The cambridge history of the english language volume 2 part 4

James Milroy
spoken dialects and of medieval written dialects. If, at the present day,
a person has lived for long periods in different parts of the country,
his/her speech is normally affected by this experience: hence, we may be
able to detect (for example) Scottish, northern English and southern
English features in his/her speech. The traditional regional-dialect
researcher, who is normally interested in examples of 'pure' regional
speech, will not want such a person as an informant and will reject
him/her in favour of true natives of the areas concerned. In traditional
Middle English studies, the focus on precisely locatable written texts is
exactly parallel to this. Texts that appear to show mixed dialects, or
which seem to have been copied by different scribes, and which are
difficult to localise for these and other reasons, have traditionally been
undervalued and neglected as materials for dialect description. However, the Middle English scholar is in a very unhappy position here,
because the vast majority of surviving documents are of this problematic
kind. We cannot afford to be purists, and so we must devise methods for
exploring the materials that have survived and account systematically
for all the data. This, essentially, is what the LALME programme has
tried to do, and it is easiest to clarify this by referring to some
publications by the researchers themselves.
Mclntosh has consistently emphasised the. importance of this work
for linguistic theory, and has pointed out the failures of the past in the
respects I have mentioned. There is such a wealth of surviving material
that 'linguists fall regrettably far short of exploiting anything like all
there is or even of making optimum selective use' of what is available
(Mclntosh (1975) 1989: 32). The researchers have set out to correct this,
and the task has involved not only an enormous commitment of time
and energy, but also a wide range of sophisticated analytic skills. It is
important to remember that many of the relevant documents have never
been published: thus, an essential task has been to track down and
analyse unpublished manuscripts in addition to those that have been

published, and to prepare (amongst other things) what the researchers
call scribal profiles for the documents (Mclntosh 1975).
Most of the surviving literary texts are copies, and these are often at
more than one remove from the original. Some copies are in the hand
of a single scribe, but others are by two or more scribes. Mclntosh and
Wakelin (1982) discuss the case of Mirk's Festial, which is in the hands
of five different scribes - but one of these scribes seems to have copied
material in no less than fourteen different dialects. Although this may be
an extreme case, the example does make it clear how important it is to

188


Middle English dialectology
explore the scribal and linguistic make-up of the texts, and Mclntosh
et al. have therefore suggested a classification of text types in terms of
the history of copying and of the different patterns of textual mixing
that may arise.
In some cases, a single scribe seems to have translated from an
original into his own dialect, but he may have done this inconsistently
to a greater or lesser extent. Sometimes, for example, the translation is
' progressive': the scribe starts by copying more or less faithfully from
his exemplar, but as he begins to work more quickly, he resorts more
and more to the forms of his own 'dialect' (or scribal practice). Other
texts, however, are composite: two or more different copyists have been
at work on the text that has come down to us, or a single scribe has
faithfully copied an exemplar which is itself the work of two scribes.
The Cotton MS of The Owl and the Nightingale is a well-known example
of the latter, and in this case the place where one scribe finished and the
other began can be accurately determined: the scribe of the final version

seems to have made few changes. In extreme cases of mixed origin, we
encounter Mischsprachen, in which (according to Mclntosh et al. 1986)
the variation encountered is random and unpredictable. However, it is
a measure of the great progress that these scholars have made that Laing
(1988) has been able to illuminate the textual histories of two
manuscripts of Richard Rolle's English Psalter, which are effectively
Mischsprachen. She demonstrates, using quantitative methods (amongst
others), that even in these extreme cases, the layers of copying may be
separable.
As many Middle English literary texts survive in only one copy,
methods of 'internal reconstruction', in addition to comparative
methods, are essential. What we can know about the original depends
on interpretations of internal variation in the text, which lead to
hypotheses about the provenance of the original, and to some extent
these interpretations have traditionally depended on rather purist
notions about relatively uniform dialects. Mclntosh et al. (1986) have
given attention to the possibility of personal and social variation
affecting the language of the texts. They note the possibility that a writer
of mixed upbringing may betray in his usage the influence of two or
more different dialects, and that a text may be affected by mixing of what
they call a sociolinguistic kind, especially through influence from the
spread of standard English, which becomes noticeable in later Middle
English. Mclntosh et al. also note that it is possible to find, especially in
the fifteenth century, an extremely wide range of spelling variation in

189


James Milroy
the work of a single writer. Thus, we should be able to acknowledge

that when a text contains variation, this may not all be due to the
activities of copyists from different regions: the copyist of the extant
version, or his speech community, may have tolerated a good deal of
variability in usage. Indeed, one thing is obvious: the copyists evidently
thought that the mixed and variable usage of their copies was acceptable
in some way,'and Mclntosh, Samuels and their colleagues have given
much more weight to this aspect of variation than have previous
scholars.
3.2.9

