Hristov, B. (2020) - Grammaticalising The Perfect and Explanations of Language Change
Hristov, B. (2020) - Grammaticalising The Perfect and Explanations of Language Change
Hristov, B. (2020) - Grammaticalising The Perfect and Explanations of Language Change
Series Editor
Jóhanna Barðdal
(Ghent University)
Consulting Editor
Spike Gildea
(University of Oregon)
Editorial Board
volume 10
By
Bozhil Hristov
LEIDEN | BOSTON
Library of Congress Cataloging-in-Publication Data
Typeface for the Latin, Greek, and Cyrillic scripts: “Brill”. See and download: brill.com/brill-typeface.
ISSN 2211-4904
ISBN 978-90-04-41432-7 (hardback)
ISBN 978-90-04-41405-1 (e-book)
∵
Contents
Acknowledgements xi
List of Abbreviations xiii
Transliteration from Cyrillic XV
1 Introduction 1
2 Theoretical Preliminaries 9
2.1 Evolutionary and Functionalist Models of Language Change 9
2.1.1 Evolutionary Linguistics 9
2.1.2 The Concepts of Function and Functional Load, the Invisible Hand
and Teleology 12
2.1.3 Problems and Challenges 16
2.1.4 Interim Summary and Preview 19
2.2 Grammaticalisation and Reanalysis 20
2.2.1 Grammaticalisation 20
2.2.2 Reanalysis in the Context of Grammaticalisation 25
2.2.3 The Role of Frequency and Contact in Grammaticalisation 27
2.3 Other Preliminaries 28
2.3.1 Progress 28
2.3.2 Randomness 29
2.3.3 Justification 30
2.4 Conclusion 30
9 Conclusions 329
References 335
Author Index (Modern Authors) 356
Language Index 360
Subject Index 362
Index of Sources 367
Acknowledgements
First of all, I would like to thank Mira Kovatcheva, my teacher and esteemed
colleague, for her generous professional guidance, and more specifically for
feedback on abstracts, book recommendations and the unstinting expendi-
ture of time and effort to improve the draft (which aroused the curiosity of
Charlie the cat, too). I am indebted to Mitko Sabev, Mark Hadley, Michael
Everdell, Vladimir Bondar, Florian Dolberg and Steven Kaye for supplying me
with papers and/or various insights, as well as to Evgenia Pancheva for ad-
vice on picking mediaeval literary texts. During pleasant conversations about
things literary, philosophical and philological, she and I were unable to decide
whether we should opt for the spelling Hilendar or Hilandar. Stefan Stefanov
and Theodore Todorov readily and enthusiastically helped with Latin even
during unsocial hours. The anonymous reviewers for the LAGB, ISLE 2016 and
ICEHL XX conferences made useful contributions and raised some very per-
tinent issues. The financial support and travel grants from Sofia University’s
Faculty of Classical and Modern Languages and Doctoral School, as well as
the Erasmus+ exchange programme, are gratefully acknowledged. Part of the
research was finalised and made possible by a Fulbright grant at the University
of Texas at Austin. Thanks go to the audiences at the LAGB 2016 meeting in
York, the ‘Gradience and Constructional Change in English’ workshop in
Edinburgh (2017), the 20th International Conference on English Historical
Linguistics (2018), as well as the Syntax/Semantics and Historical Linguistics
discussion groups at UT Austin, for their intelligent questions and comments,
especially David Beaver, John Beavers, Theresa Biberauer, Mary Blockley,
Vladimir Bondar, Bernard Comrie, Hubert Cuyckens, Patience Epps, Michael
Everdell, Teresa Fanego, Carole Hough, Cristian Juares, Danny Law, Benjamin
Molineaux, Muriel Norde, Peter Petré, Susan Pintzuk, Tamara Bouso Rivas,
Hendrik De Smet, Graeme Trousdale, Turo Vartiainen and Steve Wechsler. I
would also like to express my deep sense of gratitude to the participants in the
2017 Winter School at the New Bulgarian University, especially Elena Savova,
Maria Neikova, Maria Stambolieva, Milka Hadjikoteva and Zhivko Hristov,
for their warm reception and thought-provoking feedback. I am grateful to
the members of staff at the University of Roehampton, London, who made
it to the Research Seminar in February 2017, especially Eva Duran-Eppler and
Evangelia Sifaki. As ever, Alexandra Bagasheva was a knowledgeable point of
contact for references and many fruitful and enjoyable discussions. Andrei
Stoevsky, John H. McWhorter and Matti Kilpiö kindly provided me with some
xii Acknowledgements
very important material, going to a lot of trouble posting it and making sure it
reached me. Members of Sofia University’s Faculty of Slavonic Languages, in-
cluding Tatyana Slavova, Anna-Maria Totomanova, Gergana Ganeva and Petko
Petkov, allowed me to sit in on their classes on Old Bulgarian and the History
of Bulgarian. My interest in these subjects was first sparked by Catherine
Mary MacRobert at Oxford. Yoana Sirakova from the Department of Classics
let me join her introductory Latin course. Thomas Kohnen was a lovely host
and granted me access to the library collections at the University of Cologne.
Andrei Stoevsky, Tatyana Slavova, Gergana Ganeva and Mary Dalrymple made
helpful comments and suggestions on earlier drafts. Gergana Ganeva further
assisted with various matters, especially those related to Middle Bulgarian and
the history of the language. Relying on his meticulous eye for detail and native
intuitions, Steven Kaye read the text and made numerous stylistic and linguis-
tic improvements. Colleagues from the English Department at the University
of Sofia, including A. Bagasheva, E. Slavova, M. Kolarova, M. Kovatcheva, N.
Tincheva, N. Yakimova, R. Ishpekova, and T. Venkova, provided feedback on the
first draft and/or words of encouragement during a stimulating departmental
discussion. Denitsa Bozhilova double-checked some of the basic maths and
statistics. Last but not least, I would like to thank Brill’s editors Elisa Perotti
and Maarten Frieswijk, as well as the series editor Jóhanna Barðdal, for their
high professional standards and commitment, including the numerous help-
ful suggestions on improving the draft. I am likewise much obliged to Bridget
Drinka for taking on the task of reviewing the complete manuscript and for
her constructive feedback and valuable pointers. Like the families of many
other academics, mine have also been unfailingly patient, supportive and un-
derstanding during the long hours of writing, including even my little nieces.
Needless to say, all mistakes, omissions and flaws remain my own.
Abbreviations
A adjunct
acc accusative (case)
act active (voice)
adj./adj adjective
adv adverb
AdvP adverb(ial) phrase
AmE American English
aor aorist (tense)
aux auxiliary (verb)
BNC British National Corpus
BrE British English
CO object complement
COCA Corpus of Contemporary American English
COHA Corpus of Historical American English
comp complementiser
conj conjunction
CONTAmE Contemporary American English
CONTBrE Contemporary British English
conv conversation
dat dative (case)
def definite (article)
DP determiner phrase
EME Early Middle English
EModE Early Modern English
F feminine (gender)
fut future
gen genitive (case)
HPSG Head-Driven Phrase Structure Grammar
imperf imperfect (tense)
impers. impersonal
inf/inf. infinitive
injc interjection
ipfv imperfective (aspect)
L Latin
LFG Lexical Functional Grammar
LOE Late Old English
m masculine (gender)
xiv Abbreviations
ME Middle English
ModE Modern English
n. noun
neg negative
nom nominative (case)
NP noun phrase
nt neuter (gender)
O object
OB Old Bulgarian
OCS Old Church Slavonic
OE Old English
OED Oxford English Dictionary
ON Old Norse
pass passive (voice)
PDE Present-Day English
pfv perfective (aspect)
pl plural (number)
PP prepositional phrase
pptcp past participle
Pred predicate
pref prefix
pres present (tense)
ptcp participle
refl./refl reflexive
S subject
sb. somebody
sbjv subjunctive (mood)
SG singular (number)
V verb
voc vocative (case)
VP verb phrase
Transliteration from Cyrillic
а a п p ы y
б b р r ѣ ĕ
в v с s ѧ ę
г g т t ѫ ǫ
д d у u
е e ф f
ж ž х x
з z ц c
и i ч č
й j ш š
к k щ št
л l ъ ă
м m ь ’
н n ю ju
о o я ja
Note: The jer vowel symbols, ъ and ь, are retained for earlier stages of
Bulgarian. Alternatively, they can be transliterated as follows: ъ = ŭ, ь = ĭ.
A Greek-style ω (omega) appears in some of the Cyrillic texts and is
transliterated as o.
chapter 1
Introduction
As the title suggests, this book aims to cover ground at the intersection of three
broad subfields of linguistics: grammar (in the sense of morphology and syn-
tax), the linguistic history of English and Bulgarian, and historical linguistics.
I set out to investigate aspects of the morphosyntactic structure of the two
languages selected, as well as how their grammars came to be the way they
are today. Unsurprisingly, the discussion of each enlightens the discussion of
the other in ways unattainable by considering them separately. Throughout
the text, I highlight parallel explanations of what appear to be nearly identi-
cal phenomena in a pair of rather distant relatives, one of them much less
studied than the other. It is therefore attempted to fill some of the research
gaps (even for a language as extensively investigated as English) and situate the
work in the context of a more general theory of language change as applied to
the events of interest here. Marshalling extensive data from two languages as
opposed to one, and at the same time touching upon other varieties such as
Latin, German and Macedonian, shifts the focus away from English, which still
dominates the literature, and also lends more empirical, areal and typological
support to the general theoretical conclusions. The book builds on an enor-
mous amount of scholarship accumulated so far, reassesses and revises some
long-held notions and shines the spotlight on certain areas that have received
less attention, such as the role of ambiguity in actual usage. The detailed anal-
ysis of rich, contextualised material from a selection of texts dovetails with
large-scale corpus studies, complementing their findings and adding value to
our understanding of the processes under discussion. This monograph thus
presents a happy marriage of traditional philological techniques, which retain
their worth and usefulness, and recent advances in theoretical linguistics and
corpus work.
The following chapter addresses the theoretical preliminaries. It contains
a brief overview of evolutionary models of language change, adopted into lin-
guistics after the momentous breakthroughs of 19th-century biology, includ-
ing some functionalist ideas associated with the same theoretical affiliation.
A cautionary note is sounded to the effect that, neat and attractive though
these models may seem, there are certain problems arising from the transfer
of biological ideas to the study of language. Special attention is paid to central
notions such as function and functional load, as well as the ‘invisible hand’
in language and teleological (i.e. purpose-driven) change, some of which are
taken up in later chapters. I then proceed to flesh out the basic principles of
grammaticalisation and reanalysis, since this monograph is informed by the
broad field of grammaticalisation studies. Some of the pertinent properties of
grammaticalisation are singled out and briefly explained, among them the role
of frequency and language contact in grammaticalisation, as well as some well-
established tenets with a bearing on its nature and modus operandi.
Having thus set the scene by going through the major points of theoretical
interest, in Chapter 3 I trace the story of the perfect in English. Both have- and
be-perfects are considered, starting from Old English times, and are compared
to other periphrases, such as the passive. The origins of the English perfect
(and passive) go back to copular and possessive constructions with verbs
meaning ‘be’, ‘become’ and ‘have’, combined with past participles which origi-
nally showed agreement either with the subject or with the object (depend-
ing on the construction). The earlier participial (and hence adjectival) subject
or object complement gradually came to be reanalysed as a lexical verb. The
emphasis here falls on the morphological marking on the participle, includ-
ing ambiguous cases which are often said to have triggered this reanalysis and
the subsequent loss of agreement. Crucially, I stress that there are attestations
which demonstrate that overt agreement on the participle is not incompatible
with a bleached meaning of have serving as a perfect auxiliary (rather than a
lexical verb indicating possession).
The chapter’s second concern is with the origin and status of the be-perfect
in English, as well as the competition between be and have in this domain, cul-
minating in the ultimate disappearance of be as a viable perfect auxiliary for
so-called ‘mutative’ intransitive verbs. This is interesting because it will throw
into sharp relief the properties of both auxiliaries, at the same time inviting
some more general conclusions regarding the nature of language change.
Tying these points to the content of the opening chapter, I recount some of
the traditional explanations offered in the literature for the displacement of
be by have. One explanation relies on the greater type and token frequency
of have in this function. A particularly popular hypothesis rests on claims that
be was ill-suited for the role in question since it came to be overloaded with too
many functions – in perfect, passive, as well as resultative stative clauses. This
could lead to rampant ambiguity, which is thought to be generally disfavoured
in some evolutionary-based functionalist models of language change. The ulti-
mate victory of have is assumed to have resolved such ambiguities and relieved
be of some of its undesirably heavy functional load.
And yet, many problems beset such a functionalist story, neat though it
may seem at first sight. One major issue is the unexpected proliferation of the
functions of have itself. If it won the ‘battle’ for the domain of expressing the
Introduction 3
context or individual usage elsewhere in the language user’s output. The mate-
rial thus collected has been divided into three categories – examples where
overt agreement is present, examples where overt agreement is missing, and
examples where zero marking is expected anyway. Both have- and be-perfects
are taken into account and comparison is sought with passive or potentially
stative constructions with beon ‘be’ and weorðan ‘become’.
The study confirms that there is a substantial amount of expected zero
morphology in the perfect, as well as in (proto-)passives built with beon and
weorðan. However, while the prevalence of expected zero exponence may ap-
pear to have led to an almost complete loss of agreement in the perfect, par-
ticularly evident in the plural, this is certainly not the case in the passive. Both
types of passive construction, with beon and weorðan, demonstrate a situation
in which zero morphology is used in the singular (legitimately with the mascu-
line and neuter, not so with the feminine), whereas -e appears consistently in
the plural (unlike the perfect). This state of affairs is replicated by the ambigu-
ous passives or statives. Crucially, these findings call into question the claims
that cases of expected zero morphology were responsible for the loss of agree-
ment across the board, and possibly point to different rates at which the gram-
maticalisation of the passive and the perfect proceeded. Finally, this chapter
also offers some corroborating evidence that the displacement of perfect be by
have was already underway in Old English and might be at least in part due to
contact with Old Norse.
