Microsoft Word - 2. Proofs_Allen.docx

SELIM. Journal of the Spanish Society for Medieval English Language and Literature. 28 (2023): 19–38.
ISSN: 1132-631X
DOI: 10.17811/selim.28.2023.19-38

© Ediuno. Ediciones de la Universidad de Oviedo.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0
International License.

What Did(n’t) Happen to English? A Re-evaluation of Some

Contact Explanations in Early English

Cynthia L. Allen
ANU College of Arts and Social Sciences

(Received 19 May 2023; revised 21 July 2023)

McWhorter (2002) argued that contact with Norse caused simplifications in English
grammar that set English apart from other Germanic languages. This paper focuses on one
of the losses McWhorter attributed to the linguistic impact of the Scandinavian invasions,
External Possessors. An investigation of electronic Old and Early Middle English corpora
reveals that the construction was already on the decline in the Old English period, and that
Norse contact cannot explain the Early Middle English data. There is no support for the view
that the loss of the construction spread from the Scandinavianized areas southwards. The
facts are consistent with the view that while Celtic influence did not cause the loss of the
construction in Old English, Celtic speakers shifting to English may have played a role in
triggering the initial decline of the construction. Study of non-standard variants of other
Germanic languages is needed to increase our understanding of the history of External
Possessor constructions in those languages.1

Keywords: External possessor; language contact; Norse; Celtic; Germanic languages

1. Introduction

In 2002, John McWhorter published a thought-provoking article with the title “What
Happened to English?”. McWhorter’s basic thesis was that the Scandinavian invasions
that took place in England caused a general simplifying trend in English grammar setting
it apart from all other Germanic languages. The language the Scandinavian invaders
spoke was not very differentiated at this time, and this paper will follow common usage
in referring to it as Norse and use “Norse hypothesis” for the idea that it was contact with
Norse that caused these simplifications. Since there have been significant developments
in our understanding of the history of English in the twenty years since McWhorter’s
article was published, it is worth re-evaluating McWhorter’s ideas about what happened
to English. Such a re-evaluation must rely on detailed case studies of the individual
features like those that McWhorter looked at, making use of electronic resources that
were either very new or non-existent when McWhorter’s article was published, and
which did not play a role in his presentation of the facts.

This paper looks in detail at the loss of one of McWhorter’s features, the External
Possessor construction. Before the discussion of this construction, Section 2 makes some
brief comments about developments since the publication of McWhorter’s article that
justify new scrutiny of the conclusions presented there. Section 3 summarizes

1 I am grateful to George Walkden for comments on an earlier draft of this paper. The usual
disclaimers apply.

https://orcid.org/0000-0002-1904-4666

CYNTHIA ALLEN

McWhorter’s paper. Following the introduction to External Possessors in Section 4,
Section 5 looks in more detail at the Norse Hypothesis, and Section 6 evaluates this
hypothesis in light of Old and Early Middle English data. Section 7 considers an
alternative explanation for the loss of the External Possessor construction that
McWhorter dismissed. Some further observations and concluding remarks are made in
Section 8.

2. Recent developments

Two recent developments that have deepened our understanding of what happened to
English are the explosion of digital resources and the increased scholarship in contact
linguistics. An example of how digital resources have played a central role in upending
widely held assumptions about the impact of Norse on English comes from the history
of the plural personal pronouns. Until recently, it was nearly universally assumed that
they, their, and them are borrowings from Norse. For example, in mentioning that these
pronouns had not yet spread into the Southwest Midlands in the period covered by his
scholarly and careful investigation of Scandinavian influence in the Southwest Midlands,
Richard Dance indicates without comment that they derives from Old Norse þeir (2003,
308–9). This widely-assumed etymology has been used as evidence of very intimate
degree of contact between the Scandinavian and English linguistic communities.

However, it has long been recognized that the vocalism of some forms of the pronouns
that are found in Middle English texts is hard to explain by this etymology, and Cole
(2018) makes a convincing case for a native origin in Northumbrian: a merger of the
personal pronouns starting with h- and the demonstrative pronouns starting with a
dental fricative. A digital resource covering many texts not in print, Laing’s (2013–)
Linguistic Atlas of Early Middle English (LAEME), enabled Cole to make a major
empirical contribution by carrying out the most extensive study to date of the diffusion
of the they forms across the focal areas of Anglo-Scandinavian contact. Cole’s conclusion
is that this pattern of diffusion is highly problematic for the assumption of the Norse
origin of these pronouns. It is fair to say that we no longer “know” one of the things that
we thought we knew about Norse effects on English. At the least, the publication of this
paper in a highly regarded journal necessitates some reconsideration of this tenet.

A native source for these pronouns would not by itself mean that there was not
intimate contact between the Scandinavians and the English. There is no reason to doubt
a significant amount of intermarriage, which would be expected to affect the language of
the children; see for example Trudgill (2011), and for archaeological evidence see
Kershaw and Røyrvik (2016). The point is simply that recent research, aided by digital
resources, has caused a re-evaluation of one of the things that everyone used to think
they knew had happened to English, but may well not have happened.

A second development, theoretical advances in the area of language contact, has
opened up new interpretations to the findings of earlier studies. In particular, until fairly
recently, the idea that contact with Celtic languages had any syntactic effects was rejected
by most linguists, including McWhorter (2002, 252). However, what is known as the
Celtic Hypothesis has since then become one that can no longer be ignored. It should be
noted that McWhorter (2009) abandons this stance and credits Celtic influence with
playing a role in some syntactic changes in English, specifically in the use of the auxiliary
do. However, in this later paper he maintains his view that Norse contact was responsible

What Did(n’t) Happen to English?

for the loss of features, whereas the new uses of do represents an instance of elaboration,
not simplification.

McWhorter’s paper does not specify what model of contact linguistics he is assuming.
Any discussion of language contact effects is best embedded in a theoretical framework,
however, and this paper assumes the model of contact effects of van Coetsem (1995) and
(2000). This model focuses on the role of speakers in the transmission of contact
phenomena, as opposed to older models that tend to focus on overall changes within
societies. One important point about this model is that it sharpens the distinction
between borrowing and other types of transmissions. In van Coetsem’s model, borrowing
refers only to changes initiated by speakers of the recipient language, for example when
English speakers borrowed words from French into their native language. In contrast to
this recipient language agentivity we have source language agentivity when speakers
import features of their native language into a language they are learning. This is termed
imposition. If Norse or Celtic speakers brought features of their native language into Old
English, that would be imposition. In earlier theories, “borrowing” has tended to be used
for any importation of a feature from one language into another, regardless of who
initiated it.

Van Coetsem does not assume that these are the only type of contact effects. Of
particular relevance here is his assumption that language-internal changes may be
triggered by contact by resulting in simplification in the grammar that may not be the
direct result of either imposition or borrowing. These are changes made possible by the
nature of the system, for example the smoothing out of irregularities in paradigms, and
are not part of the transfer proper.