Some applications

I pointed out in section 3.1 that the goal of a dialect survey is primarily
linguistic and is specifically to describe and account for variation in
language. But it is clear that such an extensive exploration of manuscript
sources can have applications to other kinds of research. The most
immediately obvious applications are literary, editorial and textual. The
case of Havelok the Dane (MS Bodley Laud Misc. 108 (A) plus some other
fragments) is important linguistically, but it is also of interest for literary
history. To demonstrate the LALME method, I shall now briefly
review some of Mclntosh's arguments about the localisation of Laud
Misc. 108.
Mclntosh (1976) argues that this text may be from Norfolk, southwest of King's Lynn, a long distance south of the town that is preeminently associated with it - Grimsby, in north Lincolnshire. The -es
verbal ending that we have noted above (p. 169) makes it quite possible
that the original was composed in north Lincolnshire, but we do not
know how many copies intervene between the putative original and the
manuscript we have to hand. It has long been clear that Havelok has
much mixing of forms that are not characteristically northeast midland
(e.g. a fairly high incidence of < o > for OE /a:/, on which see map
3.1), and, despite the attribution to north Lincolnshire by Dickins &

Wilson (1956: 34) and others, the surviving manuscript has never
seemed to be from as far north as Grimsby. Mclntosh uses the 'fit'
technique to suggest a more precise location for the text than has been
suggested before - an area in west Norfolk south-west of King's Lynn.
Various comparisons are used here, but an important one is the
comparison with the work of scribe D of BM Cotton Cleopatra C vi.
Mclntosh shows that this is probably from west Norfolk rather than
Lincolnshire, and it displays many similarities to Havelok. The majority

190


Middle English dialectology
of the key forms in Have/ok that Mclntosh uses to refine the ' fit' turn out
to have a distribution to the south of west Norfolk (e.g. in material
located in Ely, near Cambridge) more often than to the north of it.
Indeed, when some variants (e.g. togidere 'together') do have a
distribution north of Norfolk, they are usually also found to the south.
The incidence of these more southerly variants is greater in Havelok than
in Cleopatra C vi; therefore, Havelok may be from somewhere to the
south of Cleopatra C vi.
Another theme running through this research programme is the
application of its findings to questions of importance in the history of
English. In a very influential paper, Samuels (1963) has considered
changes in the London dialect of the fifteenth century and the varieties
that may be said to have been competing at that time for pre-eminence
as the basis of modern literary standard English. These varieties are
classified into four types, of which the Chancery standard is the ancestor
of the modern literary standard. Samuels also argues tha. the main
regional influence on London English and the early standard language

is not the whole east midland area, or areas to the east of it, but the
central midlands. Again, the relative precision here is made possible by
work on the atlas project.
I have noted above that the Ls4LME researchers have taken more
account than previous scholars of sociolinguistic factors. Unlike
traditional scholars, they have pointed out that some of the variation
encountered may be inherent in the written language of one particular
scribe, and they have mentioned the acceptability of variant forms to the
copyists. It is appropriate, therefore, to go on to consider in section 3.3
another perspective on variation in Middle English documents, which
is not primarily about geographical provenance and not primarily
devoted to reconstructing textual histories, but which may be seen as
complementary to these. This perspective is informed mainly by the
results of variation studies on present-day dialects: it depends on the
perception that variation may itself be structured and is not necessarily
the result of errors or carelessness. Therefore, it can be studied in itself
as a matter of linguistic interest and as a contribution to historical
linguistic theory.

191


James Milroy

3.3
3.3.1

Variation theory and Middle English dialectology
Introduction

Present-day dialectology has developed an additional dimension of
interest which focuses on variation in the speech of individual speakers
and speech communities rather than on broader geographical patterns,
and it is the purpose of this section to consider how far this perspective
can contribute to the analysis and interpretation of variation in Middle
English. The primary interest of social dialectology is in tracing the origins
and diffusion of linguistic changes, and these patterns are typically
discovered in language variation within communities in the different
speech styles of individuals and of social groups. From an analysis of
these patterns, changes in progress can be located, and their path
through the community can be described. The most important principle
is that languages (or dialects) are never 'pure' or uniform states of
language, and further that variation in speech is itself structured and
functional; e.g. it may be shown to serve social purposes. As Weinreich,
Labov & Herzog (1968) have pointed out, structuredness should not
be equated with uniformity; for a language state to be structured it does
not have to be uniform. As Middle English language states are very far
from being uniform, they should in principle be suited to this kind of
analysis.
The claim that variation is structured in communities has been tested
by numerous studies (Labov 1966; Trudgill 1974a; Milroy & Milroy
1978-to name a few), which have demonstrated regular patterns of
variation according to speech style, social context and social group, and
the basic perception has been formalised in the idea of variable rules
(Labov 1972a; Sankoff 1978, etc.). Rules of this kind specify the
constraints on the variation that has been discovered by empirical
observation. In a present-day community, these appear as constraints on
variation in speech: in Middle English we must locate these constraints
initially through the writing system.
The consequence of this is that, in general, variable texts can become

more valuable for our researches than relatively uniform ones. Let me
clarify this by comparing a modern case with a medieval case. Suppose
we show that in a present-day vernacular, there is structured variation
in verb forms of the type he doesI he do (see Cheshire 1982 for a relevant
study), with one form perhaps being preferred in formal styles and the
other in casual styles; we may also - by comparing the speech of
different social groups and age-groups - additionally show that one

