II, III, IV, V, VI, VII, VIIa, VIIIb, VIIIc, IX, Xa, Xb, Xc, XI, XII, XIII, XIV, XV, XVIb, XVII, XVIII, XIX, XX, and XXI. The LFDed structures are distributed among the majority of them, as illustrated in

Table 1. The distribution of LFDed structures in the YCOE file cowulf.o34.

Homily	LFDed Structures
	Demonstratives	Personal Pronouns
Ib	1	—
IV	4	1
V	3	—
VI	2	6
VII	1	5
VIIIb	3	1
VIIIc	2	2
IX	—	2
Xa	—	1
Xb	1	3
Xc	16	1
XIII	—	2
XIV	—	1
XVI	—	—
XVIb	—	1
XVII	—	3
XVIII	3	2
XXI	—	2
cowulf.o34	36	34
Total	70

below:

The extraction of the relevant material from YCOE was easy, as CorpusSearch (http://corpussearch.sourceforge.net/, accessed on 26 February 2024) allows us to enter a query with the -LFD tag. Of course, I had to cull some examples irrelevant to this study. Non-target data involve LFDed free relatives, illustrated in (3), and LFD with resumptives that are neither personal pronouns nor demonstratives, as shown in (4).

(3)

Swa

hwæt

swa

ge

gebindað

her

ofer

eorðan

fæstum

So

what

as

you

bind

here

over

earth

fast

bealubendum

for

yfelum

gewyrhtum,

eall

hit

wyrð

on

grevious-bonds

for

evil

deeds

all

it

will-be

in

heofenon

swa swa

on

eorðan

mid

Godes

yrre

gebunden

heaven

as

on

earth

with

God’s

anger

bound

swiðe

fæste.

very

fast

‘Whatsoever you bind here on earth with secure, pernicious bonds for wicked deeds, it will all be bound quite securely in heaven as on earth with God’s anger’

(cowulf,WHom_17:57.1405)

(4)

Ac

se

ðe

þyder

mid

clænum

mode

færð

&

þær

bliðe

but

he

that

thither

with

clean

mind

goes

and

there

joyful

bið

for

godcundre

þearfe

&

to

Gode

georne

þencð

is

for

God-given

need

and

to

God

eagerly

thanks

&

clypað,

his bena

beoð

Gode

gecweme

and

calls

his prayers

are

to-God

pleasant

‘but he who to that place [to church] with pure mind goes and there is joyful for God-given need and thanks and calls to God eagerly, his prayers are pleasant to God’

(cowulf,WHom_18:119.1490)

Since the YCOE does not contain a complete collection of all Wulfstan’s homilies, some examples had to be drawn from supplementary sources like Bethurum (1957) or Napier (1883).

Contemporary corpora are excellent tools to carry out quantitative analyses, as they can crunch vast amounts of data with astonishing speed. It is sufficient to say that the biggest corpora, like iWeb, which are now available at english-corpora.org, contain billions of words and grow bigger practically every day. In contrast, historical corpora, especially those that cover the earliest period, are far smaller. In fact, the biggest OE corpus, the Dictionary of Old English Corpus, comprises around 4 million words (3 million running words of Old English plus 1 million running words of Latin). The second biggest corpus—the YCOE—contains 1.5 million words. Consequently, quantitative investigations in historical studies are conducted on a much smaller scale. At the same time, any statistical trends, no matter whether indicated in contemporary or historical corpora, should be interpreted with due caution. As remarked above, resumptive elements in LFD constructions in OE show a consistent trend: personal pronouns are always more common than demonstratives in OE texts. However, the homilies of Wulfstan show the opposite preference, as demonstrated in Table 1. The question is as follows: is it a genuine anomaly? If yes, what is the reason behind it? An alternative is that the figures are skewed, and the higher number of demonstratives can be accounted for in a different way. In my analysis, I intend to show that the unusually high number of demonstratives is in fact due to one homily and one type of construction. A general quantitative analysis might easily overlook such fine points. Let us now turn to LFDed structures with demonstrative resumptives.