An instance of paradigm levelling in Early Middle English that is likely to have been
promoted by contact happened with the determiner se. In most OE dialects, the
paradigm for this determiner was irregular in that the initial consonant was an alveolar
fricative in the nominative singular masculine and feminine se and seo, respectively, but
an interdental fricative in the rest of the paradigm, for example masculine accusative
singular þone. In Early Middle English, the interdental fricative was leveled out through
the paradigm, resulting in þe and þeo before the category distinctions of gender, case and
number were lost. As is well known, the new forms with the interdental fricative were
first found in northern areas, so contact is likely to have played some sort of role in this
development. However, this could not have been imposition by Norse speakers learning
English, because Old Norse had the same irregularity in its paradigm in the singular, as
illustrated in Table 1:

Table 1. Old Norse determiner paradigm (after Gordon and Taylor 1957, §109, §111).

Si
ng

ul
ar

Masculine Feminine Neuter
Nominative sá sú þæt
Accusative þann þá þæt
Genitive þess þeir(r)ar þess
Dative þeim þeir(r)i þ(v)i

Pl
ur

Masculine Feminine Neuter
Nominative þeir þær þau
Accusative þá þær þau
Genitive þeir(r)a þeir(r)a þeir(r)a
Dative þeim þeim þeim

CYNTHIA ALLEN

Norse speakers should have had no trouble with an irregularity that paralleled their
own language. But this sort of regularization, which mirrors the regularization that
children attempt when acquiring their native language, is the sort of thing that happens
when the enforcement of norms is weakened. As Milroy and Llamas (2013) note, contact
situations typically result in weaker social ties, and weaker social ties weaken linguistic
norms and promote language change.

McWhorter (2002, 266) comments that it is hard to avoid the conclusion that the
timing of the simplifications in English was due to poor language learning abilities on
the part of adults. However, if these changes are to be attributed to Norse contact,
McWhorter’s arguments seem more in line with an explanation in terms of one of
contact-induced internal simplification than with poor adult language-learning, as we
shall see. Let us turn now to a summary of McWhorter’s paper before considering in
detail one of the constructions he discusses.

3. McWhorter (2002)

McWhorter’s (2002) basic thesis is that language change can generally be classified as
either complexification or simplification, with Scandinavian contact being responsible
for the loss in English of a number of morphosyntactic features inherited from Common
Germanic by all early Germanic languages. McWhorter supplies a table of 10 such
features. In this table, the shaded cells represent the retention of a given feature in a
language, and a white cell indicates the loss of that feature. The table indicates that
English has lost every one of these features, losses that McWhorter views as
simplifications.

Table 2. Losses in English compared to other Germanic languages (McWhorter 2002 Table 3, copied by
permission).2

As McWhorter notes, English comes out as having lost more of these Germanic
features than Afrikaans, widely considered a “semi-creole”, but which he nevertheless

2 Key to abbreviations for languages: G=German, Du=Dutch, Y=Yiddish, Fr=Frisian, S=Swedish,
N=Norwegian, Da=Danish, I=Icelandic, A=Afrikaans, OE=Old English, E=(Modern) English.

G Du Y Fr S N Da I Fa A OE E

inherent reflexives

external possessors

gender beyond noun

loss of prefixes

directional adverbs

be-perfect

passive become verb

singular you

indefinite pronoun

What Did(n’t) Happen to English?

deems a “well-behaved” Germanic language compared with English on the basis of its
retention of more of these features. McWhorter (2002, 265) makes the sweeping claim
that “where a subset of Germanic languages have departed sharply from the original
Germanic ‘typology’, English never fails to be a member”. However, in this comparison,
McWhorter fails to mention one distinctively Germanic feature that Afrikaans has lost
but English has retained, namely the distinction between strong and weak verbs
(Donaldson 1994, 495). As Donaldson notes, this change is also found in some other
Germanic varieties, including Yiddish and Pennsylvania German. English is thus more
typically Germanic than these languages in this respect.

4. (Dative) External Possessors

It is beyond the scope of this paper to comment on all of McWhorter’s features, and I will
focus on the results of an extensive corpus-based investigation of the loss of External
Possessors (EPs). The West Germanic type of EP discussed by McWhorter is generally
known as the Dative External Possessor (DEP), e.g. by Haspelmath (1999). This term is
used here except when including other types of EPs in the discussion. As will be discussed
in Section 5, North Germanic languages also have an EP, but of a different type. The Old
English DEP is particularly interesting because both Norse and Celtic contact have been
argued to have been responsible for its loss. First, what are DEPs? This construction in
German is illustrated in (1):

(1) Die Königin schlug dem König den Kopf ab
The queen:NOM cut the:DAT king:DAT the:ACC head:ACC off
‘The queen cut off the king’s head’ (Vennemann 2002, 206, ex. 1)

Here, the possessor of the head is in the dative case, so the dative part of the DEP is
obvious. The external part is less obvious. The idea is that the possessor König is an
argument external to the phase that contains the possessum den Kopf. This contrasts
with the internal possessor (IP) found in the English translation, where the king’s and
head are dominated by the same phrase. The difference is illustrated schematically in
(2):

(2) a. DEP (German) b. IP (Modern English)

These trees do not represent a proper analysis of the two constructions, the details of
which differ greatly on the syntactic theory employed, but give an overall picture of the
main difference between the two constructions. Triangles cover the internal structure of
the Noun Phrases. As McWhorter (2002) reports, Old English had such a construction:

CYNTHIA ALLEN

(3) Seo cwen het þa ðæm cyninge þæt heafod of aceorfan
the queen ordered then the:DAT king:DAT the:ACC head:ACC off cut
‘the queen then ordered the king’s head cut off’
(coorosiu,Or 2:4.45.6.852)

This example is parallel to the German example in having an EP in the dative case. It
is also parallel in that the possessum is the direct object of the verb, in the accusative
case. DEPs are also found with subject possessa:

(4) Ac him ða eagan of his heafde ascuton
But him:DAT the:NOM eyes:NOM from his head out.shot
‘but then his eyes shot out of his head’
(cobede,Bede_1:7.40.7.332)

The discussion in this paper will be restricted to (nominative) subject and (accusative)
object possessa. DEPs were also found in examples in which the possessum played the
role of object of a preposition, parallel to Modern English fixed expressions like look x in
the eye and stare x in the face, but this construction was not subject to the affectedness
restriction that will be discussed presently. Statements made here about Old English
DEPs should therefore be interpreted as limited to subject and object possessa. The
discussion will also be restricted to body part possessa, as these were by far the most
common possessa in DEPS.3

I make the usual assumption that, however it is to be analysed, the Old English DEP
was structurally parallel to the Modern German one. It was also parallel in putting the
focus on the effect on the possessor of the body part of an action, rather than just on what
was done to the body part. However, the evidence of the texts indicates two important
differences in the use of the DEP in the two languages.