192


Middle English dialectology
form is progressing at the expense of the other. As it happens, Middle
English texts also frequently exhibit variation in verbal inflections.
Suppose, for example, a text (such as the Bestiary, discussed in section
3.2.2) exhibits third-person singular verb-form variation in -es, -eth and
syncopated forms (e.g. stant for standes, standeth): it may be possible in
many cases to show that the text is composite and expose the ' layers' of
copying (see Laing 1988). On the other hand, it may also be the case that
all three forms (or perhaps two of them) were current in the underlying
dialect of the scribe (or of the author), or - more properly - of the
speech community to which he belonged. Indeed, as the writing system
was not standardised, it is likely that structured variation of this kind
would enter more readily into the texts than it would today.
Clearly, in this approach, the exact geographical provenance of texts
is no longer the primary, or exclusive, interest (important as it is to
establish this as far as possible). The method can be seen as
complementary to geographical dialectology: the goal is to contribute
to theories of change, and within this to our understanding of the
history of English, which is of course a multidimensional history

focusing on variation of all kinds. One possible result may be to show
that variation attested in later periods of English can be traced back to
these early sources.
3.3.2

The neglect of structured variation in Middle English studies

Variability in Middle English has sometimes been perceived as an
obstacle rather than a resource, partly because of the broadly literary
emphasis on which we have commented above. In editorial and
descriptive commentary, it is very easy to find comments about chaotic
or 'lawless' spelling (e.g. Sisam 1915: xxxvii) and even editorial
judgements to the effect that a given scribe could not have been a native
English speaker — so variable is his orthography. This last judgement
(although it is commonly made) is speculative, of course, as the scribe
is normally anonymous. However, judgements of this kind can
effectively block further investigation of variable constraints in the texts
in question: they can be dismissed as ' corrupt' or' unreliable' specimens
of language. One way in which variation of this kind is discounted is to
claim that the scribe was Anglo-Norman, or that the spellings are
Anglo-Norman and therefore not valid evidence for the history of
English.
The Anglo-Norman argument goes back to Skeat (1897), who

'93


James Milroy
specified particular features of spelling as Anglo-Norman. These are
discussed by Milroy (1983), and it is noticeable that many of these

features, such as < w > for wh, have reflexes in later English. As a result
of Skeat's claims, the very fact of variable spelling in an Early Middle
English document became in itself a reason for concluding that the
scribe was Anglo-Norman and that his spelling could be corrected by
editors and ignored by historical commentators. The work of scribes
writing centuries after the Conquest has even been dismissed in this
way, seemingly mainly because it is variable, and not because we can
(usually) know whether the scribe was a first-language speaker of
Anglo-Norman, or whether it would have been relevant if he had been.
Leaving aside this argument, we must also recognise that scholars
have sometimes been more generally influenced by the notion that
written language should be uniform, even in a period in which it plainly
was not uniform, and they sometimes appear to chide the scribes for
spelling variably. Scragg (1974: 26), for example, comments that 'The
existence of regional orthographies, and their confusion in the copying
of texts resulted in a very lax attitude to spelling in most scribes.' In
the context, this 'very lax attitude' seems to be measured against
circumstances (such as Late Old English or the present day) in which
there is a uniform standard of spelling: thus, all this really means is that
in Early Middle English there was no uniform standard. Scragg adds
that these scribes had ' no conception of a spelling standard' and then
-much more dubiously - that they used 'variant forms at will'.
However, if the scribes really had used variants 'at will', we would
actually be unable to read the texts, as there would be no system in the
spelling; but there must always be some order in any spelling system that
we can read, even if it is a variable system. Therefore, the scribes did
not spell 'at will', but according to variable (and historically mixed)
conventions. It is our task to attempt to specify the constraints on
spelling under which they were working, always admitting that even
after we have done this, there may well be residues of apparent

randomness that we cannot explain.
3.3.3

Orderly variation in spelling

The existence of variable orthographies is an advantage to the Middle
English dialectologist in exactly the same way that the existence of
spoken variation is an advantage in present-day research. Although the
scribes no doubt made 'errors', it should be possible to investigate

194


Middle English dialectology
variable texts in extenso to determine the extent to which the variation in
spelling (or indeed in other linguistic dimensions) is in fact orderly, and
whether this variation can help us to work out what might have been
happening in spoken English at the time. As an example, let us briefly
consider some aspects of spelling in Have/ok the Dane.
The Have/ok text is one of those sources that has been traditionally
thought to be the work of an Anglo-Norman scribe (Sisam 1915) on the
grounds that the spelling is highly variable in the respects specified by
Skeat. However, although it doubtless contains some forms that are
simply 'errors', 8 it also exhibits the kind of orderly variation that could
be captured within a variable-rule framework, but in spelling variation
rather than phonology. The scribe does not have a free hand with spelling
variation: there are constraints on the variants he uses. OE postvocalic
/ht xt/, for example, can be represented in the spelling of Have/ok by
< s t > , < h t > , < t h > , < c h t > , < c t h > (in words of the type riht,
niht), but not by, e.g. < gt > , < ght > or by random and unpredictable

forms such as tc or m. The variation is constrained in much the same way
as present-day phonological variation in speech communities is observed
to be constrained. Therefore, just as present-day phonological variation
can be used as a clue to change in progress, so it may be possible here
to use orthographic variation in the same way.
The spelling variants for OE (ht)9 overlap with spelling variants for
other forms (from different sources in Old English), just as phonological
variants in present-day studies are found to overlap (see Milroy &
Harris 1980; Milroy 1981). Thus, if we take the realisation th, we find
that this can be used word-finally, not only for (ht), but also for (t) and
(th). The result of this is that a spelling like with can realise three separate
classes: OE iviht (' wight, person'), OE wip (' with') and OE hwit
('white'), and this of course applies to other items of these types. To
formalise this — the following (Old English) classes can appear with
final th:
1
2
3