3. Discussion

3.1. Demonstratives

As already mentioned, the most striking quantitative fact about LFDed structures in Wulfstan’s homilies is that demonstrative resumptives outnumber pronominal resumptives. This does not happen in nine texts with the highest overall number of LFDed constructions in the YCOE, as shown above. The high number of demonstrative resumptives could suggest that Wulfstan needed them to mark some pragmatic functions. For instance, we could expect a lot of topic-shifting in the text. However, a closer look at the text reveals that this anomaly is only apparent. Out of 36 resumptive demonstratives, as many as 16 instances appear in one homily, namely Xc. Moreover, 15 of them are used in a repetitive structure: se þe wære … weoðre se, as observed by Orchard (2004, p. 84). Orchard also notes that the ‘binary’ structure of these examples is clearly modelled on Latin. One example and its Latin equivalent are given below (for the full passage, see Appendix A).

(5)	Se	þe	wære	weamod,	weorðe	se	geþyldmod.
	he	that	should-be	angry	should-become	he	patient
	‘he who was angry, let him become patient’ (cowulf,WHom_10c:126.950)
(6)	Qui	fuit	iracundus	sit	patiens
	who	was	irascible	should-be	patient
	‘whoever has been irascible, let him be patient’
	(Wulfstan, Xb 99–100; Bethurum 1957, p. 197)

The OE passage is not a close translation of the Latin text (see Orchard 2004, pp. 85–86 for details and Appendix A below). Apart from lexical choices, the structure of the LFD constructions and their Latin equivalents supports this claim too. Firstly, the Latin structures are not LFDs at all because there are no resumptive elements. Secondly, both verbs in the OE text, i.e., wære and weorðe, are subjunctives, whereas the Latin excerpt uses only one subjunctive form, i.e., sit. The other verbal form, i.e., fuit, is indicative.

Another interesting fact about Wulfstanian formulaic passages is that there is some variation in the usage of resumptive elements. Consider the following example:

(7)

Se

þe

wære

gifre,

weorðe

se

(C/E he)

syfre;

he

that

were

greedy

should-become

he

temperate

&

se

ðe

wære

galsere

on

fulan

forligere,

and

he

that

were

lustful-man

in

foul

fornication

weorðe

se

(C/I he)

clænsere

his

agenre

sawle

should-become

he

cleanser

his

own

soul

‘he who was greedy, let him become temperate; and he who was a libertine in foul fornication, let him become the cleanser of his own soul’

(cowulf,WHom_10c:124.948–949)

As registered in Bethurum (1957, p. 206), in two manuscript versions, he replaces se. Specifically, the first resumptive se is replaced by a personal pronoun in Corpus Christi College, Cambridge, 201 (manuscript C) and Bodleian, Oxford, Hatton 113 (manuscript E)—both from the late eleventh century. The other personal pronoun form is additionally found in manuscript I (British Museum, London, Cotton Nero A I), dating from the first quarter of the eleventh century.3 Since the two manuscripts (C and E but not I) come from the post-1023 period, scribes might have introduced some changes to the pronominal system used earlier (cf. Allen 2022, p. 117). This hypothesis might explain the variation found above, which seems to have nothing to do with different pragmatic functions of the pronouns used in LFDed structures here.

When it comes to the subject matter, the portion of homily Xc described above lists instructions about how to salvage your soul; hence, it contains so many subjunctive forms. The examples have generic references that juxtapose good and evil human traits. Moessner (2020, p. 69) notes that there is a certain link between the use of the subjunctive and the antecedent of the se, seo, þæt paradigm in relative clauses. She illustrates this point with the only LFDed structure from homily Xc that is not part of the passage described above and is therefore not found in Appendix A. Consider the following:

(8)

Se

þe

secge

þæt

he

on

Crist

gelyfe,

fare

se

he

that

should-say

that

he

in

Christ

should-believe

should-go

he

þæs

rihtweges

þe

Crist

sylf

ferde

the

right-way

that

Christ

himself

went

‘The one who says that he believes in Christ shall go on the right way on which Christ himself went’                 (cowulf,WHom_10c:12.835)

Indeed, whenever a subjunctive verb is used, the relative starts with se þe. However, it is fair to say that se þe relatives are also found with the indicative mood. This is illustrated by Example (9) below.