First, the use of the DEP in Old English with subject and object possessa was nearly
completely limited to negative effects on the possessor, as in (3) and (4). My investigation
only found one example where a positive effect on the possessor is clear, involving a body
“lightening” or becoming relieved of pain after a treatment, and one example describing
someone’s hair growing, where an effect is not clear.4

In Modern German, in contrast, DEPs are freely used to describe actions resulting in
beneficial effects, such as rubbing someone’s back (Lee-Schoenfeld 2006, 108). The
near-total lack of DEPs of this type cannot be a data gap, because examples of IPs in
which the effect of the possessor is clearly positive are easy to find:5

3For a discussion of possessa in the role of object of preposition as well as possessa referring to
the mind or soul, see Allen (2019).
4 Space does not allow for a presentation of the examples, but for a discussion see Allen (2019,
74–5).
5 In examples with IPs, the genitive components are generally clear without glossing of inflectional
features, and these features are only glossed in this paper when such glossing is likely to be helpful
in understanding the syntax of the sentence, as in this example. In contrast, fuller glossing is given
in examples of DEPs.
The DEP equivalent of (5) would be … þæt we gefyllon þæm þearfan þa wambe.

What Did(n’t) Happen to English?

(5) … þæt we gefyllon þæs þearfan wambe
… that we fill the:MASC.GEN.SG poor:MASC.GEN.SG belly:(F).ACC.SG
mid urum godum?
with our:DAT.PL riches:DAT.PL

‘… that we fill the poor man’s belly with our riches?’
(coblick,HomS_14_[BlHom_4]:39.19.521)

The second important difference is that even in situations that most favoured the DEP
in Old English, i.e. with negatively affected possessors, that construction was in variation
with the IP. In the same text in which (3) is found, we find (6):6

(6) hie het gebindan, & … mid æxsum heora heafda of

them:ACC ordered bind and with axes their heads off
aceorfan
cut
‘(He) ordered them to be bound, and …their heads to be cut off with axes’
(coorosiu,Or 2:3.40.18.766)

This is not the case in Modern German, according to Theo Vennemann, a native
speaker of German, who states that “the dative is obligatory for affected possessors in
German” (2002, 208). The use of an IP in Modern German gives a clinical account of an
effect to a body part without indicating that the possessor of that body part was affected.
The situation in modern German is clearly different from earlier German as described by
Havers (1911), whose examples show that both the IP and the DEP were used in Old High
German to express possessors who were clearly affected, and that in fact this variation is
ancient in the Indo-European languages. The two constructions did not express different
situations, but gave a rather different perspective on them. Both English and German
have changed from what seems to have been the earlier situation: while English has
changed by dropping a construction that was in variation with another, German has
changed by sharpening the distinction between the two constructions.
Summing up so far, the DEP in Old English can be seen as a marked way of expressing
possession, especially of body parts, since IPs were the normal way of expressing
adnominal possession in Old English generally. The DEP was an option to use to
highlight the negative effect on the possessor. With this basic background information
on the DEP in Old English, we can turn our attention to explanations for the loss of this
construction, starting with the hypothesis that this development was caused by contact
with Norse.

5. The Norse hypothesis

Let us begin looking at how the mechanics of what I will call the Norse hypothesis might
work. McWhorter attributes the loss of the EP in English to the disparity between the
type of EP that Old English inherited from West Germanic with the Scandinavian EP. As
discussed above, in the Old English type, the possessor was in the dative case. In the

6 This is not to suggest that the constructions were equally common in similar situations. The
mention of a beheading in particular nearly always brought on DEP; the IP of (6) is a bit unusual,
but as will be discussed, IPs were by no means rare overall with negatively affected possessors.

CYNTHIA ALLEN

modern Scandinavian languages, the possessor is typically encoded in a locative phrase,
as in the Norwegian example given in (7):

(7) Jeg brekker armen på ham

I break arm.DEF on him
‘I break his arm’ (Lødrup 2019, ex. 15)

As McWhorter mentions, an apparent stumbling block to the Norse hypothesis is the

fact that Old Norse already had this type of EP:

(8) hvé þar á Herkju hendr sviþnuþu
how there on Herkja hands:NOM.PL burned
‘How Herkja’s hands burned there’

(Gþr. III 10, 2 as cited in Havers 1911, 269)

Assuming with McWhorter that the Norse speakers learning Old English had this
construction in their grammars, we might expect that they would impose their own EP
construction on their English, with the result that Modern English speakers would be
saying things like I broke the arm on him instead of I broke his arm. McWhorter argues
that this is not really a problem for the Norse hypothesis:

Faced with the disjunction between external possessor encoding in the two languages,
Scandinavians may have taken the choice of eliminating the distinction altogether, given that
it was not vital to the expression of the relevant concepts. (McWhorter 2002, 257–8)

Using van Coetsem’s terminology, it is clear that Norse speakers did not impose this
construction on English, and indeed the contact situation between Norse and English
speakers would not lead us to expect such a development. Imposition is typical of
language shift, especially rapid shift, that is accompanied by poor language acquisition,
and in arguing against the Norse hypothesis, Peter Trudgill (2011, 53) notes that the
adstratal contact situation with Old English and Norse should not have resulted in poor
acquisition assumed by McWhorter. It should also be noted that simple imposition
would not have resulted in the loss of the DEP, since Old Norse had a DEP just like that
of Old English, e.g. (9):

(9) hrafnar skulu þer … slíta sjónir ór

ravens shall thee:DAT tear eyes:ACC.PL out
‘ravens shall tear your eyes out’
(Fj. 45,1 as cited in Havers 1911, 268)

If Scandinavians who were learning Old English had this construction in their
language, there is no reason why they could not have used the construction in their
English.

However, it is certainly possible that in the early contact period before the Norse
speakers had such intensive contact with Old English speakers, the speakers of both
languages might have simplified their interactions by limiting themselves to the default
possessive construction in both languages, namely the IP. Then we could imagine that

What Did(n’t) Happen to English?

Old English speakers generally were exposed to a reduced frequency of DEPs and
reduced that frequency further, with the construction eventually dying out in the heavily
Scandinavianized areas. This would be a type of internal simplification of English
grammar. A problem arises, however, when we try to imagine how this Norsified syntax
supposedly spread to the dialects further south that became the basis for Standard
British English. There is no evidence whatsoever of such a spread of a DEP-less grammar.

With these considerations in mind, we turn to the corpus-based investigation of the
DEP and IP in Early Middle English.