Final (postvocalic) dental fricatives: /)? 6/ e.g. with (OE wip,
PDE with).
Final (postvocalic) dental stop: / t / e.g. with (OE whit, PDE
white).
Final /ht xt/: e.g. with (OE wiht, PDE wight).

The potential realisations of these three classes are, however, different:
(ht) items can also appear with < st > , < cht > , etc. (e.g. wicht): the
other two classes cannot; (th) items can also appear with final <]?>,

James Milroy
< 6 > (e.g. wip): the other two classes cannot; (t) items can appear with
final single / (e.g. wit, whit): the other two classes cannot. Thus, ' with,
wight' cannot appear as wit, whereas 'white' can. To this extent,
therefore, the variation is constrained, and not random. Applying the
principle that change in progress is manifested in variation, let us
consider its possible implications for spoken variation in Middle
English.
The study of (ht) in Have/ok is of course relevant to the date at which
the velar fricative [x] before [t] (in right, might, etc.) was lost in English.
The prima facie conclusion to be drawn is that in the variable phonology
of the 'underlying' (east midland/East Anglian) speech community,
loss of the fricative and merger of wight, white or close approximation
and overlap, had already taken place. It is also possible that in this
variable phonology there was some tendency to merge final /]>/ with
/ t / . If developments of this kind were not in some sense in progress,
then there would have been less likelihood of the scribe observing
precisely this pattern of orderly spelling variation, because, given the
variable state of the orthographic conventions known to him, he could
have chosen to vary in other ways. Of course, it is quite another matter
to go on to argue from this very limited piece of evidence that loss of the
fricative in /xt/ was generally accepted as a completed sound change in
the English language as a whole at this early date. Yet, if we take this
together with the fact that many other forms characteristic of Modern
English spread in these centuries from the east midlands and the north
(see the discussion of morphological dialect indicators in section 3.2.5),
we can advance the hypothesis that this change was in progress in the
east midlands around 1300 and look for further evidence to support or
refute this. If, however, we insist that many Middle English scribes were
simply careless or poorly acquainted with English, we shall be inclined

to reject the evidence and date this sound change much later — at a time
when it was actually completed in ' standard' English. This, of course,
will not bring us anywhere near the origin of the change.
Loss of the velar fricative is a change that was finally adopted in
standard English and formal styles. Middle English sources, however,
also contain variation that may be relevant to non-standard varieties and
casual styles of speech; hence, there may be considerable time-depth to
these variables also. In section 3.3.4, therefore we consider how far
studies of variable spelling in Middle English are capable of throwing
light on this.

196


Middle English dialectology

3.3.4

The time-depth of non-standard variants

A number of present-day non-standard and casual speech forms appear
to be indicated by some features of variable Middle English spelling.
Some of these are recognised as regional and have been studied as such
(e.g. in Wakelin & Barry 1968 the study of the voicing of initial
fricatives in southwest England); others are more widespread in
English. One of these is 'final-stop deletion' (loss of /t, d/, and
sometimes other stops, infinalclusters in words such as mist, mend). This
is today very common in many varieties of English (Guy 1980; Romaine
1984), but not common in careful styles of Received Pronunciation
(hence its exclusion from many accounts which claim to be accounts of

'English'). The LALME maps show a distribution of final-consonant
loss also in medieval written English, and I have noted a number of
examples in Have/ok and other texts. Thus, the phenomenon may have
been part of variability in English for many centuries - more common,
perhaps, in some dialects than in others, receding at some periods and
progressing at others. Yet it plays little part in standard accounts of
the history of English before about 1600, and Middle English stopdeleted forms (such as bes, Ian 'best', 'land') are amongst the forms that
are typically corrected by editors as errors.
There are other features that may have much earlier origins than is
generally believed. These include: (a) the (casual style) -in' ending on
present participles; (b) certain widespread socially or regionally marked
alternations in Modern English, such as 'stopping' of dental fricatives
in, e.g. thick, that, and [h]-dropping. One of the most important points
arising is that studies of these variables contribute to the history of the
language as a multidimensional phenomenon. They accept as a principle
that, just as English is variable today, so it has constantly incorporated
variation through the centuries. Indeed, as some of this 'stable'
variation may have been very long-lasting, we may have to reconsider
what it means to say that some categorical change was completed at some
specific date in history. Bearing in mind also the points made above on
the structured nature of variation, I now consider as an example the case
of [h]-dropping in English, i.e. variable loss of [h] in stressed syllables
initially before vowels.
Although scholars have noticed instability in initial < h > spellings
in Middle English, the traditional view (e.g. Wyld 1936: 296) is that
there is little reliable evidence for '[h]-dropping* in English much
before the end of the eighteenth century, and earlier instability in