The remaining examples of LFDs with demonstrative resumptives are distributed rather evenly among a number of homilies (cf. Table 1 above). Each homily contains no more than four such instances. Let us consider the following portion:

(9)

And

se

þe

þær

þæt

deð

þæt

his

þearfa

beoð,

se

and

he

that

there

that

does

that

his

need

is

he

gegladað

God

&

his

englas.

pleases

God

and

his

angels

‘and he who does there what his need is, he pleases God and his angels’

(cowulf,WHom_18:46.1445)

(10)

ða

þe

her

nu

deofle

fyligað

&

his

unlarum,

those

that

here

now

devil

follow

and

his

evil-teaching

þa

sculon

þonne

mid

deofle

faran

on

ece

those

must

then

with

devil

go

on

eternal

forwyrd

helle

wites.

perdition

hell

torments

‘those who now follow the devil here and his evil teaching, they will then have to go with the devil into the eternal perdition of hell torments’

(cowulf,WHom_6:209.384)

(11)

And

þæt

sealt

þe

se

sacerd

þam

cilde

on

muð

deð,

þæt

and

the

salt

that

the

priest

the

child

in

mouth

does

that

getacnað

godcundne

wisdom;

betokens

divine

wisdom

‘and the salt that the priest puts in the mouth of the child, that betokens divine wisdom’                      (cowulf,WHom_8b:22.554)

All the examples in this set resemble correlative structures (cf. Mitchell 1985, §1887–1896; Sweet 1900, §372; Curme 1912): the first demonstrative, either se or ða, that heads a relative clause, is copied in the main clause where it serves as the subject of the main clause. This is shown in (9) and (10). Sometimes, the copy is not exact, as only the demonstrative is resumptivized, and the relative nominal head is omitted, as shown in (11). The main function of these LFDed structures is generalizing, i.e., the function that makes generalizing statements about groups of people, their characteristics, traits, etc. Contrast can be involved as well.4 For instance, in Example (10), people who follow the devil’s teaching are contrasted in the immediately preceding context with those who follow God’s instruction (Appendix B (i)). The former will be punished in hell, whereas the latter will rejoice in the heavenly kingdom. Another common function is to introduce new topics in the discourse or reintroduce them after a long gap. Thus, sealt in (11) is mentioned for the first time. The anaphoric function is fairly rare (only 4 such instances). One clear example is given in (12), where the underlined demonstratives refer to Christ. Consider the following:

(12)

þæt

se

ðe

ah

geweald

heofones

&

eorðan

&

ealra

that

he

that

owns

power

heaven

and

earth

and

all

gesceafta,

se

let

hine

sylfne

for

ure

neode

þam

earmlicestan

creatures

he

let

him

self

for

our

need

that

miserable

deaþe

lichamlice

acwellan?

death

bodily

kill

‘that he who rules over heaven & earth and all creatures he let himself bodily be killed for our need by that miserable death?’       (cowulf,WHom_7:59.425)

Anaphoric demonstratives in OE are normally topic-shifters, as already found in a number of constructions (cf. Mitchell 1985, p. 320 and, more recently, Los and van Kemenade 2018 and Bartnik forthcoming for LFD structures). Indeed, in the immediately preceding context of Example (12), we find five non-nominative pronominal forms, i.e., hine and him, referring to our Savior (Appendix B (ii)). Thus, the topic shifts from men, who put Jesus to death, to Christ, who agreed to this death. Similarly, the other anaphoric examples exhibit topic-shifting as well, as the referents of the demonstrative resumptives are in a non-nominative case in the preceding context. One interesting example is presented below:

(13)

Se

sylfa

deofol

þe

on

helle

is,

þæt

is

se

þe

þonne

the

same

devil

that

in

hell

is

that

is

he

that

then

wyrð

on

þam

earmsceapenan

men

Antecriste

will-be

in

the

wretched

man

Antichrist

‘the same devil who is in hell, that is the one who will then be within the wretched man Antichrist’                   (cowulf,WHom_4:71.148)

In (13), the nominal relative head is resumptivized by the demonstrative þæt and the reference is specific (the devil). In the preceding context, we find a non-nominative form, þæne deofol, which suggests topic-shifting. However, it is worth noting that homily IV, De Temporibus Anticristi, from the YCOE cowulf.o34 file, is based on manuscript H, the latest manuscript from the second half of the twelfth century. However, this version omits a large portion of the text, which describes the conflict between two apostles, Peter and Paul, and the magician Simon Magus, which exemplifies the wrongdoings that the Antichrist can resort to (see Appendix B (iii) for the omitted excerpt). The other two surviving manuscripts, C and E, contain this passage. What is important, though, is that the missing exemplum is placed right above Example (13).5 This means that in C and E, the preceding context for topic-shifting is different. In particular, there are three nominative forms, both nominal and pronominal, referring to the devil. This reminds us that corpus texts, however extensive, can be incomplete, and the textual selection and edition can influence the conclusions we draw.

Finally, there are two examples of non-nominative demonstrative resumptives. They are illustrated below (Example 1 is repeated as 14 for convenience):6

(14)

þa

þonne

þe

his

leasungum

gelyfað

&

him

to

those

then

that

his

lies

believe

and

him

to

gebugað

þam

he

byrhð

her

for

worulde

turn-back

those

he

protect

here

for

world

‘those however who believe in his lies and turn back to him, them he protects here before the world’                  (cowulf,WHom_4:40.123)

(15)

þa

þe

he

elles

beswican

ne

mæg,

þa

he

wyle

those

that

he

otherwise

deceive

not

may

those

he

will

neadunga

genydan,

gyf

he

mæg,

þæt

hi

Godes

ætsacan

forcibly

compel

if

he

may

that

they

God

should-deny

&

him

to

gebugan.

and

him

to

should-bow

‘those whom he may not deceive otherwise, them he will compel by force, if he may, so that they renounce God and worship him’   (cowulf,WHom_4:43.129)

As expected, the non-nominative demonstratives in (14) and (15) land high, i.e., above the pronominal subject (cf. Bartnik 2024). The constructions are generalizing, with a contrastive function. Specifically, those who believe in the devil are contrasted in the preceding context with those who are clean and go to heaven (Example (14)). In (15), those who resist devilish tricks are juxtaposed with those who are easily charmed by his illusions.

In sum, the LFDed demonstrative system in Wulfstan’s homilies does not seem to differ from the one used in other texts. Quantitatively, the unusually high number of demonstrative resumptives is accounted for by the repetition of the same formulaic structure with subjunctive forms in one homily. Though they are not a slavish translation of the Latin text, they are formulaic because they contain the same, repetitive structure. Functionally, the overwhelming majority of examples are generalizing because LFDed structures introduce new discourse topics. They can be accompanied by contrast, especially high-landing non-nominative demonstrative resumptives. When anaphoric reference is involved, demonstratives are topic-shifters. When a larger context is taken into consideration, it is important to check whether we are not dealing with a truncated version of a (corpus) text, because this fact might influence the conclusions we draw.

Let us now turn to personal pronoun resumptives in LFDed structures in Wulfstan’s writings.