6. The data

The data presented here are a slightly modified subset of the data resulting from the
corpus-based investigation of Allen (2019). It is first necessary to describe the study
briefly. Since IPs were the normal way of expressing possession generally in Old English,
as in other languages that also have DEPs, it would not be illuminating simply to compare
the overall numbers of the two constructions. Instead, the focus needed to be on
collecting DEPs in the circumstances where they are most likely to be found, in order to
see how the frequency of DEPs compared with that of IPs in those circumstances. This
focus of the study on body parts is justified by the fact that Payne and Barshi (1999) place
body parts at the top of the hierarchy of accessibility of possessa to EPs, as well as by
initial observation of the texts. The list of body forms referring to body parts used in the
investigation was drawn up using lexicon queries run on the York-Helsinki Parsed
Corpus of Old English Prose (YCOE, Taylor et al. 2003), The York Poetry Corpus (Pintzuk
and Plug 2001), and the Penn-Helsinki Parsed Corpus of Middle English 2 (PPCME2,
Kroch and Taylor 2000).7

Since the DEPs found in searches using these words were nearly completely confined
to descriptions of events with negative effects on the possessors, the next step was to
compile a list of transitive verbs describing an action that would usually have a such a
negative effect, such as cutting and stabbing. Another list of intransitive verbs that
described a negative state or action was used for subject possessa. Queries were then run
on the three corpora to collect examples of DEPs and IPs. The results for object possessa
in Old English are given in Table 3.

Table 3. IPs and DEPs in Old English texts: object possessa with highly affecting verbs.

IP DEP Blended
Poetry Total 11 7 0
EWS 9 12 1
Other Early 0 2 1
9thC(OE) 22 13 3
General OE 45 11 1
LWS 80 4 0
LWS(Late) 29 3 0
Prose total 185 45 6

7 See the Appendix for the texts included in the study. Supplementary online resources used
include the Thesaurus of Old English (Roberts et al. 2017), An Anglo-Saxon Dictionary Online
(Bosworth et al. 2014), and the Middle English Dictionary (Kurath and Kuhn 1956).

CYNTHIA ALLEN

Some explanations are needed here. IP and DEP have already been explained, but the
rightmost column is for a third type not yet mentioned, which is traditionally seen as a
blend between the two, exemplified in (10), where the dative Leone is blended with the
possessive his.

(10) Her Romane Leone þæm papan his tungon forcurfon &

Here Romans Leo:DAT the:DAT pope:DAT his tongue:ACC cut and
his eagan astungon,
his eyes:ACC stabbed
‘In this year the Romans cut out Pope Leo’s tongue and put out his eyes’

(cochronA-1,ChronA_[Plummer]:797.1.597)

Nothing more will be said about this construction here, except to record numbers. I
will use DEP to refer to the pure DEP type with only a dative, but EP to cover both the
pure DEP construction and the blended type.

The text types of the table also need some explanation. These types are groupings of
the texts included in the YCOE according to time and dialect, where possible. It should
also be mentioned that the study excluded some texts that are included in the YCOE but
are based on editions completely or largely based on copies of much earlier compositions,
and also some texts that were actually composed in the twelfth century, which should
actually be treated as Early Middle English rather than late Old English.

The bulk of the Old English texts are in a West Saxon dialect, and the labels EWS and
LWS stand for Early and Late West Saxon, respectively. The EWS group comprises only
the West Saxon texts that are agreed to have been written by the early tenth century,
meaning that for the Anglo-Saxon Chronicle, only the annals up to 924 are included.
Texts agreed to have been composed in Early West Saxon but found only in much later
manuscripts are excluded. LWS(late) means that the text was composed in Late West
Saxon (primarily by Ælfric or Wulfstan), but is now found only in a manuscript from
considerably later. The reason for this division of the Late West Saxon texts, as with the
exclusions, is to try to keep the data as clean as possible.

The label “poetry” is clear, but it should be mentioned that the York Poetry Corpus
does not include all Old English poetry and is not exhaustive for all the poems that it
contains parts of. I have therefore supplemented it with my own reading to make the
coverage of the larger poems complete. “Other Early” includes texts from up to the EWS
period but not in the West Saxon dialect. “9thC(OE)” includes Alfred’s laws as well as the
Leechbook and the two versions of Gregory’s Dialogues, which were presumably
composed in the late ninth century in Alfred’s court but are only found in manuscripts
that are dated considerably later but before the third quarter of the eleventh century.
“General OE” comprises two main types of writings: those that cannot be dated very
precisely but are found in manuscripts from before the same deadline and some that can
be dated precisely as late Old English, but are mixed as to dialect, such as Cnut’s laws.8
In some instances, I have split the YCOE files up and put the contents into different
categories because the editions of the texts use different manuscripts. For example, part
of the translation of Bede’s Ecclesiastical History goes into the Other Early category
because it is from an early manuscript, while the remainder goes into the General OE

8 YCOE files colaw1cn.o3 and colaw2cn.o3. The YCOE gives the dialect of both these sets of laws
as West Saxon/X.

What Did(n’t) Happen to English?

category because it is found only in later manuscripts. Similarly, some of Ælfric’s
Homilies found in Pope’s (1967) edition of supplementary homilies go into LWS, while I
have put others in the LWS(late) category. This may seem overly fussy, and the lag
between composition and manuscript dates may not be important for the syntax under
investigation, but such exclusions reduce as much noise in the data as possible.

Looking at the figures in Table 3, we see that IPs outnumbered DEPs except in the two
earlier prose categories, where numbers are not overwhelming. Note that IPs were
slightly more common than DEPs in the poetry. This is important because the syntax of
this traditional Germanic genre is the least likely to be affected by contact phenomena.
There is a sharp decline in the incidence of DEPs in the Late West Saxon period. A caveat
is necessary, however, because we are not dealing with random samples here. In
particular, the LWS texts are dominated by the works of Ælfric. It is not impossible that
Ælfric’s preference for IPs was a personal one that does not reflect a general decline in
the late Old English period. However, the decline in LWS hardly comes out of the blue.
As noted, the DEP is less frequent than the IP with object possessa in all earlier text types
except EWS. Dominant patterns tend to become more so, and it is plausible that the DEP
had become a marked construction in late Old English and was reserved for particularly
drastic effects.

Turning to subject possessa, the occurrence of IPs, DEPs, and blended constructions
is given in Table 4. One salient fact is that subject possessa, in any construction, were
less common than object possessa. The second thing to notice is that when compared to
IPs, DEPs were proportionally more common with subject possessa than with object
possessa, all through the Old English period, although there is again a big drop in the
incidence of DEPs in LWS.

Table 4. Subject possessa with highly affecting verbs.