197

James Milroy
spelling is usually dismissed as unreliable in handbook accounts of the
history of English sounds (e.g. Brunner 1963; Ekwall 1975). One
reason given for the alleged lateness of the phenomenon is its apparent
absence from colonial English (Wyld 1927: 220). From a variationist
point of view, the reasoning here is not necessarily acceptable, as
colonial forms of English may have changed; for example, there is
evidence that, although Australian English is [h]-ful now, it used to have
[h]-dropping (Trudgill 1986: 138-9). The evidence of variable spelling
in Middle English seems to point to an early origin, and if the arguments
for this can be sustained, they have a clear relevance to understanding
patterns of variation in Middle English.
In modern times [h]-dropping — like -in' for (ing) — is extremely
widespread and well established: it is not confined to a particular region
(as voicing of initial fricatives is, for example). In fact, most people in
England and Wales drop their [h]s to a greater or lesser extent.
Therefore, if the origin of the phenomenon is as recent as the late
eighteenth century, it is difficult to explain how it could have become so
geographically widespread in so short a time. It was already highly
salient and overtly stigmatised by the latter half of the nineteenth
century (for evidence of this, see Milroy 1983: 40). It is reasonable to
assume that if a linguistic variant is so widespread and strongly
established, it probably has quite a long history in the language. The
late-eighteenth-century evidence adduced by Wyld and others is
therefore likely to indicate the date at which it had become stigmatised
as a ' vulgarism', rather than its date of origin.
The most important reason for questioning the traditional view,
however, is that variation in initial < h > usage is a very common
pattern in Middle English texts. Whereas we have discussed orderly

variation in spelling (above) by looking at distribution within a single
text, the evidence for early [h]-loss depends on spelling variation across
a number of texts. Many Middle English sources exhibit variable use of
the letter < h > in syllable-initial positions (i.e. in words like hate,
hopper). Sometimes it is omitted where it is historically expected to be
present, and sometimes it is added where it is not expected.
This pattern of variation is widespread in Early Middle English, and
the LA.L.ME maps also show a distribution at later periods. It has been
very widely noted by careful editors such as Hall (1920), and (although
the atlas map shows some west midland distribution) it seems in the
early part of the period to be most common in texts originating in the
east midlands, East Anglia and the south. It is quite common in

198


Middle English dialectology

southern texts of ca 1200, such as Poema morale (Lambeth and Trinity
MSS) and The Owl and the Nightingale, in early east midland/East
Anglian texts such as Genesis and Exodus, King Horn, Havelok. It is found
in the Otho MS of La3amon's Brut, but not in the Caligula MS, which
is certainly southwest midland. It is not characteristic of early texts
known to be west midland, such as those of the Katherine group. The
geographical distribution of relevant texts from ca 1190-1320 is from
Lincolnshire or Norfolk (in the north) to the southern counties, but the
instability seems to be greatest in the east midlands. Certain later texts,
mostly of a non-literary kind, display the same phenomenon. It is found
in Kristensson's (1967) northern onomastic sources in the period
1290-1350, and Wyld (1927, 1936) documents a number of later

examples, from sources that include the Norfolk Gilds (late fourteenth
century), The Paston Letters (fifteenth century), and the mid-sixteenthcentury Diary of Henry Machyn (for a fuller discussion, see Milroy 1983:
48-9). In my own investigations of many of these texts, I have noted
additional examples. The following selective lists are from the
thirteenth-century Genesis and Exodus (Morris 1873a), which is believed
to originate in East Anglia. They include examples additional to those
given by Wyld. List 1 documents omission of h, and list 2 addition of
'unhistoricaF h:
1

2

a, adde, adden, as, aue, auede, aued, auen, aue (parts of the verb
'have': lines 239, 240, 1251, 1505, 1760, 2388, 2425, 2720, and
very commonly — considerably more so than forms with h);
algen, aligen ('hallow'): 258, 918; ail ('hail'): 3066, 3183; ate
('hate'): 373, 3638; alt ( < infin 'hold'): 924; atted ('is
called' < OE baton): 813; e (' he, they'): 2341, 2708, 4094; egest
('highest'): 143, 1224; eld ('held'): 2999; elles ('of hell'): 4157;
ere ('of them' < OE heora): 2855, 3773; eden ('hence'): 2188;
eui ('heavy'): 2559; is ('his'): 482, etc; opperes ('hoppers' i.e.
'locusts'): 3096; ostel ('hostel' i.e. 'lodging'): 1056; om
('home'): 2270; oten ('called'): 1131.
hagte (' wealth'): 431; hagt (' grief): 486, 2044, 2082; halle (' all'):
2340; ham ('am'): 926; helde (i.e. elde 'age'): 457, 1527; her
('before'): 801; her/ (i.e. erf 'cattle'): 2991; herde (i.e. erde
'land'): 806; hie ( T ) : 34, 2783; hinke (i.e. inke 'dread'); his
(' is'): 2935; hore (i.e. or ' before'): 958; hunframe (unframe): 554;
hunkinde (unkinde): 534; hunne ('grant'): 2249; hunwreste
('wicked'): 537; hure ('our'): 322, 2206.