3.2. Personal Pronouns

As already shown in Table 1, personal pronouns are not accumulated in one homily. Rather, they can be found across the majority of homilies, with the greatest number being found in homily VI (six examples) and VII (five examples). Some typical examples are given below:

(16)

se

ðe

forsyhð

eow,

witod

he

forsyhð

me.

he

that

forsakes

you

truly

he

forsakes

me

‘he who forsakes you, he truly forsakes me’                                     (cowulf,WHom_17:55.1402)

(17)

Se

ðe

Godes

cyrican,

he

cwæð,

rype

oððe

he

that

God’s

churches

he

said

should-rob

or

reafige

oððe

halignessa

grið

scyrde

oððe

wyrde

should-plunder

or

holiness

sanctuary

should-injure

or

should-violate

a

he

forwyrðe;

always

he

should-perish

‘He who should rob or plunder God’s churches, he said, or injure or violate the sanctuary offered by holy places, may he perish forever’

(cowulf,WHom_10c:51.867)

(18)

ealle

þa

þing

þe

beoð

fram

þære

circan

afyrsode

buton

all

the

things

that

are

from

the

church

removed

without

tweon

hi

beoð

Criste

ætbrodene.

doubt

they

are

Christ

taken-away

‘all the things that are removed from the church, without doubt they are taken away from Christ’              (cowulf,WHom_10b:35.824)

(19)

And

se

man

þe

bið

bedæled

ealra

ðissa

seofan

gifa

nis

he

And

the

man

that

is

deprived

all

these

seven

gifts

is-not

he

na

Gode

wyrð

no

God

dear

‘and the man who is deprived of all these seven gifts, he is not dear to God’

(cowulf,WHom_9:51.711)

There are several points to make when we confront LFD with personal pronouns and demonstratives in Wulfstan’s writings. First, the choice of relative heads in LFD is similar to the one witnessed in Section 3.1: the relatives are mostly headed by demonstratives, like in (16) and (17), but nominal heads are also possible, as illustrated by (18) and (19). Second, the subjunctive mood found with all the verbs except for cweðan in (17) is found across a few homilies. Since it is instantiated only five times, it does not create an artificial increase in the number of LFDed structures with pronominal resumptives. Interestingly, Example (17) is the only subjunctive example found in homily Xc (compared to 15 with demonstratives discussed above). In other words, homily Xc contains only LFDs with the subjunctive mood. Moreover, the subjunctive form forwyrðe does not come before the resumptive, as was the case in the formulaic set in Section 3.1. Its Latin equivalent from homily Xb reads as follows:

(20)	Si	quis	ecclesiam	Dei	denudauerit	uel	sanctimonia
	if	anyone	church	God	should-plunder	or	sanctuary
	uiolauerit,		anathema	sit;
	should-violate		anathema	should-be
	‘If anyone should plunder the church of God or violate its sanctuary, let him be anathema’         (Wulfstan, Xb 41–43; Bethurum 1957, p. 195)

As evidenced in (20), the verbal moods are mirrored in the OE translation: both denudauerit and uiolauerit in the subordinate part as well as sit in the main clause are subjunctives. At the same time, however, (20) does not seem to be an exact translation of the Latin equivalent, since the original does not even start with a relative clause; rather, it starts with a conditional structure. The third point concerns the lack of correlative similarity in pronominal structures. While demonstrative resumptives are always repeated in the relative heads, this is not the case with personal pronouns because personal pronouns do not head relative clauses in the LFDed structures above (cf. Allen 2022).

Functionally, the pronominal system resembles the demonstrative one. Most examples are clearly generic and indefinite, as shown by Examples (16)–(19) above. Contrast is not common, though one clear example is presented below. Consider:

(21)

Se

ðe

hine

aht

þisses

tweoð

&

his

gelyfan

he

that

himself

anything

this

doubts

and

it

believe

nele,

ne

cymð

he

æfre

to

Godes

rice

gyf

he

on

not-wish

not

will-come

he

ever

to

God’s

kingdom

if

he

in

þam

geendað.