IP DEP Blended
Poetry Total 6 5 0
EWS 4 4 0
Other Early 1 1 0
9thC(OE) 8 17 2
General OE 24 18 2
LWS 39 5 1
LWS(Late) 2 4 2
OE Prose total 78 49 7

Note that DEPs actually outnumber IPs in the LWS(late) category. This is not likely

to be because scribes who were copying works by Ælfric or Wulfstan were changing IPs
in the original into DEPs. It is more likely to be due to an accident of the nature of the
examples. It so happens that three of the four DEPs in LWS(late) and both blended
examples are accounts of people’s bowels falling out, either as a punishment from God
or the result of a disembowelment:

(11) swa þæt him aeode ut eall his innoð togædere

so that him:DAT went out all his innards together
‘so that his bowels all fell out together’

(coaelhom,ÆHom_10:159.1490)

CYNTHIA ALLEN

That is, five out of the six examples of EPs found in the latest texts are possessors who
have lost their bowels. One of Ælfric’s few examples of a DEP in his texts in my LWS
category is similar to (11). Combining Ælfric’s works in the two text categories, we find
six examples describing people losing their bowels, and all six have a dative, in either a
pure DEP or a blend. This may or may not be significant, but it illustrates how a small
number of examples may influence the picture; it is likely that EPs were used more
commonly in descriptions of some drastic effects than others. At any rate, the numbers
in the LWS(late) category are too small to give any evidence of a trend.
Summing up, we see that these tables suggest a marked decrease in the use of DEPs
by late Old English in the texts available to us from the dialect least likely to be influenced
by Norse. This decline cannot be due to Norse contact.
Now let us consider the facts available from Early Middle English. Unfortunately, after
the Norman Conquest, original English texts become sparse until the thirteenth century.
“Original texts” refers to texts composed for the first time in English. Texts written in
English are not scarce in the twelfth century, but with a few precious exceptions, they are
copies or adaptations of texts composed in the Old English period. It is possible that
these scribes might not have updated syntax that was a bit old-fashioned but still possible
and perfectly comprehensible. For this reason, apart from a few comments on later texts,
this study is restricted to texts in the PPCME2 assigned to their m1 period, 1150-1250
and which are known or generally thought to have been composed after the middle of the
eleventh century; it does not cover texts designated by the PPCME2 as mx1.
The results are set out in Table 5. In this table, IPs are only reported for highly
affecting verbs; DEPs were only found with such verbs in these texts.9 The texts are
separated into a broad division between more and less Scandinavianized dialects
(northern and eastern versus southern and western, respectively). The reason for this is
that, as McWhorter mentions, the Norse hypothesis predicts that the DEP would last
longest in the least Scandinavianized areas. Unfortunately, texts from the north and
northeast are sparse in the earliest Middle English period.

Table 5. DEPs (all) vs. IPs (highly affecting verbs only), body words, m1 period.

9 Although it is no simple matter to determine whether these texts all retained a distinct dative
case category, these EPs are treated as dative here because they continue to appear in the positions
of EPs in Old English.

Affected Obj IP DEP Blended Total Affected Subj IP Dat Blended Total

cmancriw-1.m1 4 2 1 7 cmancriw-1.m1 1 0 0 1

cmancriw-2.m1 1 0 0 1 cmancriw-2.m1 3 0 0 3
cmhali.m1 2 0 0 2 cmhali.m1 6 0 1 7
cmjulia.m1 2 2 0 4 cmjulia.m1 2 0 0 2
cmkathe.m1 5 2 0 7 cmkathe.m1 0 0 0 0
cmkentho.m1 1 0 0 1 cmkentho.m1 0 0 0 0
cmlamb1.m1 0 0 0 0 cmlamb1.m1 0 0 0 0
cmmarga.m1 5 1 0 6 cmmarga.m1 5 0 0 5
cmsawles.m1 0 2 0 2 cmsawles.m1 0 0 0 0
cmvices1.m1 1 0 0 1 cmvices1.m1 1 0 0 1
Total South 21 9 1 31 Total South 18 0 1 19
cmorm.po.m1 6 0 1 7 cmorm.po.m1 1 1 0 2

What Did(n’t) Happen to English?

At first glance, the prediction that DEPs have nearly disappeared in the

Scandinavianized areas seems to hold. In the PPCME2 sample (53,182 words) of the
Ormulum of around 1150, which can be precisely located to Bourne, South Lincolnshire,
the queries only found one example of a pure DEP:

(12) & all himm wærenn fet & þeos Tobollenn & toblawenn.

And all him were feet and thighs puffed.up and swollen
‘and his feet and thighs were all puffed up and swollen’

(CMORM,I,280.2293)

There is also a blended example, not presented here. But this small number of EPs
tells us nothing about the frequency of IPs and DEPs in the Ormulum, because the
combination of highly affecting verbs and body parts was very small. The Continuations
of the Peterborough Chronicle yield one subject possessum DEP and no IPs. So we
cannot say that EPs are not in a minority in this small sample, and example (12) does not
seem like a fixed expression.

Turning to non-Scandinavianized areas, the texts that offer the largest number of
relevant examples are from the Southwest Midlands dialect AB; this is due not only to
the much larger size of the corpus in this dialect but also to the fact that several of the
texts are martyrologies. Here, we have eleven examples of EPs, when we combine the
pure DEP and the blended construction, and there can be little doubt that DEPs had
some productivity. Two examples are given in (13) and (14):

(13) ha duluen me þe fet & þe honden

they dug me the feet and the hands
‘they dug into my feet and hands’

(CMANCRIW-1,II.215.3101)
(14) & swipte hire of þt heaued

and swept her off the head
‘and swept her head off’

(CMKATHE,52.533)

With subject possessa, however, a decline from Old English is very apparent. While
EPs were used at a higher rate with subject possessa (42.65 per cent) than with object
possessa (21.61 per cent) in Old English prose overall, they are used in only 5.26 percent
of the southern m1 examples. There is also a striking disparity between the object and
subject possessa in m1—if EPs were used for object and subject possessa, we would
expect four examples of EPs with subject possessa.10 This disparity goes against the
Norse hypothesis—why should Norse speakers be able to learn to use DEPs with objects
and not with subjects? A differential decline of this sort seems more consistent with a
continuation of the reduced popularity of the DEP with subject possessa already evident

10 The numbers are small but large enough for a Fisher’s exact test, which yields a p-value of
0.03515, meaning that we can reject the hypothesis that the difference is due to chance.

cmpeterb.m1 0 0 0 0 cmpeterb.m1 0 1 0 1
Total North 6 0 1 7 Total North 1 2 1 3

CYNTHIA ALLEN

within the Old English period. A plausible explanation for the faster decline of DEPs with
subject possessa than with object possessa comes from the fact that early Old English
word order was more flexible in response to pragmatic conditions than was true later.
With subject possessa, a dative argument is typically placed at the front of the sentence,
and such flexibility was greatly reduced by the Early Middle English period. While the
topicalization of contrastive objects has remained into Present Day English, the fronting
of non-contrasting objects was greatly reduced in the Dialect AB texts (Allen 1995, 247),
and this may have led to less frequent sentence-initial dative possessors. In contrast, with
object possessa, both external and internal possessors were in positions typical of
objects, making it possible to use the DEP to emphasise the effect on the possessor
without resorting to unusual word order.