Chapter 5 charts the further development of the perfect from Middle into
Modern English based on a selection of texts. In view of the primacy of speech
in language change, I have sought texts which would approximate the spoken
medium. The first genre of proven suitability can roughly be summed up as
comedies or comic tales and comprises excerpts from Chaucer (14th c.), a 15th-
century mystery play, a Shakespearean comedy (16th c.), and a Restoration
comedy (17th c.). I additionally examine the late Middle English correspon-
dence of the Paston family and the 17th-century correspondence of Samuel
Pepys. A quick peek is taken at Victorian English, analysing some material from
Dickens. The developments in English are compared to German, represented
in Chapter 6 by a famous 18th-century play by Lessing. Standard German is a
close relative which took a distinctly different course from English, but offers
some curious analogues nonetheless. The examination (albeit cursory) of an-
other Germanic language might help to disentangle genetic ‘predispositions’
from ‘external’ factors such as language contact – English and German started
from a similar starting point, but something must have caused them to come
out rather differently with regard to their be-perfects. At least in part, the di-
vergent developments must be due to the different historical pressures and
contacts that the two languages experienced.
Introduction 5
Bulgarian National Revival, still allows be-perfects and aorists to narrate non-
witnessed events. What is new here is the frequent omission of the perfect
auxiliary in order to express non-witnessed events with a stand-alone parti-
ciple terminating in -l, i.e. the appearance of evidentials proper. Crucially, the
advent of evidentials has not led to an immediate deployment of have-perfects
in order to alleviate the burden of be. As in the earlier centuries, there are
nevertheless examples of delexical have which might be seen as intermediate
stages in the grammaticalisation of this verb. It would be perilous to posit a
causal relationship between the rise of evidentials and the new have-perfect, in
view of the rather long period of reasonable stability in a system which coped
marvellously without have-perfects for centuries, traditionally relying on be-
perfects (with or without auxiliaries) for mediated reports, as well as in other
situations. That is to say, the material does not show a significant widening or
narrowing in the functional scope of the be-perfect for hundreds of years – we
only see structural omission of the auxiliary and shifting frequencies, but the
range of available functions remains the same. I finally conduct searches in the
Bulgarian National Corpus and mid-20th-century fiction in order to ascertain
how well established the have-perfect is in contemporary Bulgarian. Judging
by my findings, this construction still proves to be rather rare and insufficiently
entrenched.
My general conclusion regarding functional explanations is that, in English
and Bulgarian alike, the grammaticalisation paths of be and have are more
likely to have been steered by their non-specific meanings, which easily lend
themselves to reanalysis and greater abstraction (instigated by contact or
just everyday monolingual interaction), rather than by any perceived ‘func-
tional needs’ or ‘preferences’ of the languages or their speakers. One should
be mindful of the dangers of equating categories from different languages or
even different stages in the history of the same language, since the nature of
each depends on links and oppositions within the synchronic system. And yet,
some thought-provoking parallels could tentatively be drawn. Despite instanc-
es of default agreement, agreement still appears more robust in Present-Day
Bulgarian compared to Old English, which might have to do with the bet-
ter preserved overall morphology of Bulgarian and/or the large-scale loss of
Germanic morphology due to lack of phonological prominence on inflectional
endings. Language-internal and cross-linguistic comparisons show that am-
bivalent marking does not play as important a role of bridging context in re-
analysis and grammaticalisation as has been claimed.
In sum, the two language groups chosen for the present study, Germanic
and Slavic, exhibit both fascinating similarities and palpable differences
with regard to building their perfects. They have been selected as typical
8 chapter 1
Theoretical Preliminaries
This chapter lays the theoretical groundwork on which the rest of the book will
be based. Section 2.1 traces the adoption into linguistics of ideas from biology
and economics, including functional adaptiveness and functional load, teleol-
ogy, as well as non-teleological intentional mechanisms such as the invisible
hand in language change (2.1.2). Problems and challenges for these models
are discussed in Section 2.1.3. Section 2.2 then goes on to outline the central
concepts of grammaticalisation and reanalysis. Also discussed are some of the
features of grammaticalisation that are particularly relevant to the processes
surrounding the rise of the perfect. Special attention is paid to the roles of fre-
quency and language contact in grammaticalisation (2.2.3). Finally, Section
2.3 briefly touches upon the notions of progress and randomness in language
change, asserting that a great deal of change is simply arbitrary; the chapter is
rounded off by justifying the need for a study of this kind (2.3.3). The conclu-
sion summarises the main points and looks ahead to the rest of the book.
process, dependent on various aspects of the context as the speaker sees them
(the microlevel). Selection, however, is not an entirely cognitive process but a
predominantly social one (the macrolevel). The contexts are no longer in the
control of the individual speaker, hence an invisible controller takes over (to
be discussed below; cf. Mathieu and Truswell 2017). Speakers and their utter-
ances can thus be viewed as objects or microlevel actants, while language is a
self-organising system on the macrolevel.
In essence then, a linguistic change is often taken to comprise several stages,
including the actuation or innovation stage, i.e. the beginning of the change
when someone produces a novel form/meaning/sound etc., followed by its
propagation/implementation or selection – the spread in the linguistic com-
munity. In at least some functionalist evolutionary models, the motivations
and mechanisms for innovation are believed to be functional, whereas the
mechanisms responsible for propagation are evolutionary selection mecha-
nisms, and they are social (Croft 2000: 8, 31–32, 38, 166ff., 184–185, with ref-
erences). These social factors, essentially sociolinguistic concepts such as
prestige, social status or identifying with a social group, have been compared
to the ecological factors in biology. Propagation can be seen as the adoption
of a convention in society. Other functionally-minded scholars, however, seek
functional explanations for the selection stage too, assuming an analogy be-
tween functional explanations in linguistics and adaptive explanations in evo-
lutionary biology (see Haspelmath 1999). In defence of this latter view, it could
be stressed that adoption is also a cognitive process to a certain extent because
the new element needs to be integrated in the individual speaker’s grammar.
As a supporter of the former hypothesis, Croft (2000: 39) counters that
Croft (2000: 195, n. 10) backs up this statement by pointing to the existence of
social situations which might favour linguistic changes that are not considered
to be ‘optimal’ or ‘natural’ on functional grounds. This furnishes evidence that
functional considerations are unimportant in selection (propagation). Rather,
complex older forms, or elaborate ‘dysfunctional’ innovations, might end up
surviving for a long time or even spreading through a speech community.3
3 The dividing lines between the stages are often blurred. For instance, we could think of the
gradual displacement of be by have (discussed below in the context of the English perfect) as
both propagation and also innovation (or a series of micro-innovations). This is highlighted
12 chapter 2
2.1.2 The Concepts of Function and Functional Load, the Invisible Hand
and Teleology
As noted above, fitness in language has to do with function. If a construction
has too many functions, i.e. if it has too great a functional load, it will be too
ambiguous and might be subjected to change (see Durie 1995: 276ff.). As Lass
(1980: 65) reminds us, the idea that function is a significant factor in change
largely goes back to the pioneering work of the Prague School, whose pro-
ponents tended to see change as ‘purposeful’ or ‘teleological’. Of course, the
intellectual roots of functionalism can be traced back much further, as synthe-
sised by Givón (2013). Function is often associated with Humboldt’s Universal,
which requires that (ideally) one meaning should correspond to one form.4
Croft (2009: xiii) states the functionalist position quite succinctly:
one cannot deny the existence of grammar, that is, knowledge of linguistic
conventions in a speaker’s mind; one cannot disentangle this knowledge
from the act of using language; language function influences language
form through the dynamics of language use and language change.5
in the framework of Construction Grammar (see Traugott and Trousdale 2010, 2013, inter
alia).
4 See Lass (1980: 71–75) for counterexamples as well as further references. Itkonen (2013) con-
tains modern echoes of Humboldt’s view, and argues against them.
5 Durie (1995) explores what is meant when we say that a linguistic structure has a certain func-
tion. On the divide (and bridges) between formal and functional linguistics, see Newmeyer
(1991), Hill (1993), Beaugrande (1994), Łęcki (2010: 44ff., with references), Vincent and Börjars
(2010), Bischoff and Jany (2013), Hengeveld et al. (2017). In a nutshell, functionalists see com-
munication as the primary raison d’être of language, to which it is geared and which in turn
shapes it, while formalists view the communicative function of language as largely irrelevant
to its analysis. Haspelmath (2008a: esp. Section 4, 2018) provides a good overview of current
functional explanations in typology, focusing on ways of accounting for language universals
and comparing the functionalist approach to generative syntax, including the Principles and
Parameters framework and Optimality Theory. Bischoff and Jany (2013) and Hengeveld et al.
(2017) are collections of recent work within the functionlist paradigm.
Theoretical Preliminaries 13
6 Whenever such personification occurs in the rest of this book, the caveats from this section
should kick in. For instance, wherever I have used loose metaphorical turns of phrase such as
to recruit an item for grammaticalisation, no purposeful single-mindedness or grand designs
on the part of the language or its speakers have been implied.
14 chapter 2
a teleological process, since the goal of the process needn’t be built in – what is
needed is simply a mechanism that causes different interactors to end up with
differential survival rates. Adaptation is such a mechanism – in it certain prop-
erties of a given organism interact with properties of the environment in such
a way that the likelihood of the organism’s survival is increased or decreased.
Adaptation thus ultimately results in directed selection, whereby a greater pro-
portion of organisms in possession of adaptive properties survive.
On this view, adaptation is neither teleological nor deterministic, since
being in possession of adaptive properties merely increases the chances of
survival, but without guaranteeing survival. However, Croft still believes that,
in a population of ‘organisms’, given enough time, the laws of probability will
direct change in favour of the better adapted organisms. Although this stance
is more sophisticated, in the spirit of the literature on self-organising complex
adaptive systems, it also seems to indirectly amount to a teleological explana-
tion, albeit not a deterministic one. Even when you see directional language
change as probabilistic, rather than deterministic, you still attribute to it the
same functional causal mechanisms.
A careful treatment will thus distinguish different intentional mechanisms.
Teleological intentional mechanisms assume that a speaker innovates in order
to change the linguistic system – essentially Keller’s artifactual phenomena
whereby the linguistic system is shaped by the speaker to have the structure
that it does, as well as to change the way it does. Croft rejects those. He is in fa-
vour of non-teleological intentional behaviour – the speaker produces an inno-
vation because he or she is trying to achieve some other goal in using language.
Intentional processes do not necessarily aim to produce linguistic innovations.
Croft (2000: 65) states that he ‘will use the term intentional to describe mecha-
nisms that are not teleological but involve the intention of a speaker to achieve
some other goal in language use’. He refers to teleological mechanisms as ‘sys-
temic functional explanations’, reserving the term ‘functional explanations’ for
non-teleological intentional mechanisms.
There exists yet another type of causal mechanism proposed for language
change which is neither teleological nor intentional, since language change
is not seen as an intended means aimed at achieving some other goal in lan-
guage use. On this view, a change simply happens as a result of the process
of utterance production. This phenomenon has sometimes been labelled ‘me-
chanical’, especially when applied to production in phonetic articulation. Such
explanations could be proposed for higher-level cognitive processes in the use
of language as well. This kind of causal mechanism is therefore often labelled
as broadly non-intentional (corresponding to Keller’s natural processes). The
distinction can be clarified as follows:
Theoretical Preliminaries 15
The subtle difference between the two hypotheses lies in their different as-
sumptions about what speakers actually ‘wanted’ – to preserve the linguistic
system for its own sake or to avoid misunderstanding (and thereby indirectly
preserving the linguistic system). In other words, does avoidance of homon-
ymy happen for the sake of propping up the system as such or does it follow
from the intention of interlocutors to avoid being misunderstood? This is an
example of how a teleological and an intentional mechanism would basically
amount to the same explanation, despite the fine line that divides them.
whether or not a gap exists in the system. On the one hand, there are many pro-
nominal systems worldwide that have number gaps but do not show any signs
of changing. On the other hand, there are instances of new plural forms aris-
ing in addition to pre-existing ones, i.e. where no gaps existed (see Croft 2000:
69–70, with references). So linguistic systems are not always as symmetrical as
we might imagine; rather, they tolerate a high degree of messiness. To be more
accurate and careful, we should say that speakers exhibit a high level of ‘toler-
ance’ to ambiguity. Croft himself (2000: 71) concedes that
7 See Lass (1980: Chapter 4) for more philosophical arguments against the explanatory power
of functional load.
18 chapter 2
8 Apart from near-mergers, Ringe and Eska (2013: 101–103) report studies that demonstrate
the failure of functional load to account for mergers in phonology more generally. As they
conclude, “[t]hat confirms the consensus of opinion among rigorous historical linguists that
functional load has no demonstrable effect on mergers and should be disregarded in framing
hypotheses about sound change (Ringe and Eska 2013: 102)”. Below, I extend this to morpho-
syntactic change.
Theoretical Preliminaries 19
2.2.1 Grammaticalisation
Grammaticalisation studies had their beginnings in the 18th century. The term
itself is credited to Meillet (1912: 131), dating back to the early 20th century.
However, this field of research really picked up only in the 1970’s, when Givón
(1971: 413) famously observed that ‘[t]oday’s morphology is yesterday’s syn-
tax’ (see Heine 2003: 575–578, 599 n. 4, Hopper and Traugott 2003: Chapter 2,
Traugott 2003: 624, 646 n. 2, 2010, Łęcki 2010: 3ff., Macleod 2012: 5, for an over-
view of the history of research into grammaticalisation).
Grammaticalisation can be regarded as a process (mechanism) of innova-
tion in grammatical change. In it, constructions which are made up of specific
lexical items develop grammatical meanings, which leads to reinterpreting the
lexical items as serving grammatical functions (Croft 2000: 156, cf. Traugott
2002: 22ff., 2003: 645, Macleod 2012: 5ff.). In addition, grammatical forms can
give rise to the development of even more grammatical forms, i.e. ones even
more firmly embedded in the morphosyntactic system. It is also important to
stress that grammaticalisation does not happen to items in a vacuum but to
entire constructions in context, including not just the syntactic environment
but an all-encompassing idea of context as situation and discourse, paradig-
matic relationships, etc. (see Bybee 2003: 602–603, Heine 2003: 575, 580–581,
Hopper and Traugott 2003, Traugott 2003, 2008, Traugott and Trousdale 2013:
Chapter 5). For instance, English has the constructional schema in (1).
Theoretical Preliminaries 21
9 This view is highly compatible with synchronic models of language analysis such as
Construction Grammar (with its definition of construction as a ‘form-meaning pairing’),
LFG, HPSG, as well as hybrids such as Sign-Based Construction Grammar (SBCG) (see
Hollmann 2003: 17–21, Traugott 2003: 625, 2008, 2010: 277–279, Bergs and Diewald 2008,
Sag 2010, Traugott and Trousdale 2010, 2013, Vincent and Börjars 2010, Hoffmann and
Trousdale 2013). Trousdale (2008, with references) and Coussé et al. (2018), among others,
provide diachronic discussion of grammaticalisation in Construction Grammar. Bergs
and Diewald (2008), Hoffmann and Trousdale (2013) and Traugott and Trousdale (2013)
deal with constructions and language change more generally.