199


James Milroy
The most immediate 'explanation' for such substantial instability in
the use of < h > is that syllable-initial [h] was not present, or only
variably present, in the speech of the relevant regions. The letter was,
however, present in the orthographic tradition (regardless of the mixed
origins of the tradition in Old English, Anglo-Norman and Latin):
thus, in the absence of strong orthographic standardisation, the scribes
would omit it on some occasions and insert it 'hypercorrectly' on
others.
As instability of < h > is extremely common, it is remarkable that
careful scholars, such as Wyld, could have been so much aware of this
type of evidence, but could nevertheless have rejected it. I have
suggested above some of the reasons why this should be so and have
elsewhere (Milroy 1983) reviewed some of the arguments that have been
used to reject variable evidence; however, as it happens, instability of
< h > is one of the putative 'Anglo-Norman' features distinguished by
Skeat (1897).
Frequently, this orthographic evidence for variation in Middle
English is rejected on the grounds, not that the scribe was literally an
Anglo-Norman, but that uses such as variable < h > are originally
scribal importations from French or Latin usage. However, the origin
of scribal habits is not in itself valid proof that variable use of the
conventions in written English do not also relate to variable usages in
spoken English. This is because variable scribal usage is likely to be
functional in some way, and the most immediately obvious function of
an alphabetic writing system is to relate writing to speech forms

(however complicated this relationship may be). Thus, especially in a
time of unsettled orthography, it is extremely likely that current sound
changes will be admitted into writing, whatever the historical origins of
the writing conventions may be. Moreover, theprima facieevidence for
[h]-dropping continues well into Early Modern English - long after
there can be any suspicion of direct Anglo-Norman scribal interference.
The evidence from spelling strongly suggests that (h) has been a variable
in English for many centuries: [h]-loss may have gone to completion in
some varieties at particular times and places, but in general speech
communities have used the variation over these centuries as a stylistic
and social marker. In other words, whatever the origin of the
phenomenon may be (in phonotactic constraints, in rapid speech
processes or in language contact, for example), it has probably had a
social and stylistic function in the language for centuries.
Although a sociolinguistic perspective does suggest some possible

200


Middle English dialectology

interpretations of the social and stylistic functions of [h]-dropping
and other kinds of variation at different times and places, we are not
primarily concerned with these here and will refer to them briefly in
section 3.4. The case of [h]-dropping is discussed here as an example of
the possible contribution that a variationist perspective may make to a
multidimensional account of linguistic variation in Middle English, and
through that to a multidimensional history of the structure of the
English language.

3.4

Concluding remarks

In this chapter I have concentrated on methods of ascertaining the
distribution of linguistic variants in Middle English rather than
attempting to give a full account of the details of variation. This latter
task would involve reporting massive variation, which is best appreciated by direct study of the texts themselves and of the surveys and
maps that have been discussed above. I have also been very much
concerned with how the variation discovered is to be interpreted, as the
aims of dialectology are more far-reaching than merely to record the
distribution of variants. Ultimately they are concerned with explaining
linguistic change, the seeds of which are manifested in variation.
Historical dialectology has always been modelled to some extent on
the methods and principles of present-day researches, and investigators,
such as Kristensson (1967), have normally emphasised this dependence.
This chapter has focused on two branches of the subject - regional and
social dialectology - and we have assessed what each of these can
contribute to the study of variation in Middle English. Of these two,
however, it is social dialectology that has been most explicitly concerned
in recent times with the theoretical issue of how linguistic change is to
be explained. In this concluding section, therefore, I should like to take
up two points connected with social dialectology that are relevant to the
exploration of past states. The first concerns the idea of uniformity in
language as it applies to historical description and interpretation. The
second concerns the social nature of language and, within this, how far
the framework of social dialectology can help us to understand the
social motivations of change and variation in the past.
Most branches of linguistic enquiry have been influenced by the
doctrine that only uniform language states can be regular or structured.

Therefore, when variation is encountered, it may well be discounted as
'irregular'. We have noticed above that this doctrine has in the past

201


James Milroy
influenced the analysis of Middle English in the tendency to dismiss
variant forms as errors or to explain them away as 'Anglo-Norman'. I
have suggested in section 3.3 that, although there may often be errors in
the texts, it is appropriate in the first place to determine the extent to
which the variation displayed is actually structured, and I have
attempted to demonstrate this by looking at constraints on variation in
Have/ok. The belief in uniformity has, of course, had a more general and
diffuse effect on the historical description of English, chiefly in the form
of emphasising the history of standard English, at the expense of
'vernaculars'. Although this is less relevant to Middle English than to
later periods, it has resulted in some selectivity in reporting the data in
handbooks of Middle English. Thus, the time-depth of such phenomena
as final-stop deletion and [h]-dropping may well have been underestimated by many. The general effect of this is to understate the
multidimensionality of language and its history.
As for the social motivations of change, it is clear that although
linguistic changes are initiated and diffused by live speakers, they become
apparent in changes in the language system. What we have to explain is
how innovations initiated by speakers find their way into language
systems, at which point, of course, they become linguistic changes. For
this reason it is useful in social dialectology to bear in mind a distinction
between speaker-based and system-based accounts (Milroy & Milroy
1985), and to look at how speakers are motivated to innovate and to
accept innovations by others. Yet, whereas present-day dialectologists