Whatever the explanation for the disparity between subject and object possessa DEPs,
by the m2 period (1250-1350), DEPs of subject and object body parts are limited to fixed
expressions in the PPCME2 texts.

Although it is not possible to be certain exactly how quickly EPs died out in English,
there is no evidence whatsoever that their loss had any relationship to the amount of
contact with Norse in a given dialect.11 Finally, it is difficult to explain the mechanics of
how this supposedly Northern and North Midlands syntactic innovation would have
spread south into Standard English. So one of the things that seems not to have happened
to English is that contact with Norse speakers triggered the loss of EPs in English.

7. Celtic

We turn now to contact with Celtic speakers as a possible explanation for the
disappearance of the DEP. The idea that a Celtic substratum had profound effects on Old
English grammar is not a new one, and the specific suggestion that the sole use of the IP
in English is due to Celtic influence goes back as far as Pokorny (1927, 253). The reason
why Celtic influence has some plausibility for this development is that the Brythonic
Celtic languages Welsh and Breton lack DEPs, using only IPs. However, until this century
the Celtic Hypothesis did not really enter the linguistic mainstream, mainly because of
the paucity of Celtic loanwords in the Old English period and also because of the general
belief that the Celts mostly fled or were wiped out during the Anglo-Saxon invasion,
leaving their numbers too few to affect English. More recent evidence points to a much
greater Celtic presence in England and it is also now better understood that a conquered
population forced to shift to a new language is likely to impose much of their grammar
but unlikely to bring much vocabulary into their new language.

Publications including by Tristram (2004), Filppula and associates (Filppula,
Klemola, and Paulasto 2008; Filppula, Klemola, and Pitkänen 2002) and Lutz (2009,
2011) have presented a case for the Celtic Hypothesis that at least deserves serious

11 It is also worth mentioning that when a text appears in versions in different dialects, we might
expect, by the Norse hypothesis, that the more northern versions would have IPs where southern
versions have DEPs. This does not seem to be the case. For example, the Ancrene Riwle in the
PPCME2 is from the Southwest Midlands manuscript Cotton Cleopatra C vi, but a more northerly
version is found in Cotton Titus D xviii, which LAEME identifies as being penned in a Northwest
Midlands dialect. As indicated in 5 the Southwest Midlands version found in the PPCME2 has
three examples of EPs (one of which is presented in (13)). The Titus text, which is not included in
the PPCME2 but was edited by Mack (1963) has the same syntax in its version of all three
examples.

What Did(n’t) Happen to English?

attention. There is no single version of the Celtic Hypothesis, but whatever version is
adopted, the loss of the DEP is one of the grammatical changes that adherents of this
hypothesis point to as at least possibly caused by Celtic influence.

One apparent problem for the Celtic Hypothesis generally is that most of the syntactic
changes that are attributed to Celtic contact only show up in Middle English, much later
than we might expect. But a possible counter to this objection has been around for a long
time. Tolkien (1963, 184) observed that writing was mainly learned or aristocratic and it
is not until the Old English period is over that we get “a glimpse of what is going on
beneath the cultivated surface”. More recently, contemporary adherents of the Celtic
Hypothesis generally adopt the idea that Celticisms were suppressed in Old English
writing because this was controlled by an Anglo-Saxon elite that suppressed a “Brittonic”
English that arose through language shift and was spoken by the majority of the
population. When the Norman Invasion smashed the West Saxon Standard, Early
Middle English scribes were free to write more as they spoke. I will refer to this idea as
the “suppression hypothesis”. The contact scenario that fits best with this hypothesis is
rapid language shift by Celtic speakers who had poor access to English as they were
acquiring their new language.

The suppression hypothesis warrants serious consideration, not least because it poses
a challenge to the study of syntactic change in Early Middle English that the Norse
hypothesis does not. We can argue about whether a particular syntactic change was
caused by contact with Norse, but no one imagines that such influence seriously distorts
what is found in the Old English texts. With the Brittonic English suppression
hypothesis, however, we can no longer assume that syntactic changes found in Early
Middle English reflect changes made to their grammars by language learners, as is
typically done in generative grammar, because the assumption is that the Early Middle
English texts do not continue the history of the language spoken in the Old English
period, but represent an entirely new language.

The suppression hypothesis as applied to possessive constructions goes like this: The
general populace used only IPs, and this had become the Old English of all but a small
number of Germanic elites. But these elites controlled the production of texts, resulting
in the use of DEPs in writing. With the Norman Invasion, the DEP disappeared with the
rest of the West Saxon Standard.

The most detailed presentation of this idea is by Vennemann (2002), who comments
that his impression is that the “genitival’ construction” (that is, the IP), is “relatively rare”
in “earlier Old English”. He assumes that the DEP was the usual Germanic construction,
and his explanation for why the IP sometimes showed up in Old English writing at all is
that it occasionally slipped through because of social mobility in the church (2002, 220).
In other words, Vennemann is assuming that the Celts imposed their grammar on the
English they learned, since he is assuming, incorrectly as we have seen, that the more
purely Germanic DEP was the dominant construction with affected possessors.
Vennemann also has to assume that the IP would be a stigmatised Celticism. However,
there is no reason why this should be. As has been discussed, the variation between DEPs
and IPs in the Indo-European languages, including the early Germanic languages
generally, is ancient. This means that the IP expressing affected possessors was in a very
different category from some other constructions that have been proposed as due to
Celtic influence, for example periphrastic do. Supposing that Celts had imposed
something like periphrastic do on their Old English, it would have stuck out as a
Celticism, and might well have been suppressed by a Germanic-controlled scribal

CYNTHIA ALLEN

tradition. However, if a scribe who spoke a Brittonic English used only IPs, that would
hardly be likely to be noticed and “corrected”.

To reiterate, Tables 3 and 4 prove that the IP was clearly not suppressed in Old
English, even in poetry, the most Germanic of genres, or in the works of Ælfric, who
made corrections on the manuscripts from his scriptorium to improve the consistency of
case marking (Kitson 1993, 5).

It is also clear that the DEP did not simply disappear from the scene once the West
Saxon Standard was no longer enforced, since it was apparently reasonably productive
with object possessa in Dialect AB. The continued presence of the DEP in Dialect AB
much later than expected by the Suppression Hypothesis is particularly problematic for
the Celtic Hypothesis, given the evidence that is available for the contact situation in the
West Midlands. Gelling (1992, 70), on the basis of the scarcity of Celtic place names in
this area generally, concludes that Welsh speech must have disappeared in most of this
area by the end of the ninth century. If that is correct, we are looking at a fairly rapid
language shift in the West Midlands generally, but Gelling points to Herefordshire, the
area long associated with Dialect AB, as being exceptional and argues for the
continuation of Welsh in this area all through the Anglo-Saxon period. This suggests a
more gradual shift to English in this dialect area, but whether the shift was gradual or
sudden, there is no reason to suppose that the AB scribes, long freed from any shackles
of the West Saxon Standard, did not represent the language of the general populace,
insofar as written language ever represents spoken language, in the use of DEPs.