10 The discussion here is primarily based on Croft (2000: 156–157) – cf. Danchev and Kytö
(1994), Heine (2003: 580, 584–586), Traugott and Trousdale (2013: 102, 105–107, 116ff., 217ff.),
among others.
22 chapter 2
particular function. It has been argued that this might be driven by expressivity –
a speaker comes up with a novel periphrastic way of expressing a function
which stands out and is thus more effective (see Croft 2000: 159; cf. Heine 2003:
578 and the preceding discussion of the place of expressivity in evolutionary
models). There has been a long tradition of claims that expressivity could be an
important driving force behind grammaticalisation. It was discussed as early
as Meillet (1912, cited in Łęcki 2010: 15), who attributed the emergence of new
grammatical constructions to the loss of the vividness (i.e. the bleaching in
modern parlance) of the older, well-established ones, coupled with the desire
of language users to be expressive. Traugott and Trousdale (2013: 124ff.) provide
critical assessment of such proposals, remarking that speakers are often un-
aware that they might be initiating or propagating a change.
Another related mechanism offered to explain the rise of periphrasis is the
avoidance of misunderstanding (see Heine 2003: 578 and Section 2.1.2 above).
All too often, in fact, it has been proposed that speakers introduce a certain
amount of redundancy in interaction in order to avoid being misunderstood
(Croft 2000: 159, with references). As Keller (1994: 109) puts it, ‘to increase re-
dundancy beyond the acoustically possible, one must use lexical means’, that
is, a periphrastic construction. As the subjunctive inflections in English be-
came eroded to a point where they were hardly distinguishable, speakers re-
sorted to expressing the same non-factual semantics with modals instead.
Semantically, when a linguistic expression A is recruited for grammaticali-
sation, it acquires a second pattern of use – B; there is then an overlapping
stage of ambiguity between A and B, before A is potentially lost and only B
remains (Heine 1993: 48–49, 2003: 579, 587, 589–590, Macleod 2012: 8–11). So
ambiguity is important in this overlap model, as it enables the shift of mean-
ing. Such overlaps will be pivotal in discussing the morphosyntactic marking
of Old English periphrastic constructions in the chapters to follow.
Grammaticalisation can be demonstrated to involve pragmatic inferenc-
ing, also labelled as ‘context-induced reinterpretation’ (see Croft 2000: 160–161,
Traugott 2002, 2003: 634–636, 2010: 279–280, Heine 2003: 580, 586–588, 591–
592; cf. Levinson 2000: Chapter 4, Traugott and Trousdale 2013: 26 fn. 19, 217ff.).
There is normally a continuum of situation types which bridge the gap be-
tween the original meaning of a construction and the new meaning or function
which has arisen as a result of grammaticalisation (with significant overlaps).
Pragmatic inferencing allows speakers to gradually extend the meaning from
one situation type to another (contiguous one). In Henry is going to town/to
the library,11 we have a purely allative situation type of motion from point A
11 The examples in this paragraph come from Croft (2000: 161), sometimes with only cos-
metic modifications.
Theoretical Preliminaries 23
reanalysis, but not all (see Heine 2003: 592–593, 600 n. 6, Hopper and Traugott
2003: 58ff., 69, Łęcki 2010: 32–33, Traugott and Trousdale 2010, with references).
It has been contended, notably by Haspelmath (1998), that grammaticalisation
and structural reanalysis are two distinct processes which should in principle
be kept apart. According to Haspelmath, genuine structural reanalysis involves
no loss of syntactic autonomy or of phonological substance, and it is poten-
tially reversible, unlike grammaticalisation, which involves loss of autonomy
as well as substance, and is usually unidirectional.12
So, some scholars are inclined to see reanalysis as synonymous, or near syn-
onymous, with grammaticaliastion. Others maintain that the two are distinct
and grammaticalisation involves, but is not coextensive with, reanalysis. That
is to say, reanalysis is a mechanism for grammaticalisation, as well as for other
changes. Yet others believe that grammaticalisation does not require reanaly-
sis and the two are not necessarily related. I tend to side with the second group.
Form-function reanalysis is likewise a pervasive mechanism of change
and it is often crucial in innovation, its causes being an integral part of
communication:
12 On the unidirectionality hypothesis, see Heine (2003: 582–583), Hopper and Traugott
(2003: Chapter 5), Traugott (2003: 627ff., 646 n. 9, 2010a: 272–274), Danylenko (2005: 351ff.).
Theoretical Preliminaries 27
one way (see Croft 2000: Chapter 5, Traugott and Trousdale 2013: 217ff.). As with
the material I present in later chapters, the duality of motion and/or future
readings of be going to persists to this day.13
This may often be the case, but ambiguity is not a necessary condition.
Although it is assumed by many that reanalysis presupposes the existence
of alternative surface structural parses, such as ambivalent case marking or
alternative constituency structure possibilities, ‘remappings between gram-
matical form and conventional function may violate existing unambiguous
conventional mappings’ (Croft 2000: 120, cf. Traugott and Trousdale 2013: esp.
199ff.). Therefore, it might be too strong a claim to require formal grammatical
ambivalence for reanalysis to be able to occur. Both reanalysis and ambiguity
will play a central part in the account of the development of the perfect below.
13 On grammaticalisation, reanalysis and implicatures, see Haspelmath (1998), Hopper and
Traugott (2003: Chapter 4), and also Levinson (2000: Chapter 4, esp. 263ff.), where he dis-
cusses the pragmatics of the development of the be-going-to future construction, as well
as the pragmatic aspects of grammaticalisation more broadly, as already noted in the pre-
vious sub-section.
14 Also consult Bybee and Hopper (2001); cf. Hollmann (2003: 102–104, 108 fn. 9, 110 fn. 11,
116–117, 125–126, with references), for alternative and complementary usage-based mod-
els, where (type and token) frequency is paramount too. See Traugott (2010a: 280) for
a brief critical assessment. Traugott (2010b) warns that, although frequency tends to be
construed as a contributing factor, it is not actually a cause. Hoffmann (2005) points to
grammaticalised forms which remain infrequent even post-grammaticalisation. I return
to problems with frequency in Chapter 9.
28 chapter 2
cunnan ‘to know’ as a modal auxiliary – ModE can, Bybee suspects that cunnan
began to grammaticalise as an auxiliary because it was already quite frequent,
having undergone some bleaching of its semantic content. The frequency of
the various competing auxiliaries can likewise be regarded as a crucial ingredi-
ent in their survival during the development of the perfect, as will emerge in
chapters to come.
Finally, it has been suggested that constructions (in the Construction
Grammar sense) can be transferred from one language to another in contact
situations, with areal and contact-based research on grammaticalisation car-
ried out for a number of languages (see Traugott 2010a: 282–283, Drinka 2017:
14ff.). Thus, it has been confirmed that grammaticalisation can sometimes be
motivated externally. In such cases, it is potentially not always easy to tell if
one is dealing with externally-prompted grammaticalisation or with the bor-
rowing of structure. Below we shall see that contact-induced scenarios have
been proposed for the development of the perfects in English and Bulgarian,
as well as Germanic and Slavonic more widely. For instance, the (relatively re-
cent?) have-perfect of Bulgarian is less grammaticalised than those of contigu-
ous Macedonian and Greek in that it does not normally occur with intransitive
verbs (see Chapter 7). Since grammaticalisation has not been implemented to
the same extent, this might suggest that Bulgarian did not straightforwardly
copy the structures themselves but replicated the model of its more advanced
neighbours in an externally-motivated round of grammaticalisation, imitating
the movement towards more grammatical forms (see Aikhenvald 2002, Heine
and Kuteva 2005, Drinka 2017: 285–286).
Before closing off this chapter, I ought to quickly tie three more loose ends –
discrediting progress as a legitimate way of thinking about language change
and endorsing randomness instead (as avowed already), as well as justifying
why it is worthwhile to embark on the present exploratory journey.
2.3.1 Progress
Croft (2000: 78–79, 82) notes that teleology in evolution and in language change
can be associated with the idea of progress, which can be seen as movement
towards some specified (positive) goal. Even if change were progressive (at
least locally, if not globally) in order to achieve better adaptation to the en-
vironment, it is usually hard and subjective to specify what is desirable in the
first place. Furthermore, the conditions (both in biology and in language use)
Theoretical Preliminaries 29
are so changeable that today’s adaptive trait that confers a distinct advantage
might turn out to be rather dysfunctional tomorrow. There are competing mo-
tivations, such as economy of expression, a preference for shorter or even zero
forms, as opposed to iconicity – expressing concepts overtly so that no mis-
understanding occurs. Although it was widely popular in the 19th century, the
idea that there is (global) progress in language is universally condemned by
serious linguists today. As Croft (2000: 80–81) cautions, linguistic subsystems
are all ‘well enough adapted to effective communication that no one structural
type has a decided functional advantage over the others’.15
2.3.2 Randomness
Above I outlined one possible evolutionary model of language change, devel-
oped and articulated by Croft (2000). It can accommodate both teleological
and non-teleological mechanisms, as well as intentional and non-intentional
behaviour. Croft also envisages other alternatives. A fairly simple alternative
theory to his evolutionary account involving non-teleological intentional
behaviour is that language change is random. Randomness can easily be
subsumed under a generalised theory of selection, since it is a mechanism un-
derlying replication. Croft (2000: 42) goes on to make the following admission:
15 See Croft (2000: 78–84) for further refinements and a more sophisticated treatment of the
topic. Cf. Sampson et al. (2009) and Aitchison (2013).
30 chapter 2
chapters, they do not appear to follow from such intentional human behaviour
as has been suggested.
2.3.3 Justification
Some of the anonymous reviewers for the LAGB and ISLE 2016 conferences
cast doubt on the seriousness of the issues involved here, arguing that an ap-
peal to (naïve) teleology might be a straw man. One of them states that
2.4 Conclusion
Function and functional load, including the avoidance of ambiguity, have been
central in this introductory overview. It was demonstrated that accounts ap-
pealing to linguistic fitness, in the sense of avoiding ambiguity and/or func-
tional overload, are highly suspect, even though they have been, and continue
to be, invoked on countless occasions. Their predictive power will be chal-
lenged in the chapters to come, where I zero in on the role of ambiguity and
functional load both on the level of the (abstract) linguistic system, as well as
in actual output. By now, the intellectual climate has shifted: naïve teleology
Theoretical Preliminaries 31
Synthesising and reassessing a vast body of research, the present chapter docu-
ments the development of the perfect in English. Section 3.1 deals with per-
fects built with have, paying special attention to the morphological marking
and the ambiguity which accompanied the reanalysis of have as an auxiliary in
the Old English period (3.1.2). Section 3.1.3 is an excursus into a short-lived rival
construction relying on a verb for ‘own’, before I consider how have really took
off from the Middle English period onwards (3.1.4). Section 3.2 is dedicated
to perfects with be, whereas Section 3.3 chronicles the competition between
the two auxiliaries, culminating in the loss of be. Echoing the major themes
of the previous chapter, in Section 3.4 I recount some of the explanations for
the demise of perfect be from the specialist literature, in particular: the greater
frequency of perfect have; the functional overload and the ambiguity of be as
a copula and an auxiliary for the perfect, the progressive and the passive; the
disappearance of alternative passive auxiliaries, which ‘conspired’ to overload
be even further; as well as unclear morphosyntactic contexts involving coordi-
nation, contraction and ellipsis. Section 3.5 raises certain concerns about the
validity of some of the earlier explanations: I dwell on the interaction between
morphological marking and perfect readings (3.5.1), the polygrammaticalisa-
tion of have itself, which resulted in an explosion in the number of functions
it took on (3.5.2), as well as the development of alternative perfect and passive
constructions, notably with get, in addition to a comeback of perfect be (3.5.3).
In the light of frequency explanations, Section 3.5.4 presents some statistical
trends which emerge from the analysis of corpus data and lend credibility to
frequency as a contributing factor. In Section 3.5.5, I propose other convincing
scenarios about the displacement of be from the perfect domain, including
explanations appealing to language contact, before finally wrapping things up
in the conclusion.
1 See Denison (1998: 135); cf. Mustanoja (1960: 499), Strang (1970: 352), Smirnickaja et al. (1977),
Bybee and Dahl (1989: 67ff.), Rydén (1991: 345–346), Elsness (1997: Section 4.2.2), Łęcki (2010:
Chapter 4, esp. 204), Macleod (2012: 18).
The Story of the English Perfect 35
The syntactic origins of the perfect clearly illustrate the point made in
Chapter 2 that it is constructions that grammaticalise, not isolated lexical
items (cf. Heine 1993: 30). Clauses like (1) are the prototype of what was later
to become the periphrastic verbal construction which is our topic here. The
stative meaning of possession gradually gave way to a dynamic reading, and
what used to be a transitive clause with an object and a participial object
complement, as a result of reanalysis, became a periphrastic perfect with an
auxiliary and a main verb.2 After the auxiliation of have, the original object of
have came to be the object of the participle (now the main verb). The semantic
and pragmatic change behind this development was from a meaning of pos-
sessing/having something in a certain state, via denoting a state as a result of
previous action, to the meaning of completed action.3
The different stages of this development can be schematically represented
as follows:
table 1 Stages in the development of the English perfect (based on Łęcki 2010: 149)
2 Current syntactic theory is increasingly coming to treat traditional auxiliary verbs as heads
which take what is traditionally labelled the main verb as a dependent complement (see
Warner 1993: 19ff., Huddleston and Pullum et al. 2002: 104, 1209–1220; cf. Heine 1993: 6–7,
99, 106, Chankova 2008: 40ff., Hristov 2015). On this analysis, has written is not part of the
paradigm of write and, according to many contemporary researchers, the two verbs project
their own separate clauses or phrases – i.e. has and written appear in their own separate VPs.
Although there is solid evidence in favour of this analysis, I will stick to the traditional and
constructionist view, whereby the auxiliary forms a unit with the main verb, of which it is
a dependent. This choice has been dictated mostly by reasons of expository convenience,
sacrificing syntactic faithfulness and rigour (cf. Bowie et al. 2013: Section 1.2). In an elegant
synthesis, Brown and Miller (2016: 89–90) reconcile the two views.