have access to speakers in social contexts and can therefore form
hypotheses of a social kind on the basis of fieldwork and empirical
explorations, Middle English dialectologists must attempt to get access
to speaker motivations by very indirect means. One source is the general
sociopolitical situation as studied by historians; within this branch of
enquiry we also gain insights from, for example, comments by
contemporary observers on the language situation and documentary
evidence of population movements. Thus, from Ekwall's (1956) study,
we can deduce that the change in London dialect from a southern/southeastern to a midland type is related to large-scale immigration into London from the east midlands and north. Another
method we can use is to project the social argumentation of sociolinguistics on to the past.
One aspect of this argumentation concerns evaluation of linguistic
variants by communities. This may be relevant to explaining the change
in the character of London English noticed above, as population

2O2


Middle English dialectology
movement does not in itself explain why the dialect of the in-comers
should prove to be dominant: why was their speech not simply
assimilated into the pre-existing London dialect? It seems that certain
features of the in-coming dialect were evaluated more highly, and
sociolinguists might explain this kind of pattern in terms of prestige or
— preferably — in terms of the changing identity functions of language.
Present-day studies in the rise and development of urban vernaculars
(Milroy 1981, and Harris 1985, on Belfast, for example) help to provide
a framework additional to the findings of historical investigations, in
which these historical phenomena can be further considered.
It has also been clearly established that in the course of time
evaluation of particular variants can change or even be reversed.

Therefore, it is most unlikely that present-day stigmatised forms have
always been stigmatised, and I have suggested elsewhere (Milroy 1983)
that in the Middle Ages [h]-dropping may have been a marker of more
cultured speech. Although in a particular instance like this such an
interpretation may be debatable, the belief that 'vulgarisms' have
always been 'vulgarisms' is much more dangerous. Apart from specific
cases, however, there are broader trends in the history of English for
which sociolinguistics can provide an interpretative framework. One of
these is the trend toward simplification that was mentioned in section
3.2: it seems fairly clear that such a sweeping change is at least to some
extent associated with language contact.
Language-contact studies form an important background to presentday social dialectology in the work of Weinreich (1953), and the topic is
further developed in Trudgill's Dialects in Contact (1986). In a suggestive
study Anderson (1986) has examined simplification patterns in a wide
variety of European dialects and proposed a distinction between open
and closed communities - those that are open to outside influences as
against those that are not. We (Milroy & Milroy 1985) have proposed
that speakers are open to outside influences to the extent that their social
links within close-tie communities are weakened and, further, that
simplification is associated with weakening of links (Milroy 1992).
Studies of this kind seem to be suggestive as projections on to the past:
late medieval London, for example, seems to have been an open
community in this respect, and changes in the London dialect may have
depended on the development of weak personal ties resulting from
population movements.
The most extreme cases of simplification are pidgin/creole languages,
and these have been empirically studied very widely in recent years (e.g.

203

James Milroy
by Miihlhausler 1986). Using pidgin/creole arguments, the most
extreme solution to the Middle English simplification question is that of
Bailey & Maroldt (1977), who argue that Middle English was a
French-based Creole — a view that few have accepted as it stands. Yet, it
seems likely that language-contact phenomena may be implicated in a
more general way: the advanced inflectional loss in twelfth- to
thirteenth-century east midland dialects, for example, may be in some
way associated with heavy Danish settlement in these areas — even if
the language varieties that resulted from this were not Creoles. In
general, these observations on language contact, rooted as they are in
empirical research, provide a well-motivated framework in which
simplification - and variation in the speed of change at different
periods — can be discussed and debated.
There are many other matters of interest to theories of change that I
have not been able to discuss in this chapter. However, I hope that I
have said enough to make the point that continuing study of Middle
English dialects is of crucial importance to writing a realistic multidimensional history of English, and of considerable importance also to
theories of linguistic change in general.
FURTHER READING

Elementary introductions to variation in Middle English are available in
standard histories of the English language. These vary in the amount of
attention given to dialect variation, and some are quite poor in this respect.
Amongst those that give attention to variation, the appropriate chapters of the
following are recommended: Baugh & Cable (1978), A History of the English
Language, 3rd edn; Bourcier (1981), An Introduction to the History of the English

Language (English adaptation by Cecily Clark); Strang (1970), A History of

English. The collections of Middle English texts by Dickins & Wilson (1956)
and by Bennett & Smithers (1966) have useful general introductions to Middle
English dialect variation and useful commentaries on the individual texts.
Martyn Wakelin's English Dialects: an Introduction (1972a) contains a good deal
of historical material.
Amongst more recent writers on regional dialects of Middle English, the
work of Kristensson and Sundby is recommended, together with the classic
work of Ekwall, which underlies their work. However, the most important
contributions to Middle English dialectology are those of Mclntosh, Samuels
and their colleagues. The introduction to the first volume of LALA4E is very
important. The complexity of the work that has gone into Middle English
dialectology can be further investigated in, for example, Mclntosh's study of
the provenance of Havelok the Dane and the Mclntosh & Wakelin study of