So it seems that another of the things that did not happen to English is that Celtic
learners of Old English simply failed to learn the DEP and imposed their Celtic possessive
construction on spoken Old English. However, there is more than one way that Celtic
speakers might have led to the ultimate demise of the DEP. Markku Filppula includes
the loss of the DEP in his 2008 discussion of grammatical effects of contact with Celtic
on English. Like Vennemann, Filppula incorrectly states that the DEP was the prevailing
construction in earlier English. Unlike Vennemann, however, he concludes that there is
no timing problem for the Celtic Hypothesis to overcome in this instance. He does not
exclude the possibility of “mutually reinforcing adstratal influences” in the centuries
following initial contact, leading to the unusualness of both languages compared with
other European languages (2008, 39).

It is not implausible that Celtic speakers shifting to Old English did not simply fail to
learn the construction, but affected Old English by reducing its range. How might this
happen? These people may well have picked up on the fact that the DEP was a marked,
special use construction, with the result that in their English they reserved it for the
situations that were at the core of its use, namely with affected possessors. Since the
effect on a possessor that was most likely to call for empathy with the possessor was a
negative one, they would have restricted the range of DEPs to negatively affected
possessors. If the English of these speakers affected the wider community, and made its
way into writing, it could account for the absence of DEPs representing beneficially
affected possessors in most of the texts. This reduced range would be the beginning of
the end for DEPs. Some sort of influence of Celtic on possessor constructions like this
seems plausible, although not provable.

What Did(n’t) Happen to English?

8. Final observations and conclusions

In the twenty years elapsing between the publication of both McWhorter’s and
Vennemann’s papers, there has been an explosion in the tools linguists have available for
studying the history of English and related languages. This article has attempted to
illustrate by way of a case study that a picture painted with broad brush strokes about
what happened in English may look appealing when viewed from a distance, but serious
flaws with the picture may become apparent when we take the closer look that is made
possible with the use of digital corpora. Specifically, it has focused on how two strikingly
different claims about what brought about the disappearance of one construction in
English fare in the face of more detailed empirical data. The evidence leads to the
conclusion that we can discard the idea that it was contact with Norse that led to the loss
of this construction. The idea that contact with Celtic played some sort of role comes out
looking more plausible, but it is clear that this role was not simply the imposition of Celtic
grammar on Old English in the course of language shift. This case study says nothing
about the other losses of Germanic grammar that involve simplification, other than to
say that they all need closer scrutiny.

To understand what happened to English, we also need further research on what
happened in other Germanic languages, especially non-standard dialects. Note that
McWhorter’s list of features is only for codified varieties of the languages in his table. We
need to look at non-literary dialects of these languages to see whether they have
undergone some changes similar to those McWhorter listed as making English unique in
the family. A case in point is the fact that McWhorter’s classification of Dutch as a
language that has retained EPs is open to challenge. Different scholars have given
different answers about the grammaticality of EPs in Dutch; for example, Vandeweghe
(1987) found that the construction was not productive with subject and object possessa.
One problem is how productive a construction has to be before we classify it as one that
is generally grammatical rather than one that is limited to some fixed expressions. More
recent and extensive research into Dutch dialects by Scholten (2018) shows that while
the DEP can be considered productive in some dialects, it is so limited in southern and
western dialects that it can be said to have been lost to at least as great an extent as it
apparently was in Early Middle English. Interestingly, Scholten found that the dialects
where the DEP has the greatest productivity are those in contact with German. In the
case of Dutch, contact with a language that makes substantial use of DEPs has led to a
retention of a construction inherited from Germanic in some dialects in contrast to a loss
of it in dialects with less contact of this sort. What are we to make of the Dutch dialects
where the construction has diminished? I am not aware of any suggested contact
explanation for the disinclination to use the DEP in Flemish,12 and neither Norse nor
Celtic explanations seem plausible. But what is clear enough is that no variety of any
Germanic language can be assumed to reflect exactly the use of EPs in an earlier Common
Germanic stage. This is presumably also true of the other constructions on McWhorter’s
list. In short, to understand what happened to English, we need to know more about what
happened to related languages.

12 According to Dankert and Haegeman (2013), an EP construction does exist in colloquial West
Flemish; however, it does not use a determiner, but requires a possessive pronoun, similar to the
“blended” construction in OE. I view West Flemish therefore as having lost the DEP.

CYNTHIA ALLEN

References

Ahlgren, Arthur. 1946. On the Use of the Definite Article with ‘Nouns of Possession’ in

English. Uppsala: Appelbergs Boktryckeriaktiebolag.
Allen, Cynthia L. 1995. Case-Marking and Reanalysis: Grammatical Relations from

Old to Early Modern English. Oxford: Clarendon.
Allen, Cynthia L. 2019. Dative External Possessors in Early English. Oxford: Oxford

University Press.
Bosworth, Joseph, T. Northcote Toller, Sean Christ, and Ondřej Tichy. 2014. An Anglo-

Saxon Dictionary Online. Prague: Faculty of Arts, Charles University.
Coetsem, Frans van. 1995. “Outlining a Model of the Transmission Phenomenon in

Language Contact.” Leuvense Bijdragen 84: 63–85.
Coetsem, Frans van. 2000. A General and Unified Theory of the Transmission Process

in Language Contact. Heidelberg: Winter.
Cole, Marcelle. 2018. “A Native Origin for Present-Day English They, Their, Them.”

Diachronica 35: 165–209.
Dance, Richard. 2003. Words Derived from Old Norse in Early Middle English: Studies

in the Vocabulary of the South-West Midland Texts. Tempe, Arizona: Arizona Center
for Medieval and Renaissance Studies.

Dankert, Lieven, and Liliane Haegeman. 2013. “Multiple Subjects in Flemish: The
External Possessor.” In Irish Network in Formal Linguistics, Proceedings, edited by
Alison Henry, Catrin S. Rhys, and Iosad Pavel, 2–21. Newcastle upon Tyne:
Cambridge Scholars Publishing.

Donaldson, Bruce. 1994. “Afrikaans.” In The Germanic Languages, edited by Ekkehard
König, and Johan van der Auwera, 478–504. London and New York: Routledge.

Filppula, Markku. 2008. “The Linguistic Outcomes of the Early Contact.” In Markku
Filppula et al., 24–132. New York: Routledge.

Filppula, Markku, Juhani Klemola, and Heli Paulasto. 2008. English and Celtic in
Contact. New York: Routledge.