3 See Jespersen (1931: 29–30), Mincoff (1958: 131), Visser (1963–1973: §§2001–2003, §2028, §2043
(on the analogous origin of the infinitival construction to have killed him), §2116, §2189),
Mitchell (1985: §§702–733, esp. §§724–728), Rydén and Brorström (1987: 16–17), Brinton
(1988: 99–100, 107–108), Kovatcheva (1989: 135ff.), Traugott (1992: 190–192), Denison (1993:
364), Warner (1993: 97), Bybee et al. (1994: Chapter 3, esp. 68), Fischer and van der Wurff
(2006: 139ff.), Kilpiö (2007), Trask and Millar (2007: 175–177), Chankova (2008: 71, 218, 2009:
425), Łęcki (2010: 145ff.), Macleod (2012: 11–12), Johannsen (2016, for a recent reassessment),
as well as the Dictionary of Old English (http://tapor.library.utoronto.ca/doe/, accessed on
27 Feb 2017). OE adjectival paradigms can be found in Brinton and Arnovick (2011: 208–209)
and in Table 2 below.
36 chapter 3
table 1 Stages in the development of the English perfect (based on Łęcki 2010: 149) (cont.)
Reanalysis
(ultimately leading to a change in word order)
Stage III: He (subject, agent) has (auxiliary) meat (direct object, patient) cooked
(main verb)
Stage IV: He (subject, agent) has (auxiliary) cooked (main verb) meat (direct object,
patient)
Analogy
(extension to intransitive verbs)
Stage V: He (subject, agent/patient/undergoer) has (auxiliary) cooked/died (main
verb)
Mitchell (1985: §24) stresses that in the proto-periphrasis with habban, the par-
ticiple was often declined to agree with the object and could be regarded as a
passive participle used as an adjective. When the periphrasis became a genuine
verbal combination, we no longer had an adjectival use of a passive participle,
but an active verbal use instead. In other words, in the original proto-perfect
complex transitive construction, the transitive participle retained its passive
meaning (e.g. ‘He had the song [in the state of having been] sung’), which was
lost with the transition into a purely verbal, active combination (e.g. ‘He had
sung the song’).
A Modern English construction which has been claimed to be directly de-
scended from the original ‘proto-perfect’ is sometimes dubbed the conclusive
perfect (see Denison 1993: 343–344, 364, 1998: 135):
(2) The way he kept on, he had me almost convinced Hickock and Smith were
innocent. [1965, from Denison (1993: 343)]
As is often the case in linguistics, the road gets bumpy even before we have
started our exploratory journey: criticism has been levelled at the received
wisdom presented above, casting doubt on the evidence in favour of the OE
stative meaning of possession, in addition to the weak chronological evidence
for the spread to actional meanings (see Brinton 1988: §3.1.3; cf. Denison 1993:
365, Macleod 2012: esp. 234). For the time being, however, I shall uphold the
traditional account, in spite of the validity of such concerns.
Interestingly, in Old English the participle did not always agree with the ob-
ject, as becomes apparent from (3). The accusative masculine singular agree-
ment marker -ne is missing from befangen (cf. (1) above and Table 2 below), in
which case it is an open question whether we have an SOVC O clause or wheth-
er it is better interpreted as SOV AUX V MAIN (see Denison 1993: 340–341; cf. Strang
1970: 311, Traugott 1992: 190).
The traditional endings of the strong declension in the nominative and accusa-
tive are set out in the following table:
table 2 Partial strong adjectival declension in OE (based on Brinton and Arnovick 2011:
208, Kovatcheva and Sabev 2015: 145)a
Elsness (1997: 241–242, with references) reviews claims and counterclaims that
the grammaticalised OE perfect was to be distinguished from its precursor on
the basis of the presence or lack of agreement. Denison (1993: 341) offers the
following criteria for determining whether a construction is best analysed as
an SVOC O clause with lexical have or as a genuine perfect syntagm with auxil-
iary have (SV AUX OV MAIN ). SVOC O presupposes the following:
– a sentence brace in main clauses, with non-adjacency of finite have and the
participle;
– accusative adjectival inflection on the participle;
– a stative context where have can mean ‘possess’ and both have and the par-
ticiple can refer to states.
By contrast, the analysis of have as an auxiliary presupposes the following:
– adjacency of have and the participle;
– no adjectival inflection on the participle;
– similar patterns in a non-stative context;
– similar patterns with subjects that cannot be possessors;
– similar patterns with objects that cannot be possessed;
– similar patterns with participles that do not take an accusative direct object.
For problems arising from such rigid distinctions, see Mitchell (1985: §727) and
Kilpiö (2007: 337ff.) – inflection, for instance, can be present or absent in what
is essentially the same clause with the same meaning (cf. (6) below).
In fact, Mitchell (1985: §§709–710) and Denison (1993: 346) report studies
which show that agreeing inflected forms were much fewer in number than
non-agreeing ones, and became even less common over time. Brinton (1988:
101) notes that the majority of participles in OE are uninflected (cf. Fridén 1948:
41, Mitchell 1985: 283–284, 292, Brinton and Arnovick 2011: 229). It can even be
claimed that OE participles are ‘normally invariable’ (Quirk and Wrenn 1957:
75). Strang (1970: 305) points out that the participle was not inflected in the
period between 970 and 1170, though inflection was still present between 770
and 970. Traugott (1992: 190) concurs that inflected forms became less frequent
during the OE period, although they were never predominant in the first place.
Mitchell (1985: §710) quotes publications which record between fourteen and
The Story of the English Perfect 39
An attested example in which there would have been no inflection (or zero in-
flection) in any case is provided in (5). The neuter singular object interrogative
pronoun triggers no overt ending on the past participle gedon. It is impossible
to tell if this is zero agreement (i.e. agreement works but the exponent in this
slot of the paradigm happens to be a meaningful zero morpheme in paradig-
matic opposition to overt markers in other slots), or whether it is a breakdown
in agreement (i.e. agreement no longer functions and the speaker/writer did
not intend to signal the relationship between the two elements by copying the
features of one onto the other).4
4 Haas (1957) is a classic on the theoretical implications of positing zeros in linguistic descrip-
tion. In Chapters 7 and 8, it will emerge that, although there are few such morphosyntactically
40 chapter 3
We are thus dealing with two types of lack of agreement – audible/visible mor-
phology is legitimately missing in (5) because of zero exponence for this slot in
the paradigm, so agreement can be thought of as ‘covertly present’; (5) cannot
have overt agreement (cf. Table 2 above). In (6) on the other hand, the overt
plural -e marker fails to appear on one of the two participles. We can tell that
a breakdown has occurred. The first type is said to have opened the floodgates
for the second. The above sources imply that reanalysis is to a great extent due
to the syntactic ambiguity of the construction, which in turn existed because
there was no overt morphology to dispel it. Trask and Millar (2007: 177) are
ambiguous cases in Bulgarian and Macedonian, the same reanalysis has also occurred or is
occurring. So it cannot be just form that matters – there must be something else too.
5 More examples like this one can be found in Mitchell (1985: §711), Kilpiö (2007: 328) and
Łęcki (2010: 173, 229 n. 29).
The Story of the English Perfect 41
rather more cautious in stating that ‘reanalysis may be favoured by the absence
or loss of overt morphology which is inconsistent with the reanalysis’. The cru-
cial difference boils down to whether the reanalysis was caused or possibly just
favoured by ambiguity.
Conversely, inflections could appear when they were not expected, perhaps
due to scribal errors or analogical levelling. In (7), hus is a neuter plural object
and it should not trigger inflection on the past participle (which should surface
as berypt); however, the (originally) accusative plural masculine form berypte
turns up instead (see Łęcki 2010: 172, Table 2 above and Section 3.2 below).
At the end of the day, it remains a moot point whether habban was a genu-
ine auxiliary in Old English (see preceding discussion, as well as Mitchell
1985: §§729–730, §733, Fischer 1992: 250, Denison 1993: 340, Łęcki 2010: 164ff.).
Crucially, perfect progressives and perfect passives did not appear until Middle
English, pointing to an incomplete grammaticalisation cycle in Old English.6
According to Mitchell (1985: §753) and Łęcki (2010: 188), at least one of these
gaps might be due rather to the later development of the progressive. What is
more, the OE verb agan ‘own’ is also to be met with in perfect-like construc-
tions instead of habban, suggesting uncertainty and lack of a fully fixed and
grammaticalised periphrasis (see Brinton and Arnovick 2011: 228; cf. the gram-
maticalisation tendencies outlined in Chapter 2 above and the brief discussion
of agan in Section 3.1.3 below).
On the other hand, Traugott (1992: 191) reminds us that the use of the hab-
ban construction with intransitive verbs from very early on, as well as its use
with transitive verbs which had non-accusative objects (where there was no
6
See Mustanoja (1960: 440, 516–519 on perfect infinitives in ME, 590–591), Mitchell
(1985: §702, §753, §§782–785, §922 on perfect infinitives in OE), Fischer (1992: 276–277),
Chankova (2008: 138, 185–186), Brinton and Arnovick (2011: 373); cf. Łęcki (2010: 184–188 on
perfect infinitives in OE). Denison (1998: 183–184) notes that even in earlier Modern English
the combination of perfect and passive seems to have frequently been avoided, (i), as it still
is in Dutch and German (cf. Jespersen 1909–1949: IV 102–104, Visser 1963–1973: §793, §1909).
(i) but that objection is done away with [1818, Keats, Letters 66, p. 146 (21 May), quoted in
Denison (1998: 183)].
42 chapter 3
As pointed out by Łęcki (2010: 173), the problematic inflection on the participle
bewrigen-e cannot be used as evidence in favour of a non-perfect stative/ad-
jectival reading because of the semantics of the clause. The strong plural (?) -e
might be due to attraction to plural gimmas or it might somehow agree with
neuter singular treow. In that case the form should probably be emended to
strong neuter singular bewrigen, since bewrigene is unlikely to be a weak neuter
singular form (cf. Table 2 above). Mitchell (1985: §712) is more inclined to treat
the whole thing as a scribal error, noting that the participle is preceded by a
plural.
Going back to the debate surrounding the status of OE habban, Denison
(1993: 352) lists the following stages in the grammaticalisation of the perfect
construction with have as possible indicators of the establishment of the per-
fect category:
a) when the have-perfect became available for any lexical verb which did
not conjugate with be (late Old English?)
b) when it had come to be a pure tense equivalent (late Old English?) – see
Section 3.1.4 below
c) when it had developed approximately its present-day meaning (seven-
teenth century?) – which involved the loss of b)
d) when it became available for every non-auxiliary verb (late Modern
English)
The Story of the English Perfect 43
Denison settles on stage a) as the point at which have had attained true
auxiliary status, since it was used ‘without an argument structure or selec-
tional restrictions of its own’ – decategorialisation being a typical feature of
the grammaticalisation process, as is the bleaching of the original meaning,
in our case the meaning of possession (see exx. (8) and (9) and Chapter 2,
Section 2.2.1).
7 There may be alternative interpretations of these verse lines whereby ah lifes wyn and ge-
biden do not belong together – see http://www.anglo-saxons.net/hwaet/?do=get&type=text
&id=Sfr&textOnly=false (accessed on 1 June 2017).
8 See Mustanoja (1960: 504–505), Fischer (1992: 250, 257), Elsness (1997: 252–253, 264–270, 277,
340–341), Smith (2001: esp. 375, also tracing later periods), Chankova (2007a: 44, 2007b: 510,
2008: 102, 129ff., 152, 201, 216).
44 chapter 3
instances of this verb in (what is traditionally considered) the same clause, one
of them must be an auxiliary (see Łęcki 2010: 171 and the remarks in Chapter
2, Section 2.2.1 above concerning the survival of the original lexical item post-
grammaticalisation; cf. am being, going to go, there is something there).
(11) We habbeð ihaued moni burst, moni hunger, & moni þurst.
we have had much need much hunger and much thirst
[c. 1275(?1200) Lay. Brut, from Łęcki (2010: 171)]
In Middle English, the plural -e ending on the participle could even be used in
relation to singular nouns, which is taken to indicate that the grammaticalisa-
tion process had more or less reached completion by that stage (Zimmermann
1968: 36, Fischer 1992: 256; cf. example (9) above). More usually, however, in-
flection is generally lost, as noted in the preceding section. On the other hand,
the perfect often alternated in its functions/meanings with the preterite,
sometimes in different manuscripts of the same text, suggesting that it was not
yet fully grammaticalised and ‘regulated’.9 In the mediaeval period, the perfect
was still compatible with specific past time reference, as will be stressed and
illustrated in Chapter 5, Section 5.2.2. Conversely, Chaucer used a simple past,
rather than a perfect, with adverbs like never: A fairer saugh I nevere ‘I have
never seen anyone more beautiful’ (Benson 1987a: xli). Essentially both the
perfect and the preterite could indicate a past event connected to the moment
of speech. Therefore, Middle English is indeed a transition period in which the
perfect has started to come into its own more fully than in Old English but has
not yet attained the state of distinctness from the preterite and idiomatic use
it has in Modern English.
The ambiguity between a periphrastic verb phrase and a complex transitive
clause with main verb have plus a participial object complement persisted into
Middle English, and in fact the two constructions survive into the modern peri-
od, now distinguished by word order (Fischer 1992: 257).10 This is illustrated in
9 See Denison’s stage b) in Section 3.1.2 above and Denison (1993: 352–353), as well as
Mustanoja (1960: 498–499, 504, 506–507), Visser (1963–1973: §766, §772, §800, §§805–
811, §2004, §2008), Zimmermann (1968: 36), Mitchell (1985: §723, §726 fn. 173), Fischer
(1992: 250, 256–258), Elsness (1997: 247ff., 291), Tagliamonte (2000), Fischer and van der
Wurff (2006: 139), Chankova (2008: 119–124, 128, 157–158, 169, 200–203); for the preterite
and the perfect in Shakespeare’s English, consult Kakietek (1976: 50–51).
10 See Visser (1963–1973: §1964, §1979, §2001, §§2116–2122), Brinton (1988: 102, 1994),
Kovatcheva (1989: 135–137, 139), Denison (1993: 348, 358, 365). The survival of both con-
structions and our recurrent themes of ambiguity and indeterminacy resonate with the
constructionist idea of synchronic gradience resulting from constructional change (see
Traugott and Trousdale 2010, 2013, inter alia). Such indeterminacy in this book involves
the analysis of OE perfects and passives as complex transitive or copular constructions as
The Story of the English Perfect 45
There are modern instances where word order cannot help to distinguish
between these two constructions and we have to rely on contextual cues. In
other words, the duality of have + past participle still persists and might even
result in double-marking, as seen in (13), with two tokens of have and two past
participles.