204


Middle English dialectology
John Mirk's Festial. The recent collection of essays edited by Mclntosh,
Samuels and Laing is recommended. Apart from the studies mentioned, this
includes an important article by Samuels on the origins of early standard
English and one by Laing on linguistically composite texts. The maps
themselves will be very useful to investigators who have specific aims in mind
(for example, compiling a history of/h/-dropping!).
There has been very little work on sociolinguistic variation in Middle
English. Some of my comments in this chapter are discussed more fully in J.
Milroy (1983) and treated in the context of variation studied in J. Milroy,
Linguistic Variation and Change (1992). The classic essay on backward projection
of variation studies is Labov, On the Use of the Present to Explain the Past, which
is most accessible in the reprint by P. Baldi and R. Werth (eds), Readings in

Historical Phonology (University Park: Pennsylvania State University Press,
1978).

NOTES
1 There is a tremendous bibliography relevant to the study of variation in
Middle English that stretches back for well over a century. I have not
attempted here to review this in detail, as good reviews are available in the
best histories of English, such as Baugh & Cable (1978). The main focus of
this chapter is on methodology and interpretation.
2 Another manuscript from ca 1200 that is very consistent in spelling is the
Ormtilum — from the east midlands. This, however, is experimental — a
conscious attempt to devise a consistent orthographic system that
represents the phonological system (especially vowel length) accurately.
3 Recent developments — in particular the idea oi lexical diffusion (Wang 1969)
have questioned the Neogrammarian axiom, which often lay behind earlier
interpretations of sound change. For an assessment of the controversy, see
Labov 1981 and Kiparsky 1988.
4 As systematic change in language (e.g. in phonology as against lexical
borrowing of learned words) is initiated and diffused by speakers (and not
writers) in casual and informal conversational contexts, the styles and modes
of writing conspire to 'cut off' the origins of linguistic changes. They
represent 'planned' rather than 'unplanned' discourse (Ochs 1983).
5 For example, most dialects of English have merger of words of the type
pair I pear I pare and many other sets of items. However, the writing system,
here as elsewhere, retains older spelling distinctions.
6 There are many studies of particular instances in historical records of the
use of French in the medieval English speech community, and the place and
function of Anglo-Norman has been widely debated. Some commentators
have tended to emphasise its importance and its longevity as a mother
tongue (see, e.g. Legge 1941). A different view is expressed, however, by

205


James Milroy
some others, for example, Rothwell (1966, 1968, 1975). He has shown that
certain thirteenth-century scribes (who would be amongst the most
educated persons of the time) did not have a native-like command of
French, and were given to quite systematic errors in translation. Thus, the
position was fluid and changing, ultimately resulting in the disappearance
of Anglo-Norman from the speech community. It is discussed in many
histories of English, such as Baugh & Cable 1978.
7 One way of achieving ' accountability to the data' is to quantify the relevant
variation (as Labov 1966 does). There is considerable scope for doing this
in Middle English studies, and it has been used in the LALME project: for
example, by Laing (1988).
8 Here, I distinguish an 'error' from an 'orderly variant' by considering the
latter to be reasonably frequently attested in a text, and not just once.
Ultimately, there is bound to be some difficulty in making such a distinction
in every instance — partly because what appear to be 'errors' may sometimes
be the beginnings of linguistic changes.
9 The parentheses enclose variables. Thus (ht) is a variable which may be
differentially realised.

206


SYNTAX
Olga Fischer

4.1

Introduction

In many ways 'Middle' is an appropriate term for the syntax of the
period that will be the subject of discussion in this chapter.1 As Roger
Lass says in chapter 2 (section 2.2) of this volume,' middle' indicates the
transitional nature of the language in this period; 'transitional', of
course, only with hindsight. Lass further refers to the typological use
of the term 'middle' within the family of Germanic languages,
representing among other things a language with a relatively 'poor'
inflectional system. Translated into syntactic terms, a 'middle' language
tends to have a fairly strict word order, and to make greater use of
periphrastic constructions; i.e. it relies more heavily on auxiliary verbs,
prepositional phrases, etc.
Compared with the Old English period, when the syntax of the
language was relatively stable (see vol. I, section 4.1), the Middle
English period is indeed one of change. Much has been written about
the causes of the rapid loss of inflections, which started in the Late Old
English period in the northern part of the country and which was more
or less concluded in the fourteenth century with the exception of some
enclaves in the extreme south. Without doubt the fact that Old English
had initial stress played a role. It must have contributed to the
neutralisation of vowel qualities in inflectional endings and their almost
total subsequent demise. However, when we consider the fact that other
Germanic (initial-stress) languages did not all lose their inflections, it
cannot have been a decisive factor. More important may have been the
influence of the Viking settlements in the Danelaw, which, according to
some scholars, led to a process of pidginisation, with a concomitant loss
of morphological structure and the development of a more analytic

207