Filppula, Markku, Juhani Klemola, and Heli Pitkänen, eds. 2002. The Celtic Roots of
English. Joensuu, Finland: University of Joensuu, Faculty of Humanities.

Gelling, Margaret. 1992. The West Midlands in the Early Middle Ages. Leicester:
Leicester University Press.

Gordon, E.V., and A.R. Taylor. 1957. An Introduction to Old Norse, 2nd edn. Oxford:
Clarendon Press.

Haspelmath, Martin. 1999. “External Possession in a European Areal Perspective.” In
External Possession, edited by Doris L. Payne, and Immanuel Barshi, 109–35.
Amsterdam and Philadelphia: Benjamins.

Havers, Wilhelm. 1911. Untersuchungen zur Kasussyntax der indogermanischen
Sprachen. Strassburg: Karl J. Trübner.

Kershaw, Jane, and Ellen C. Røyrvik. 2016. “The ‘People of the British Isles’ Project and
Viking Settlement in England.” Antiquity 90: 1670–680.

Kitson, Peter. 1993. “Geographical Variation in Old English Prepositions and the
Location of Ælfric's and Other Literary Dialects.” English Studies 74: 1–50.

Kroch, Anthony, and Ann Taylor. 2000. The Penn-Helsinki Parsed Corpus of Middle
English 2. York: Department of Language and Linguistic Science, University of York.

Kurath, Hans, and Sherman Kuhn, eds. 1956. Middle English Dictionary. Ann Arbor:
University of Michigan Press.

What Did(n’t) Happen to English?

Kytö, Merja. 1996. Manual to the Diachronic Part of the Helsinki Corpus of English
Texts, 3 edn. Helsinki: Department of English, University of Helsinki.

Laing, Margaret. 2013–. “A Linguistic Atlas of Early Middle English, 1150–1325, Version
3.2.” Accessed July 20, 2023. http://www.lel.ed.ac.uk/ihd/laeme2/laeme2.html.

Lee-Schoenfeld, Vera. 2006. “German Possessor Datives: Raised and Affected.” Journal
of Comparative Germanic Linguistics 9: 101–42.

Lødrup, Helge. 2019. “The Accusative External Possessor with Norwegian Unergatives.”
Argumentum 15: 561–74.

Lutz, Angelika. 2009. “Celtic Influence on Old English and West Germanic.” English
Language and Linguistics 13: 227–49.

Lutz, Angelika. 2011. “Why is West-Saxon English Different from Old Saxon?” In Anglo-
Saxon England and the Continent, edited by Hans Sauer, and Johanna Story, 113–
38. Tempe: Arizona Center for Medieval and Renaissance Studies.

Mack, Frances M., ed. 1963. The English Text of the Ancrene Riwle Edited from Cotton
MS Titus D. xviii (Early English Text Society 252). London: Oxford University Press.

McWhorter, John. 2002. “What Happened to English?” Diachronica 19: 217–72.
McWhorter, John. 2009. “What Else Happened to English? A Brief for the Celtic

Hypothesis.” English Language and Linguistics 13 (2): 163–91.
Payne, Doris L., and Immanuel Barshi. 1999. “External Possession: What, Where, How,

and Why.” In External Possession, edited by Doris L. Payne, and Immanuel Barshi,
3–29. Amsterdam and Philadelphia: Benjamins.

Pintzuk, Susan, and Leendert Plug. 2001. “The York-Helsinki Parsed Corpus of Old
English Poetry.” Accessed July 20, 2023. https://www-
users.york.ac.uk/~lang18/pcorpus.html.

Pokorny, Julius. 1927. “Das nicht-Indogermanische Substrat im Irischen.” Zeitschrift für
Celtische Philologie 16: 95–144, 231–66, 363–94.

Pope, John Collins, ed. 1967. Homilies of Ælfric: A Supplementary Collection (Early
English Text Society 259). London: Oxford University Press.

Roberts, Jane, Christian Kay, and Lynne Grundy. 2017. “A Thesaurus of Old English.”
Accessed July 20, 2023. http://oldenglishthesaurus.arts.gla.ac.uk/.

Scholten, Jolien. 2018. The Ins and Outs of External Possession: A Micro-Comparative
Perspective. Utrecht: LOT.

Taylor, Ann, Anthony Warner, Susan Pintzuk, and Frank Beths. 2003. “The York-
Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE).” York: Department of
Linguistics, University of York. Accessed July 20, 2023. http://www-
users.york.ac.uk/~lang22/YcoeHome1.htm.

Tolkien, J. R. R. 1963. “English and Welsh.” In Angles and Britons. The O’Donnell
Lectures, edited by Henry Lewis, 1–41. Cardiff: University of Wales Press.

Tristram, Hildegard. 2004. “Diglossia in Anglo-Saxon England, or What Was Spoken
Old English Like?” Studia Anglica Posnaniensia 40: 87–110.

Trudgill, Peter. 2011. Sociolinguistic Typology: Social Determinants of Linguistic
Complexity. Oxford and New York: Oxford University Press.

Vandeweghe, Willy. 1987. “The Possessive Dative in Dutch: Syntactic Reanalysis and
Predicate Formation.” In Ins and Outs of Predication, edited by Johan van der
Auwera, and Louis Goossens, 137–51. Dordrecht: Foris.

Vennemann, Theo. 2002. “On the Rise of ‘Celtic’ Syntax in Middle English.” In Middle
English from Tongue to Text: Selected Papers from the Third International

CYNTHIA ALLEN

Conference on Middle English: Language and Text, Held at Dublin, Ireland, 1-4 July,
edited by Peter J. Lucas, and Angela M. Lucas, 205–34. Bern: Peter Lang.

Appendix. Note on texts used.
Information on the presumed dates and dialects of all texts included in the YCOE can be
found at https://www-users.york.ac.uk/~lang22/YCOE/info/YcoeTextInfo.htm. The
YCOE’s information is based on that given in the manual to the Helsinki Corpus compiled by
Kytö (1996). This manual is available at
http://korpus.uib.no/icame/manuals/HC/INDEX.HTM.

The excluded West Saxon texts mentioned in Section 6 are coboetho2.psd,
colawine.ox2.psd, coprefsolilo.psd, and cosolil.psd. The YCOE texts excluded as composed
later than my cut-off date of the third quarter of the eleventh century or from editions
completely or substantially based on manuscripts dating from the twelfth century are
cochronE.o34, coleofri.o4, comargaC.o34, coneot, conicodA, and covinsal. Similar-looking
examples occurring in my data when the YCOE texts present different versions of the same
basic text have only been counted once.

Information on the texts in the York Poetry Corpus it to be found at https://www-
users.york.ac.uk/~lang18/ptext-list.html.

Information about the PPCME texts can be found at https://www.ling.upenn.edu/hist-
corpora/PPCME2-RELEASE-4/index.html.

For further philological discussion of the texts, see Appendix A of Allen (2019).