(13)
In one year the Abbey Theatre has had submitted 600 plays by Irish
authors. [1960, Daily Telegraph, quoted in Visser (1963–1973: §2120)]
Instead of coming after the object at the end of the clause, submitted appears
medially with the other two verbs – the perfect auxiliary has is thus followed
by two past participles, the first one part of the perfect construction, the sec-
ond more like an object complement. Essentially, have + past participle occurs
twice. I return to similar cases of double-marking below.
Have was by no means the only perfect auxiliary in Old, Middle and Early
Modern English. Be could build perfect periphrases too, mainly, though not
necessarily, with intransitive verbs denoting a change of place or state.11 Both
auxiliaries are found from Old English on, when have occurred mostly with
transitive verbs and be was restricted to intransitives, as in Modern Dutch or
German, although some intransitive verbs already combined with have. The
original meaning of the be-construction is widely believed to have been that of
a stative use of linking/copular be followed by a predicative adjective in order
to refer to a state which resulted from an earlier action (Denison 1993: 366;
cf. Visser 1963–1973: §1898, Rissanen 1999: 213). This is shown in the following
simple OE example:
[it] is unlikely that such a clear dichotomy exists in view of the facts that
the two constructions often seem to be interchangeable and that in the
later development have ousts the be forms. Such a dichotomy would also
mean that a structure like he is come always referred to state, never to
completed action.
Occurrence with OE past participles like that of come arguably indicates some
degree of grammaticalisation, since they cannot be so naturally interpreted
as denoting resultant states which persist at the time of speaking: cf. ModE
a fallen tree and *a come man (Los 2015: 75), though both phrases would be
equally acceptable in Bulgarian (see the Epilogue to Chapter 5 too). In addi-
tion, Traugott (1992: 193) notes that the participle occurring with be is always
semantically active, because the verb is intransitive.12
originally separate verbs beon and wesan, in fact combining three Indo-European roots,
with traces of a fourth one – see Prokosch (2009[1939]: 219–222), Mitchell (1985: §§651–
664), Laing (2010: 238), Łęcki (2010: 8, 212 n. 8).
12 It will be demonstrated below that Bulgarian and Macedonian possess morphological
distinctions between active and passive past participles. This, however, has not prevented
the formation of a new have-perfect in addition to the old one which relies exclusively
on be.
The Story of the English Perfect 47
(15) Tens of thousands of phantom azalea bushes and geraniums are believed
to be alive and growing in the gardens of Northern Ireland. [news, from
Biber et al. (2002: 331)]
48 chapter 3
And yet, when applied in combination, the tests above do suggest that the
grammaticalisation of the be-perfect had not reached completion in OE.
Have started to drive be out of its domain as early as Middle English or even
earlier, in spite of occasional instances of be where have would be expected, as
in (16), where be is used in a transitive clause.13
(16) for þou art passed thy paynes alle [Tundale, 1464, quoted in Mustanoja
(1960: 501)]
Cases like (16) are rather rare and exceptional (though they survive until late
and are to be revisited below – see Chapter 5 and Rydén and Brorström 1987:
133ff.). Traugott (1992: 191) notes that throughout the Old English period al-
ready, and of course much more so later, habban could occur with intransi-
tive participles in contexts which had nothing to do with possession (cf. Łęcki
2010: 153–154, 157–159, and the Dictionary of Old English, at http://tapor.library.
utoronto.ca/doe, accessed on 27 Feb 2017). Quite a few verbs were happy with
either auxiliary, so we observe the typical variation that is to result in language
change. The following are early examples of have used with intransitive verbs
of motion:
13 See Jespersen (1931: 30), Mustanoja (1960: 500–501), Visser (1963–1973: §§2002–2003),
Strang (1970: 207), Zimmermann (1973: 109ff., for be with transitive verbs), Mitchell
(1985: §722ff.), Rydén and Brorström (1987: 16–17, 195ff., 197 fn. 11, 200), Rydén (1991: 346,
esp. Fig. 2), the last two showing a typical S-curve of the progression of have between
1500 and 1900; consult also Fischer (1992: 260), Denison (1993: 344, 363–364, 1998: 135),
Kytö (1994: 179, 1997: 17–18), Elsness (1997: 246, 271–272, 325–327, 340), Rissanen (1999:
213), Fischer and van der Wurff (2006: 142), Chankova (2008: 136–137).
The Story of the English Perfect 49
‘The devil was very near who had travelled in the dangerous journey over
a long way.’ [Genesis B 688–90, quoted and translated in Lee (2003: 390),
also quoted in Mitchell (1985: 289), emphasis mine]
By the sixteenth century, only have was allowed with transitive verbs and was
preferred with non-mutative intransitives, with variation in the case of mu-
tative ones (Rissanen 1999: 213; cf. Strang 1970: 149, Kakietek 1976, Rydén and
Brorstöm 1987: 17–18, Chankova 2008: 181–186, 205, Brinton and Arnovick 2011:
371–373). Curiously, Nakamura (1987: 43) notes that be-perfects are statistically
still decidedly preferred with mutative verbs in Pepys’s Diary from the seven-
teenth century.14
It was mainly in the eighteenth century that have gained a lot of ground,
although be was still more common with intransitive verbs (Rissanen 1999: 215;
cf. Rydén and Brorström 1987: esp. 195ff., Shannon 1989, Rydén 1991: 346–347,
Kytö 1997: 32ff.). Interestingly, prescriptive grammarians in the 18th century
picked up on the variation that existed back then. Some of them actually pro-
scribe the use of be with certain mutative intransitive verbs, as in (19), which
Lowth denounces as ‘improper’ (see Rydén and Brorström 1987: 206–209 and
Straaijer 2010: 65–66).
(19) The rules of our holy Religion, from which we are infinitely swerved [Robert
Lowth. 1762. A Short Introduction to English Grammar. p. 63, quoted in
Straaijer (2010: 66), emphasis mine]
Presumably, Lowth would have preferred have here. These views make Lowth,
and other eighteenth- and early-nineteenth-century grammarians who ex-
pressed similar opinions, innovators in terms of this change ongoing at the
time, though more research is needed to verify if they were also innovators
in practice, i.e. if they really practised what they preached. Faithful to his
more descriptive (rather than prescriptive) inclinations, in the 1768 edition
of his Rudiments of English Grammar, Joseph Priestley (1768: 127–128, cited in
Straaijer 2010: 66–67) comments on the choice of auxiliary, suggesting that I
am fallen refers more to the present (state?), whereas with I have fallen ref-
erence is also made to the past (action?) (cf. Rydén and Brorström 1987: 208
for similar contemporary accounts). According to Straaijer (2010), Priestley’s
private correspondence demonstrates that in his actual linguistic output he
adhered to the principles of differentiation which he postulated in his work.
14 A mutative verb is one which indicates a transition from one place, location, state or
condition to another, e.g. go, come, return, become, grow, fade, disappear, etc. (Mustanoja
1960: 500 fn. 2, Rydén and Brorström 1987: 22).
50 chapter 3
Furthermore, his usage to a great extent chimes in with the more general usage
of the time.
The nineteenth century saw the most rapid switchover to have in all remain-
ing contexts, especially in informal language. By the end of the nineteenth cen-
tury, have had won out as the sole marker of the perfect, despite some fossilised
leftovers such as She is gone; those are arguably no longer taken to be perfects
at all, but are once again to be thought of as copular clauses with predicative
complements, devoid of actional meaning, as suggested by so-called ‘double
perfects’ (to be discussed below).15 (20) and (21) are two late examples, from
the 19th and the 20th centuries. Both of them must be interpreted as perfect,
not passive, since the verbs are intransitive (see Denison 1998: 137).
(20) … the extremity at which we are arrived [1864?, The Federalist Paper,
COHA]
(21) The warm sea wind was risen and blew over them now. [1963, Iris Murdoch,
The Unicorn, quoted in Visser (1963–1973: §1900), emphasis added]
15 See Jespersen (1931: 31, 34), Mincoff (1958: 131–132, 188), Visser (1963–1973: §2052, §§2061–
2062), Strang (1970: 100), Mitchell (1985: §740), Rydén and Brorström (1987: esp. 116, 195ff.,
210–213), Rydén (1991: 346–347, 350, 353, n. 8, n. 11), Denison (1993: 344, n. 6, 359, 363, 367–
368, 1998: 135–136), Kytö (1994: 180, 1997: 19, 30), McFadden and Alexiadou (2006: 270),
Mugglestone (2012: 352–353); also see Harris (1984) for modern non-standard varieties.
The Story of the English Perfect 51
travel and walk, unlike change-of-state lexemes such as change and recover,
which prove more resistant to have. Also of importance are:
– the tense: present vs. past perfect, the latter reportedly favouring have
(Shannon 1995: 150, though this is disputed by McFadden and Alexiadou
2010: 396–39),
– finiteness: with perfect infinitives and (less certainly) present participles
favouring have (Mustanoja 1960: 502, Shannon 1995: 148, 150; cf. McFadden
and Alexiadou 2010: 398–399 for caveats),
– modality,
– complementation patterns,
– etymology: apparently, borrowed main verbs contributed to the rise of
have, etc.
Extra-linguistic and social factors include time, region, type of text, level of
formality, and gender. As is to be expected, there is variation depending on
the individual author’s idiolect (regarded cognitively as well as socially).
Idiolectal peculiarities stand out clearly in the material presented in Rydén and
Brorström (1987: 201–206) and Rydén (1991: 351). For instance, Dickens appears
to be innovative concerning this ongoing change, while Jane Austen is more
conservative, as are other prominent female Victorian writers, like Charlotte
Brontë and George Eliot (cf. Cortes 2019). The various factors (e.g. age, gender
and social status) interact in complex ways. This can also be seen in Straaijer’s
(2010) study of the idiolect of the 18th-century grammarian Joseph Priestley.
Kytö’s (1994: 184–186) is a particularly valuable and detailed survey of the
distribution of be and have with intransitives in Early Modern English, featur-
ing discussion of relative frequencies and linguistic and extra-linguistic factors
at work in determining the choice of auxiliary. The Bible favours the traditional
use of be with intransitives, which is hardly surprising, but strangely enough,
private letters, comedy and fiction from the Early Modern period do not favour
the use of the rising form have either. In addition, the verbs come and go show a
clear preference for the construction with be more than other verbs, as already
noted in the preceding paragraphs. For the same period, the past perfect and
the perfect infinitive, both of which highlight the perfective aspect of ‘action’,
were among the constructions that paved the way for the rise of have with
intransitive verbs (cf. Straaijer 2010; Rydén and Brorström 1987: 189ff. cover
the 18th and 19th centuries, when this still obtains to a large extent, though
with some fluctuations due to complex interactions with the other factors).
In the language of Shakespeare, be-perfects are incompatible with adverbials
of duration and with modals (Kakietek 1976: 47; cf. Łęcki 2010: 159 for similar
observations concerning Old English). In Kakietek’s (1976: 49–53) corpus of
Shakespeare’s work, be is also more restricted than have in terms of the range
52 chapter 3
3.4.1 Frequency
Apart from the factors discussed in the previous section, have-perfects were
normally preferred to be-perfects in the following contexts during the ME and
EModE periods, while they were still in competition: the presence of durative
meaning, (22), recurrent/repeated actions, (23), statements containing adver-
bial expressions, (24), and hypothetical statements, (25), as well as negated or
questioned propositions.16
(22) and while I had a while goon [14th c., The Romaunt of the Rose, 135, from
Mustanoja (1960: 502), emphasis mine]
(23) a. þis þretty winter … hath he gone and preched [14th c., Piers Plowman
B xviii 293, from Mustanoja (1960: 502), emphasis mine]
16 Mustanoja (1960: 502), Shannon (1989: 677, 1995: 147–150), Huber (2019); cf. Fridén (1957),
Johannisson (1958), Rydén and Brorström (1987: 27–29, 184ff.), Rydén (1991: 350), Fischer
(1992: 260), Fischer and van der Wurff (2006: 141–142), and McFadden and Alexiadou
(2010), who express some doubts about negation. See Łęcki (2010: 159–162) for similar
examples and observations concerning tendencies in Old English, including adverbials
(or perhaps more accurately, locative complements), (i), and conditional/hypothetical
contexts, (ii).
(i) þa Scipia hæfde gefaren to ðære niwan byrig Cartaina …
when Scipio had gone to the new city Carthage
‘When Scipio had gone to the new city of Carthage …’ [ca. 880, from Łęcki (2010: 160)]
(ii) hæfde ða forsiðod sunu Ecgþeowes …
had then died [lit. for-journeyed] son Ecgtheow’s
‘Ecgtheow’s son would then have perished …’ [Beowulf, from Łęcki (2010: 161)]
Similar preferences regarding the usage of Samuel Pepys in the 17th century are noted by
Nakamura (1987: 43). Rydén and Brorström’s (1987) data covers the 18th and the 19th cen-
turies. Denoting hypothetical events indicates further grammaticalisation that has gone
beyond the encoding of temporal and aspectual relations.
The Story of the English Perfect 53
(24) ye han entred into myn hous by violence [14th c., Chaucer, Canterbury
Tales, B The Prologue and Tale of Melibee 3001, from Mustanoja (1960:
502), emphasis mine]
(25) a. I wolde […] that I, Aurelius, Hadde went ther nevere I sholde have come
agayn [14th c., Chaucer, Canterbury Tales, F The Franklin’s Prologue and
Tale 971, from Mustanoja (1960: 502), emphasis mine]
b. I should not have been sorry if you had entered a little more into
Peninsular politics [1837, Wordsworth, from Rydén and Brorström
(1987: 186), Rydén (1991: 350)]17
Furthermore, when the mutative verb emphasised the action, this favoured
have, as opposed to emphasis on states, which favoured be (Fridén 1957,
Johannisson 1958, Mustanoja 1960: 502, Mitchell 1985: §722, §729, Rydén and
Brorstöm 1987: 17, 26, Shannon 1989: 677, 1995: 148, Rydén 1991: 350, Fischer
1992: 260, Rissanen 1999: 213; cf. Gräf 1888: 81ff.). All of this made the have-
perfect more frequent than the one with be, and since frequency is crucial
in grammaticalisation and entrenchment, this may be why have ultimately
pushed out be (Fischer and van der Wurff 2006: 142; cf. Bybee and Thompson
1997, Bybee 2003, Łęcki 2010: 24ff., and the discussion in Chapter 2). More
globally, since non-mutative verbs outnumbered mutative ones, making have
the more frequent marker of the perfect from the outset, it was poised to take
over anyway (see Traugott 1972: 145, cited in Denison 1998: 136; cf. Chankova
2008: 184).18
The token frequency of an item is how many times it occurs in a given
corpus – for our purposes, the occurrences of be or have as perfect auxiliaries
in a selection of running text. While token frequency was certainly important,
type frequency must have played a part as well. Smith (2001) notes that even in
the earliest OE, the displacement was already underway and it was influenced
by the fact that perfect have combined with many more types of verbal lexeme
than its underdog rival.