And

se

ðe

ðonne

rihtne

geleafan

hæfð

&

that

ends

And

he

that

then

right

faith

has

and

his

ealles

gelyfð

þæs

ðe

ic

rehte

(..),

he

it

completely

believes

of-that

that

I

explained

(..),

he

þæs

habban

sceal

ece

edlean

on

Godes

rice.

of-that

have

shall

eternal

reward

in

God’s

kingdom

‘he who doubts any part of this and does not wish to believe in it, he will never come to God’s kingdom if he ends his life that way. And he who then keeps the right faith and believes completely in what I have explained, (…) he will have as a result an eternal reward in God’s kingdom’ (cowulf,WHom_7:159.500/501)

In Example (21), two groups of people are contrasted through the generic he: those who doubt and do not believe in Christ’s teaching and those who keep faith and believe in it. The former will not enter the heavenly kingdom, whereas the latter will receive the eternal reward.

Anaphoric reference is very limited and found only with three examples. Anaphoric personal pronouns refer to specific individuals: Christ, his disciples, and God Almighty. One example is given below:

(22)

Nu

is

mænig

ungelæred

man

þe

wile

þencan

hu

now

is

many

unlearned

men

that

will

think

how

þæt

beon

mæg

þæt

se

ðe

gescop

on

fruman

that

be

may

that

he

that

created

in

beginning

ealle

gesceafta,

þæt

he

wearð

þus

late

geboren,

all

creatures

that

he

was

thus

late

born

‘now there are many unlearned men who will question how it may be that he who made all creation in the beginning, he was thus born late’ (cowulf,WHom_6:134.328)

LFDed structures with personal pronouns exhibit topic continuity, i.e., a nominative topic that is mentioned in subsequent discourse in the main clause in the sense of Traugott (2007, p. 426). As shown in Appendix B (iv), Example (22) does show topic continuity because the main clause he is mentioned twice in the following passage. However, this is not always the case, as in one example, the topic (disciples) is not continued in the following context (cf. Appendix B (v)). Finally, it is difficult to verify the hypothesis that (non-nominative) personal pronouns tend to land low, i.e., below the subject. I have not found such examples in the data.

In sum, personal pronouns in LFD exhibit both structural and functional differences, though they seem to be quantitative rather than qualitative. Structurally, the subjunctive mood is rare but is found across a few homilies. It is not used in formulaic structures in the data of the study; nor does the verb have to come before the resumptive. The correlative structure of relative heads and resumptives is not preserved. Functionally, contrast is fairly rare, though possible. Generic reference prevails but a couple of examples with anaphoric and specific reference are found. In such cases, topic continuity is found, though it is not obligatory.

4. Conclusions

In this paper, I have examined LFD with demonstratives and personal pronouns in Wulfstan’s homilies. Despite superficial differences, the system is similar to the one found in other OE texts. On the whole, personal pronouns are more common—if we subtract the subjunctive structures from homily Xc—and found in more texts than demonstratives. This means that they are the default option in LFD (Traugott 2007). LFD with personal pronouns is also less restricted grammatically, as shown by the subjunctive examples. Functionally, LFDed structures are predominantly generalizing because they introduce new topics; the contrastive function can accompany LFD with demonstrative resumptives. Non-nominative high demonstratives are topic-shifters but there are few such examples. Personal pronoun resumptives mark topic continuity, though again, the empirical material is rather scant. Some observations found in the literature are unverifiable, as there are no examples supporting them. For instance, there are no non-nominative personal pronoun resumptives; so, we cannot tell whether they normally land low (cf. Bartnik forthcoming).

A note on corpus searches deserves a separate comment. The YCOE is a valuable tool, but it has its limitations too. This investigation shows two such problems. First, copies of the same manuscript may contain different types of resumptives. It is important to determine whether or not this variation reflects the use of different types of LFD, i.e., CLD or HTLD and, consequently, different pragmatic functions. Second, pragmatic analysis often covers wider contexts. We need to make sure that the text we examine is complete if we want to identify discourse functions correctly. Otherwise, the preceding or following context may influence the results we obtain.