17 With regard to this last (and also rather late) example, Rydén (1991: 350, quoting Rydén
and Brorström 1987: 28, 186) notes that have is resorted to in order to ‘prevent potential
ambiguity in time-reference’, lamented as ‘disastrous’ by earlier researchers.
18 Frequency as a functional explanation of syntactic universals is discussed in Haspelmath
(2008a: Section 4, 2008b); Haspelmath (1999: Section 5) also considers the role of frequen-
cy in functional explanations more generally. See Bybee (2003, with references) on the
entrenchment of schemas with high token frequency.
54 chapter 3
Unlike the use of be as the auxiliary of the progressive, its use as a tense
[i.e. perfect] auxiliary […] is a receding feature. Although have had been
gaining ground at the expense of be with mutative intransitives since
the late Middle English period, the decisive steps in the process where
have superseded be were taken in the 18th and 19th centuries. […] One
possible conclusion to be drawn is that considering the preponderance
throughout the periods studied (1) of the copular uses of be and (2) of the
passive auxiliary uses of be compared to all the remaining auxiliary uses,
it becomes evident that the use of be as a tense auxiliary is marginal and
one which thus was perhaps more liable to be superseded by the have
perfect than not.
This explanation, however, does not tell us why be did not recede as a progres-
sive auxiliary, since the numbers of the progressive were initially not that great
either. Perhaps it did not because there was no other alternative. I return to this
and similar issues in Chapter 9.
19 See Jespersen (1909–1949: IV 41), Mustanoja (1960: 501), Traugott (1972: 145, cited in
Kakietek 1976: 47–48), Zimmermann (1973), Rydén and Brorström (1987: 17, 197),
Kovatcheva (1989: 111ff.), Shannon (1989: 677), Rydén (1991: 347), Fischer (1992: 261–262),
Denison (1993: 366, 1998: 136, 183–184), Kytö (1994: 182, 1997: 18, 28), Elsness (1997: 246),
Kilpiö (1997), Rissanen (1999: 215), Fischer and van der Wurff (2006: 142), Chankova (2008:
137–138), Łęcki (2010: 163–164), Brinton and Arnovick (2011: 372). The references range
The Story of the English Perfect 55
followed by the past participle of a verb that had both transitive and intran-
sitive uses could be construed as perfect, as well as passive, or a resultative
stative, as in (26) and (27).20
(26) Our hopes are again revived of seeing the Viceroy of Mexico. [1797, Nelson,
Letters, ed. Naish (1958) 190 p. 328 (30 Jun.), cited in Rydén and Brorström
(1987: 24), Denison (1998: 184)]
(27) … this house is turn’d bawdy-house, they pick pockets [Shakespeare, Henry
IV, Part I, III 3.69, quoted in Kakietek (1976: 46), emphasis mine]
from introductory textbooks to more serious scholarly publications, spanning more than
a century, which shows how widespread and deeply entrenched this view is. Kovatcheva
(1982: 14), among others, proposes an analogous explanation for the shift from impersonal
to personal constructions.
20 Again, see Jespersen (1909–1949: IV 41), Mincoff (1958: 187–188), Kakietek (1976: 46–47),
Mitchell (1985: §735), Rydén and Brorström (1987: 24, 100 fn. 11, 147), Fischer (1992: 261),
Denison (1993: 345–346, 1998: 183–184), Kytö (1994: 181), Elsness (1997: 243, 339), Rissanen
(1999: 215), Fischer and van der Wurff (2006: 142), Straaijer (2010: 75), Brinton and Arnovick
(2011: 372). It remains unclear, however, why English seems to ‘have no objections’ to the
verb be still serving as the auxiliary of the progressive and the passive – probably as a
result of the lack of ambiguity in the form of the following non-finite verb? The availabil-
ity of separate non-finite participial shapes in Bulgarian/Macedonian probably likewise
enabled be to function unproblematically as perfect and passive auxiliary. On the use of
the Germanic so-called ‘past (or second) participle’ in the passive and the perfect, see
Anderson (1997).
56 chapter 3
The relevant passage from Rissanen (1999: 215) is pretty much in the same spir-
it, where Rissanen is confident that the reasons are rather self-evident:
The reasons for the loss of be are fairly easy to find. The functional load
of be was heavy as this verb was not only used as the copula but also
in the be + -ing structure and in the passive. It was particularly the last-
mentioned function that easily caused ambiguity in expressions such as
was grown, was developed, etc.
Denison (1998: 136) adds that functional load and ambiguity may look a lit-
tle post hoc as explanatory factors, but he nevertheless finds them ‘plausible
enough’. The following examples are also ambiguous in the same way. Although
Visser (1963–1973: §2073) treats them as perfects, they could be passives too.
(28) how thou art altered with thy travel! [1598–9, Ben Jonson, The Case Is
Altered, quoted in Visser (1963–1973: §2073), emphasis added]
(29) The whole aspect of the room was now altered. [1911, A. Bennett, Hilda
Lessways, quoted in Visser (1963–1973: §2073), emphasis added]
Kakietek (1976: 47) notes in passing that this type of ambiguity was a rare phe-
nomenon in Old English; according to him, this was due to the fact that in
Old English most transitive and intransitive verbs had distinct shapes, and in
addition, resultative stative past participles were arguably adjectival and often
inflected (as already pointed out in earlier sections).
Confusion of analysis might be caused even by verbs which normally belong
to the class of mutative intransitives. Quirk et al. (1985: 170 n. [a]) actually refer
to cases like her friend was gone as pseudo-passives, as did many eighteenth-
century grammarians (see Rydén and Brorström 1987: 208, Denison 1998: 136,
The Story of the English Perfect 57
Rissanen 1999: 215, Straaijer 2010: 65–66). In addition, as Mustanoja (1960: 500)
notes, an important difference between OE wesan/beon ‘be’ and OE habban
‘have’ is that the former never fully lost its capacity to express states, as op-
posed to actions; this was mentioned above and can be seen clearly in (30),
where ambiguity between a passive and a copular reading persists to this day.21
(30) we could not go in because the door was closed. [Mustanoja (1960: 500),
emphasis added]
(31) a. he has been come over about ten days [Swift, Journal to Stella II 625,
from Rissanen (1999: 215)]
b. Cher Frere has been gone since four o’clock this morning to a private con-
ference. [1788, Betsy Sheridan, Journal 42 p. 131 (21 Nov.), from Denison
(1998: 137)]
c. Richard had been gone away some time, when a visitor came to pass a
few days with us. [1853, Charles Dickens, Bleak House, Chapter XXX]
21 Compare Denison (1993: 361) and Chankova (2008: 87) for counterarguments. The ambi-
guities between passive (actional/dynamic) and copulative (statal/stative) be are aptly
illustrated in Mincoff (1958: 187–189) and Quirk et al. (1985: 167–170).
22 Further discussion can be found in Jespersen (1931: 36), Visser (1963–1973: §2162), Mitchell
(1985: §735), Rydén and Brorström (1987: 25), Denison (1993: 363, 1998: 137–138), Rissanen
(1999: 215), Łęcki (2010: 164), McFadden and Alexiadou (2010: 415). Ammann (2005) deals
with double perfects in German. Also see Visser (1963–1973: §2023) and Bowie et al. (2013:
Section 3.2.3) for other double perfects with have:
(i) I have done told you. [from Visser (1963–1973: §2023)]
58 chapter 3
Only two pages after the example in (31c), the chapter’s opening sentence,
Dickens gives an indication that this construction was most probably adjecti-
val for him:
(32) However, as he is now gone so far away, and for an indefinite time, and as he
will have good opportunities and introductions, we may consider this past
and gone. [1853, Charles Dickens, Bleak House, Chapter XXX]
(33) Not to be always selfish, talking of my son, who has gone to seek his
fortune, and to find a wife… [1853, Charles Dickens, Bleak House,
Chapter XXX]
(34) also confirms that be gone was a copular construction for Dickens. He uses
it in a temporal subordinate clause introduced by while to indicate a stretch
of time, where the dynamic perfect would not be allowed (cf. *Take care of
Pa while I have gone). The meaning is clearly ‘while I am away’, as in (31)–(32)
above.
(34) Take a little care of Pa while I am gone, mama! [1853, Charles Dickens,
Bleak House, Chapter XXX]
respectively.23 It has been proposed (and also argued against) that, when
combined with the past participle of a transitive verb to form a passive(-like)
construction, beon and wesan denote a resulting state, whereas weorðan de-
notes a process/action.24 There are also attestations of wearð geworden ‘had
happened’, testifying that this verb can occur as an auxiliary and a lexical verb
in the same clause.
Unlike what happened in other Germanic languages, OE weorðan ‘become’
for some mysterious reason disappeared as an alternative passive auxiliary,
leaving be as the only marker of the passive. According to some scholars, the
disappearance of passive weorðan thus created a ‘conspiracy’ to make be even
more unpopular in the context of the perfect, again due to excessive functional
load.25 Rissanen (1999: 215) notes that German, which has no actional passives
with sein ‘be’, retains the sein/haben ‘be/have’ distinction in the perfect, unlike
standard Swedish and English, both of which rely on be-passives (cf. Fischer
and van der Wurff 2006: 140, and Chapter 6 herein).
The Middle English reflex wurthen could have played a more significant
role, just as it does in English’s West Germanic relatives German and Dutch.
However, even in the earliest attested Old English, weorðan was arguably some-
what infrequent in the passive, becoming even more so in Early Middle English,
when the most significant shifts in the newly emerging auxiliary system were
taking place. It remains unclear why exactly ME wurthen disappeared: poten-
tial reasons have been sought in the nature of the verb itself or perhaps there
might have been some foreign influence – Scandinavian or French (see Fischer
1992: 262; cf. Mincoff 1958: 188, Mustanoja 1960: 616–619, Mitchell 1985: §801,
McWhorter 2002, Petré 2010, 2013, 2014).
Kovatcheva (1989: 109–114) puts forward three scenarios for the demise of
OE weorðan/ME wurthen: the first is that OE beon acquired a dynamic inter-
pretation (which, however, it always had, as established above!); alternatively,
the whole thing was arbitrary (the second scenario); the final scenario, which
is the view Kovatcheva seems to subscribe to, is that, unlike combinations with
23 On weorðan as an alternative to the be-perfect from OE to the early fourteenth century, as
well as on ambiguous perfects/passives with OE beon and OE weorðan, see Visser (1963–
1973: §1897), Mitchell (1985: §671, §734ff.). Denison (1993: 344–346) points out that perfect
weorðan ‘is fairly uncommon in Old English and rare in Middle English’, making its status
as perfect auxiliary uncertain (cf. Łęcki 2010: 227 n. 15).
24 See Mitchell (1985: §§673–674, §§786–787, §§789–801), who doubts the universal validity
of this distinction (cf. the discussion of Modern German and of get-passives below): there
is evidence that OE beon/wesan + past participle could denote an action or process, in
addition to denoting states (more details will be given in the discussion below); likewise,
weorðan could express both states and actions.
25 See Mustanoja (1960: 437–440, 616–619), Strang (1970: 351), Zimmermann (1973), Fischer
(1992: 261–262), Chankova (2008: 185).
60 chapter 3
OE beon, those with OE weorðan were not being grammaticalised – they were
just free syntactic combinations (but that ignores the collocation of wearð
geworden). Arbitrariness might be best squared with the facts.
(36) & man cydde Harolde cyng hu hit wæs þær gedon
and one informed Harold king how it was there done
& geworden
and ?happened/?made
‘and King Harold was informed how things had gone there’ [ChronE
(1066), from Denison (1993: 345–346), emphasis mine]
Moreover, Fischer (1992: 262) suspects that the loss of perfect be was perhaps
additionally facilitated by the fact that ‘in co-ordinate constructions the aux-
iliary was usually not repeated so that have was often used where be was ex-
pected’. However, it is not entirely clear why this should not work vice versa, in
favour of be. This is probably due to the far greater frequency of have (and of
transitive verbs), but the point needs to be studied and quantified in more de-
tail. Be that as it may, coordination is one more ambiguous context in which re-
interpretation might sneak in – a construction intended by the speaker/writer
26 On the properties of the ‘past’ participle in English, perfect as well as passive, see Anderson
(1997: esp. 8), who points to its adjectival origins and relates it to the development of
the perfect periphrasis (cf. Mitchell 1985: §§23–24, §728, §734, §777). As remarked above,
in earlier English (proto-)perfects the past participles of intransitive verbs are active in
terms of meaning, while those of transitive verbs are passive (cf. Sections 3.1.2, 3.2).
27 More constructions of this kind are discussed in Mitchell (1985: §735). Compare a similar
flexibility of progressive and passive (stative?) be, (i), or progressive and perfect be, (ii),
in Victorian English, as well as of the contraction ’s in PDE, (iii). Also see the Epilogue to
Chapter 5.
(i) … lest the object of his distrust should be looking over, or hidden on the other side.
[1853, Charles Dickens, Bleak House, Chapter XLVI]
(ii) … people are going or gone to bed [1853, Charles Dickens, Bleak House, Chapter LVI]
(iii) to make sense || of all that’s changed and changing [poem by Tom Phillips, http://
msvstp.blogspot.bg/2016/12/trees-in-portishead.html, accessed on 30 Dec 2016]
The Story of the English Perfect 61
(37) Anyhow, she’s gone, walked out, slung her hook. [1955, L. P. Hartley,
A Perfect Woman, cited in Visser (1963–1973: §1902, p. 2072), emphasis
mine]
More radically, ambiguities may arise in the so-called clipped perfect. What
defines the clipped perfect is its missing auxiliary, as illustrated in (38). An am-
biguous context is presented in (39), where He come may be expanded either
as ‘is he come’ or as ‘has he come’ (Visser 1963–1973: §2054), unless come is an
infinitive/base form.
(39) ‘Tis miraculous that you tell me, sir. He come to woo our lade mistress for
his wife! [1612, Chapman, The Widow’s Tears (ed. Smeak) I, i, quoted in
Visser (1963–1973: §2054)]
Things are not as straightforward as the foregoing account might suggest. There
are problems of four kinds: firstly, there are counterexamples to the assump-
tion that agreement marking on OE past participles is incompatible with a
perfect reading, as noted in earlier sections; secondly, instead of being ‘neatly’
restricted to the perfect, have seems to have dramatically increased its func-
tional load over time, fostering a greater diversity of constructions it can par-
ticipate in (including a have-passive); thirdly, alternative passive and perfect
28 See Jespersen (1931: 30, 41), Mincoff (1958: 131), Visser (1963–1973: §1898), Rydén and
Brorström (1987: 197), Shannon (1989: 677), Fischer (1992: 262), Denison (1993: 366, 1998:
136), Kytö (1994: 181, 1997: 18, 30), Fischer and van der Wurff (2006: 142), Brinton and
Arnovick (2011: 372).
62 chapter 3
As Trask and Millar (2007: 177) observe, since the food is all gone, it cannot
be available or in someone’s possession (see Mitchell 1985: §726, who pro-
vides additional examples). This example, therefore, cannot receive a stative
29 An anonymous LAGB reviewer objects that the fourth of these points “does not appear
to problematise the traditional functional load account. This is because the frequency of
the passive (regardless of how it is constructed) presumably depends on speakers’ com-
municative needs; why should they all of a sudden wish to express the passive function
more often when be recedes from the perfect paradigm?”
30 Kovatcheva (1982) raises similar concerns regarding the transition from impersonal to
personal constructions.
The Story of the English Perfect 63
(41) If they had any parte of their liberties withdrawne [1568, Grafton, Chron. II
141, OED s.v. have 18, cited in Rissanen (1999: 216), emphasis added]
(42) Another had one of his hands … burnt. [Defoe, Robinson Crusoe II 10, OED
s.v. have 18, cited in Rissanen (1999: 216), emphasis added]
This passive-like construction, of the type He had a book given (to) him, is
paraphrasable with established passives, e.g. ‘A book was given (to) him’ or
‘He was given a book’. It can be traced back to the middle of the fourteenth
century, or arguably even earlier. Initially relatively infrequent, it became fully
established by the beginning of the Modern English period, and might now
sometimes be preferred to the regular passive, He was given a book (Visser
1963–1973: §1964, §1979, §§1980–1984, §§2116–2122). As mentioned above, the
earliest example in Visser dates back to the 14th c.:
(43) Bot of Oðer wommen tua [he] Had four suns geten him. [13.. Cursor M.,
3900, quoted in Visser (1963–1973: §1979), emphasis added]
64 chapter 3
Łęcki (2010: 196) gives the following as the earliest attested have-passive, on
the cusp between Old and Middle English.
The role of the subject in this type of construction with have is more active
than in the be-passive, and this structure often has a causative meaning too,
as in (44) (see next section). The emergence of the so-called have-passive can
be associated with the subjectivisation of the recipient argument. The noun
phrase which denotes the person in the clause gets topicalised in either type of
construction: the passive with be, He was given a book, or the rival with have, He
had a book given to him. The rise of the have-passive has likewise been linked to
dispelling ambiguity. As Rissanen (1999: 216) observes:
Moessner points out that in [He had a book given to him] there is no risk of
even momentary ambiguity as to the semantic role of the subject; theo-
retically speaking, he in [He was given a book] could be analysed either as
the direct or the indirect object of the corresponding active clause until
the post-verbal elements are heard or seen.
However, why should the have-passive come into being and gain in frequency
when it adds to what is essentially a greater functional load and potential for
ambiguity in the roles of have? As it were, it goes back to the original mean-
ing of the ancestor of the perfect – the complex transitive clause indicating
possession. It could therefore be thought to have branched off as another de-
scendant of the original prototype (see Chapter 2, Section 2.2). Even if the two
The Story of the English Perfect 65
structures have coexisted all along, what we see is hardly a reduction in the
range of functions; quite the contrary, the duality of perfect- and passive-like
uses is still with us, including intermediate cases. This ties in well with the
points made in Chapter 2 about layering and coexistence of input and output
constructions after they undergo grammaticalisation.
There can even be double-marking, as in (46), which shows the perfect and
the quasi-passive with have. The string of three verbs means that the first two
must be auxiliaries. Since it represents a new ‘layer’ of auxiliary, the second
have is still more ‘lexical’ than the first one in terms of NICE properties, in-
cluding the ability of auxiliaries to host negation and invert with the subject,
among others (more on this is given below). As the context indicates, the mem-
bers of staff do not have their passes any more, so a meaning of possession is
ruled out. There can be no causative meaning either, as they certainly did not
want to have their passes revoked (see below for the causative construction).
Similarly, the referents of the subject in (47) sadly do not have legs any more,
so the have-passive can surely exhibit semantic bleaching.
(46) Several members of staff have had passes revoked. [Sky News, 25 March
2016]
Such examples are not at all rare and can easily be multiplied (e.g. … had prop-
erty confiscated – from a History Channel documentary, etc.).
31 As Łęcki (2010: 189) correctly notes, “causative HAVE in English is only grammaticalised
to some extent in that it could be treated as an idiomatic use of HAVE rather than as an
auxiliary since it virtually does not exhibit any NICE properties […], nor are there any
signs of erosion or fusion …”
Grammaticalisation is only incipient here, involving the pragmatic-semantic dimension,
with no concomitant morpho-syntactic and phonological changes (cf. Krug 2001).
66 chapter 3
that the construction is still pretty much alive and well and shows no sign of
fading away.
(48) He … would have had me dined [edn: dine] with him [1711, Swift, Jnl. to
Stella 230. 27 (3 Apr), cited in Denison (1993: 342), emphasis added]
(49) a. He hadde þare tweie castles bi-walled swiðe faste. [c. 1205, Layamon,
quoted in Visser (1963–1973: §2118), emphasis added]
(50) one went there to have his teeth pulled, his wagon fixed, his heart listened
to … [H. Lee, To Kill a Mockingbird, quoted in Visser (1963–1973: §2118)]
32 Exemplifying each type of construction, Łęcki (2010: 196) comes up with the following
sentences (though the first three are very close in meaning and are in practice often dif-
ficult to tease apart; see also Mitchell 1985: §704, and Johannsen 2016):
(i) I had my testicles removed when I was a child. [experiential; alternatively treated as
passive here]
(ii) I had my testicles removed by the doctor. [passive]
(iii) I had my testicles removed to spite my wife. [causative]
(iv) I had the doctor (?to) remove my testicles. [causative]
The Story of the English Perfect 67
(53) You have (got) to come in now. [Huddleston and Pullum et al. (2003: 205)]
33 The causal implication of Latin fecit might also be rendered with hæfþ in OE (among
other constructions), suggesting possible causative overtones for OE habban from very
early on, as in (i) (see Mitchell 1985: §726 fn. 173). Alternatively, hæfþ in (i) is simply a
marker of the (proto-)perfect and the causality is inherent in the semantics of heal, ‘to
make whole/healthy’.
(i) fides tua te saluum fecit [Latin, Mark 10:52, from
faith thine thee healthy has.made Mitchell (1985: 293)]
þin agen geleafa þe hæfþ gehæledne [OE, Blickling Homily, from
thine own faith thee has healed Mitchell (1985: 293)]
‘thy faith hath made thee whole’ [Mark 10:52, King James Bible]
34 This is similar to the story of ought to, which derives from the OE verb agan ‘to own’ (cf.
Kovatcheva 1989: 143, Łęcki 2010: 132, and Section 3.1.3 above).
68 chapter 3
(54) This has (got) to be the worst restaurant in town. [Huddleston and Pullum
et al. (2002: 205)]
The auxiliation of have to is traced by Łęcki (2010: Chapter 3) and Fischer (2015)
(cf. Mitchell 1985: §§950–953, and the Dictionary of Old English http://tapor.
library.utoronto.ca/doe/, accessed on 27 Feb 2017). According to a traditional
grammaticalisation account (disputed by Fischer), the development went
through stages reminiscent of the formation of the perfect: from I [haveV [a
book to recommend]NP:Object], with a full transitive have and an infinitival ad-
junct/post-modifier which depends on the object noun, to I [[have to recom-
mend]VP [a book]NP:Object], with the newly established unit (see Łęcki 2010: 80,
86–91, Fischer 2015: 121).35 In the course of late Old English and Middle English,
the possessive meaning is believed to have slowly bleached and given way to
a deontic colouring of obligation in conjunction with the to-infinitive, accom-
panied by a concomitant syntactic reanalysis. As a result of the reanalysis, the
infinitive is now the complement of the deontic semi-auxiliary have and is no
longer a modifier/adjunct in the NP, while the original object of have is now
the object of the infinitive.36 The word order has been ‘reshuffled’ according-
ly. The following Old English example contains shades of necessity as well as
possession:
(56) is a less ambiguous example from the Canticles of the Cambridge Psalter
(which, unsurprisingly, must have been heavily influenced by the Latin
original):
This example shows beyond doubt that HABBAN must have operated as
an auxiliary of necessity accompanied by an inflected infinitive already
in Old English.
37 Though Fischer (2015) raises some legitimate objections to the traditional outline above,
they would take us too far afield in our pursuit.
38 Have got to was grammaticalised later than have to – see Łęcki (2010: 138–142) and Section
3.5.2.4 below.
70 chapter 3
from 21st-century American English, where have (got) to/(have) gotta is again
much more popular than must, can be found in Łęcki (2010: 225 n. 54).39
Finally, there are indications that, apart from being omissible, as in gotta
(you got to laugh at me, 1911, from Łęcki 2010: 142), have in this construction can
now be mixed up with be, as in (57)–(58). I return to this interchangeability of
be and have in Sections 3.5.3.2 and 3.5.3.3.
(57) What I’m gotta do first … [BNC, cited in Łęcki (2010: 142), quoting Krug,
emphasis mine]
(58) I’m gotta finance industry so … [BNC, cited in Łęcki (2010: 142), quoting
Krug, emphasis mine]
Another modal use of have is in the had better idiom (see Quirk et al. 1985:
141ff.). Strangely enough, had better replaced an Old English construction
made up of the OE verb beon ‘be’ accompanied by bet(ere) ‘better’ and a dative
NP followed by a that-clause (Łęcki 2010: 69–70), again showing paradigmatic
and cognitive links between verbs of being and possession:40
39 Interestingly enough, be also developed (semi-)modal uses in combination with an infini-
tive, or maybe had them all along, as in (i). Such uses, though with an active instead of a
passive infinitive, are attested as early as Old English, as in (ii), perhaps at least in part in-
fluenced by Latin originals (see Fischer 1991: 146–151, 174ff.; cf. Mitchell 1985: §§934–944).
Latin influence or reinforcement is also possible for the use of have as a deontic auxiliary
in OE (see Łęcki 2010: 114–118). In Latin, habere + infinitive could refer to the future, as
well as to obligation, (iii) (cf. the translation example above, as well as Chapter 7 for a
have-based future in Bulgarian).
(i) They are to be blamed therefore. [from Fischer (1991: 146)]
(ii) quanti psalmi dicenda sunt [Latin]
hu fela psealma … to singenne synt [Old English]
how many psalms to sing are
‘how many psalms are to be sung’ [from Fischer (1991: 149)]
(iii) Haec canta-re habe-o
these sing-inf have-1sg
‘I have these to sing’ or ‘I have to sing these’ [from Łęcki (2010: 114)]
40 Kovatcheva (1982: 17, 25), Bauer (2000: 197–203, 211), Danylenko (2005), as well as
Grković-Major (2011), highlight a similar interchangeability of generalised possessive
schemas involving ‘I have something’/‘Something is at/to me’, e.g. Latin pugnatum mihi
est ‘fought is to me’, i.e. ‘I have fought’ (cf. Heine 1993: 31–32, 1997). In Hiberno-English,
possession can be expressed with be + near/at/by, as in There wasn’t any money by that boy
‘That boy didn’t have any money’ (Brinton and Arnovick 2011: 465).
The Story of the English Perfect 71
Unlike (59), in the had-better construction, the noun phrase is in the nomi-
native and a bare infinitive is used instead of a that-clause. The infinitive,
which could appear in this context as early as OE, started to be used in ME
with a greater frequency than the that-clause. The Middle English example
below still has the verb be and an oblique argument:
Have, in its past subjunctive form had, first appeared in this construction in
late Middle English; the subject (though missing here) is now nominative and
better is followed by an infinitive rather than a that-clause (Łęcki 2010: 71–72):
3.5.2.4 Summary
At this point, I should repeat that, in addition to the other uses enumerated so
far, have still occurs in complex transitive clauses with an undeniably adjecti-
val object complement, as in (62), which comes from colloquial Present-Day
English:
(62) I had it right the first time, didn’t I? [conv, from Biber et al. (2002: 201),
emphasis mine]
(63) Make sure the plug casing is in one piece, and replace any that are cracked,
chipped or, worst of all, have bits missing that could expose live parts. [BNC]
Summarising his findings, Łęcki (2010: 207) gives the following approximate
dates for grammaticalising the various functions of have (though there might
be periods of uncertainty for most of these):
– perfect have + past participle – from 800
– causative have + past participle – from 1100
– causative have + infinitive – between 1300 and 1400
– obligative (i.e. deontic) have to – from 1300
– have got to – late 19th c./1900
– had better – from 1500
Based on these somewhat conservative dates and also adding passive have into
the mix, it is right to talk about the polygrammaticalisation of have (cf. discus-
sion in Chapter 2). If the whole point of displacing be from the perfect was to
restrict the functional load of both auxiliaries, this ‘extravagant’ increase (or
maintenance) of have’s own functional load is rather problematic, as is the
creation of yet more alternative passives in competition with be, as well as per-
fects in competition with have itself.
(64) Insteade of mentioninge his name: Jo: fox the presbyterians gott his name
changhed: & putt in George ffox ye quaker. [[HC] Fox 155, cited in Rissanen
(1999: 216), emphasis mine]
and Cable (2013: 335), he got hurt arose because he is hurt is too static in mean-
ing, whereas he became hurt is too formal.41
Indeed, some might say that the new get-passive has evolved in order to
dispel the existing lack of clarity, since potential ambiguity persists to this day
in They were frightened (passive or copular be – see Quirk et al. 1985: 167ff.).
Mitchell (1985: §796) makes similar claims regarding both OE weorðan and
ModE get (cf. Section 3.4.3 for analogous issues with weorðan-passives, which
could be dynamic as well as stative). The use of get instead of be normally gives
a dynamic, as opposed to stative, interpretation, as in (65) (see Biber et al. 2002:
171–172, among others). Get, however, is not completely free of ambiguity ei-
ther. X got frightened can describe an action, as in (66a), or a transition to a
state in a copular clause with an adjectival subject complement, as indicated
by typically adjectival modifiers such as very – (66b). Furthermore, the case
studies below demonstrate that ambiguity between passive and copular be
existed for a long time without any ‘intervention’ from get, which once again
casts doubt on the validity of functional explanations.
41 As already noted above, it is not terribly accurate to apply the term ‘auxiliary’ in these
circumstances. Although get has come to participate in passive and perfect constructions,
it shares none of the NICE properties characteristic of the earlier generation of auxiliaries
(see Huddleston and Pullum et al. 2002: 92ff.). By the time get was recruited for the roles
in question, the group of earlier auxiliaries proper had been established for a long time,
distinguished by their unique set of special features which do not seem to be extended
to ‘newcomers’ (cf. Heine 1993: 71–72, Krug 2001, Bybee 2003: 619–621). The same applies
to the later ‘auxiliary’ uses of have (see Łęcki 2010: 85–86, 219 n. 5). ‘Semi-modal’ have to,
for instance, serves as operator in subject-auxiliary inversion and hosts negative not only
exceptionally in some varieties of English. The following would be unacceptable to many
speakers:
(i) Why has it to be me, Jimmy? [M. Munro, Whispering Sands, 1961, from Łęcki (2010: 219)]
74 chapter 3
seen as a perfect, and has come to be synonymous with the stative possession
meaning of present-tense have, as in (67).
(67) The Amphibicar. It’s got little propellers in the back. <cf. It has little
propellers.> [from Biber et al. (2002: 112), emphasis added]
The erstwhile perfect auxiliary is now omissible in speech from the have got
phrase, and we appear to be witnessing the birth of something approximating
a new get-perfect (in addition to the get-passive mentioned above). In (68), get
heads a complex transitive clause with a participial object complement very
much reminiscent of the Old English ‘proto-perfect’ (Miller 2000: 344ff., 2004a:
239, and Brown and Miller 2016: 249–250 talk about resultatives with have got,
get and other verbs).
(68) You got your homework done, Jason? <cf. Have you got…?> [example from
Biber et al. (2002: 112), emphasis added; treated as causative-resultative
by Miller (2004a: 240)]
(69) Maybe Ernest Hemingway there could run me in a beer when you get done
maulin’ him. [1989, from The Dark Half by Stephen King]
have gone through a possessive stage (have got) or through a copulative (even
causative) stage, so in essence the possessive/copulative/causative schemas
have been recycled multiple times. The result of such repeated rounds of gram-
maticalisaion of generic and interchangeable lexemes is multiple functions for
a single auxiliary and multiple alternative auxiliaries serving the same func-
tion, which violates the ‘one form, one meaning’ principle (cf. Chapter 2, Heine
1993: 43).
Support for this hypothesis comes from the fact that, in addition to get-
passives and perfects, get is also being recruited as a causative in the construc-
tions ‘get someone to do something’ or ‘get something done’ in yet another
instance of polygrammaticalisation (see Hollmann 2003: Chapter 4, Miller
2004a: 240, as well as sentences (64) and (68) above; cf. also modal gotta in
Section 3.5.2.3, including the interchangeability of be and have there too). In
the following Facebook post, it is clear that someone else came to pick up the
boxes, so the implied agents of got and picked up are different:
(70) Finally got all the empty boxes picked up today for recycling/reuse! Our
shed is so much more spacious now [Facebook post, American English
speaker, 25 July 2017]
42 Bernard Comrie (p.c.) notes that he also has a be-perfect with transitive finish, as in I’m
finished my dinner (see Rydén and Brorström 1987: 93–94, for earlier examples and an il-
lustration of the rise of this construction). B. Comrie further shares that in Modern Dutch
be has come to be used in the perfect with transitive forget.
76 chapter 3
The structures in (71) and (72) above are likely to be judged as degraded by
many native speakers, but there is also a more widespread ‘flavour’ of new be-
perfect that occurs in varieties of English spoken on the British Isles, including
Scottish English and also Northern English more generally (see Miller 2000:
345–349, 2003, 2004 a, b, Brown and Miller 2016: 251–252, Weir 2016). As evident
from (73), this pattern takes the form: that be NP Pred (that followed by a form
of be, followed by a noun phrase [determiner phrase (DP) in some theories],
and then followed by some kind of predicate). The predicate can be a past
participle, among other things. The more passive-like construction (73a) is dis-
tributed more widely in the UK than the more active type, (73b)–(73d), which
is restricted to the north and Scotland.
(73) a. That’s him consulted. (≈ He has been consulted, I’ve consulted him)
[Miller (2000: 346, 2004 a, b), quoted in Brown and Miller (2016: 251)
and Weir (2016)]
b. … that’s me seen it. (≈ I’ve seen it now) [Miller (2000: 347, 2004 a, b),
quoted in Brown and Miller (2016: 251) and Weir (2016)]
c. [Is] that you left the school now? (≈ Have you left school?) [Miller (2004
a, b), quoted in Brown and Miller (2016: 251) and Weir (2016)]
43 Consult Elsness (1997) for a much more detailed discussion of the distribution of these
past-referring forms, including various kinds of perfects, as well as further refinement of
the data sources and other considerations.
44 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
78 chapter 3
Development to 1600
100%
80%
60%
40%
20%
0%
Old English Early Middle English 1350‒1400 1550‒1600
Present perfect Preterite Other
figure 1 Relative frequencies of major past-referring verb forms from Old English
to 1550–1600
source: Elsness (1997: 270)
60%
40%
20%
0%
1750‒1800BrE 1750‒1800AmE CONTBrE CONTAmE
Present perfect Preterite Other
The same can be said about the breakdown of auxiliary roles themselves, as
shown in Table 3b.
table 3b Auxiliary uses of be in the history of English [source : Kilpiö (1997: 106)]
As becomes evident from this data, the passive auxiliary function of be has
been its most important auxiliary function throughout the history of English.
The figures for the perfect are negligible in all the periods, and there are no
signs of ambiguity increasing or decreasing significantly over time. Kilpiö
(1997: 107–108) notes that the relative stability in the roles of be goes against
‘the expectations that the relative share of auxiliary uses at the expense of the
remaining two uses would rise’. Furthermore, the statistics invite the conclu-
sion that be died out as a (traditional) auxiliary of the perfect just because it
was vanishingly rare in this job from the very outset, not because of any pre-
sumed excessive functional load or any rife ambiguity.45 By contrast, 26% of
all the instances of habban in the Old English section of the Helsinki Corpus
involve the perfect, which makes the perfect periphrasis the most common
function of habban in Old English (Łęcki 2010: 184, citing Kilpiö).
45 Compare the statistical data in Elsness (1997: 264–265), where the proportion of present
perfects with be does not exceed 1.2% throughout the periods he studied. Drinka (2017:
252) reports similar results. Refer back to Chapter 2 and Section 3.4.1 for the role of fre-
quency in grammaticalisation.
80 chapter 3
46 It might be problematic that earlier Swedish had a somewhat greater amount of be-
perfects, but perhaps the important thing is that they shrank over time (see Drinka 2017:
248–250). On substratal contact in Hiberno-English as linked to the retention of earlier
English patterns, see Harris (1984).
47 Mitchell (1985: §870) casts doubt on the aspectual connotations of prefixes in OE.
82 chapter 3
3.6 Conclusion
This chapter has charted the development of the perfect in English. In order
to ensure a more comprehensive treatment, the perfect has been considered
in the context of other constructions, such as the passive, and also against the
background of the polygrammaticalisation of generic/delexical verbs such as
be, have and get. Special attention has been paid to the explanations provided
in the literature, including functional load, frequency and contact. Ambiguity
has also figured prominently as an enabling factor for reanalysis.
The opening sections traced the grammaticalisation of have in Old English –
an original complex transitive clause with lexical have and a participial object
complement came to be reinterpreted as a perfect construction with auxiliary
have and a main verb in the past participle. Contexts involving legitimate zero
inflections on the initially adjectival participle have been deemed instrumen-
tal in enabling the reinterpretation due to the lack of unequivocal marking.
Consequently, agreement was later lost across the board from the verbal per-
fect. The second major concern in this chapter has been the competition be-
tween have and be in the perfect domain, and the reasons why be ultimately
became defunct. Contributing factors that have been put forward by earlier
scholars include:
– the disparities in the relative frequencies of the two constructions;
– the functional ‘overload’ of be as copula, as well as perfect, passive and pro-
gressive auxiliary;
– the undesirable ambiguity which the above ‘functional promiscuity’ could
lead to (the latter two obviously go together and are by far the most popular
explanations in the specialist literature);
– the role played by the ancillary confusion-fostering contexts of coordina-
tion, contraction and auxiliary ellipsis.
The Story of the English Perfect 83
Problems for this neat and orderly traditional story come from several di-
rections. Firstly, the non-straightforward interaction between morphological
marking and perfect readings is damaging to the causality inherent in the OE
reanalysis account – presence vs. absence of morphological marking and pos-
sessive vs. perfect interpretations do not match neatly as expected. Secondly,
the functionalist take on the competition between the two auxiliaries faces
challenges from the polygrammaticalisation of have itself (arguably, have ought
to have remained restricted to the perfect rather than acquiring a range of ad-
ditional functions, because supposedly its ‘monofunctionality’ is what gave it a
competitive edge in the first place). Further challenges are posed by the devel-
opment of new rival perfect and passive constructions, including be-perfects.
The actual picture is very far from Humboldt’s ‘one-form-one-meaning’ ideal.
In a series of repeated and overlapping waves of grammaticalisation, English
has seen the birth of both perfects and passives built with be, have and get.
It is therefore concluded that the various cases of polygrammaticalisation in
the domains of the perfect and the passive do not obey functional laws at all;
instead they are just natural consequences of the generic meanings (often in-
terchangeable) of verbs like be, have and get, coupled with common cognitive
processes of routine reanalysis.
Corpus evidence indicates a remarkable stability in the relative shares of
the auxiliary functions of some of the relevant verbs under consideration and
suggests that ambiguity tended not to fluctuate wildly, consistently remain-
ing fairly low. Rates of perfect be also remain low throughout the history of
English, while have has seen a dramatic rise and was much more common in
that role from the outset. This lends credibility to the frequency hypothesis.
Another plausible-sounding suggestion might be that be-perfects receded as
a result of contact with Scandinavian. These claims will be substantiated in
subsequent chapters.
For now, it might be instructive to conclude with a parallel from another
area of grammar. It is a well-known fact that syncretism is omnipresent
even in richly inflected languages. Even in a highly synthetic variety such
as Old English, in many cases the ‘apparently crucial’ distinction between
nominative and accusative can itself be neutralised; in Old English, this hap-
pens with all neuter nouns, with all plurals, in the singular of consonant-stems,
as well as with heavy i-stems. Lass’s (1994: 138–139) moral from those observa-
tions is:
that there is a great difference between what a language has and what it
does with it; this should make one suspicious of any kind of facile argu-
ment suggesting that changes are ‘caused’ by the growth of morphologi-
cal ambiguity.
84 chapter 3
This chapter marks the beginning of the more empirical part of the book.
In it I follow the evolution of the perfect and the passive in the Anglo-Saxon
Chronicle, before moving on to Middle and Modern English texts in Chapter 5.
The objectives of the study (Section 4.1) have been informed by the earlier
discussion, so I will be trying to unravel the intricacies of frequency, reanaly-
sis, uncertainty of interpretation, ambiguity and morphological marking (in-
cluding the influence of zero exponence), based on a selection of original Old
English prose. Before I get to the empirical data, I discuss some of the pitfalls
that should not be lost sight of when working with earlier stages of languages
(4.2), and justify my choice of samples and my approach (4.3). Then I proceed
to analyse the actual texts, successively examining Anglo-Saxon Chronicle en-
tries from the eighth century (4.4.2.1), followed by some interim remarks on OE
agreement (4.4.2.2), and then on to the ninth century (4.4.2.3), the late ninth
and early tenth centuries (4.4.2.4), finally supplemented by entries from the
Peterborough Chronicle for the late tenth and early eleventh centuries (4.4.2.5).
In the rich and fascinating material yielded by the texts, I do find support for
potential Scandinavian influence on the development of the perfect, as well
as differential rates of the grammaticalisation of the passive and the various
types of perfect, including curious idiosyncrasies of agreement behaviour.
Morphosyntactic agreement will be a major recurrent theme in the chapter
as it is inextricably linked to the questions of grammaticalisation and reanaly-
sis. Importantly, ambivalent zero exponence is demonstrated to play no sig-
nificant part in the relevant developments, so its role has been overstated in
the existing literature. The chapter’s setting is roughly that of Construction
Grammar, with a special emphasis on gradual changes of auxiliation which
have produced synchronic gradience in terms of the entrenchment of a cluster
of periphrastic schemas in the Old English slice in time.
The following objectives have been set for a philological study of a selection of
passages, presented in this and subsequent chapters. Naturally, the objectives
reflect the major issues which emerged in reviewing the literature and analys-
ing the preliminary data in Chapter 3.
– I will ascertain how frequent the different types of perfect are in a sample
of OE texts and will then compare the rates to frequencies in analogous
ME and ModE text samples, as well as to similar nascent proto-perfects in
Bulgarian.
– An important question to keep in mind will be the following: how often
are the occurrences ambiguous between a perfect syntagm and an SVOC O
clause, taking into account the presence or absence of morphological agree-
ment marking and the other factors identified earlier? Again, comparisons
will be drawn between the selections from the different periods, in an at-
tempt to find out whether ambiguities increase or decrease as time goes by.
– What is the role (if any) of zero exponence as a bridging context enabling
reanalysis and as a contributing factor for the loss of agreement across the
board? Is the loss (or lack) of agreement a cause or a consequence of more
advanced grammaticalisation?
– The corpus research will feature an analysis of be- and have-perfects in
the material, recording any ambiguities between be-perfect and passive or
stative constructions. Middle and Early Modern English will be instrumen-
tal in establishing whether ambiguities proliferate over time. Parallels will
be drawn to Bulgarian and German; the latter retains both types of perfect,
as well as a passive with werden ‘become’ (arguably relieving the functional
load of sein ‘be’). In addition, I will trace occurrences of coordinated past
participles, of auxiliary ellipsis, as well as of ’s which is ambiguous between
is and has, and whether there are any significant diachronic fluctuations.
Compiling and working with a diachronic corpus, albeit a modest one, comes
with specific challenges of several kinds. The most important of these are
touched upon by Elsness (1997: 255), and here I summarise Elsness’s main
points. In the first place, the text samples selected for diachronic comparison
can never be one hundred percent parallel. The disparities between texts from
different periods can be due to a number of reasons, such as changing styles
of writing or changing topics of interest (cf. Macleod 2012: 122, 179ff., 226ff., for
similar issues surrounding the complexities of synchronic stylistic variation in
earlier periods). It is virtually impossible, both theoretically and practically, to
draw clear lines of demarcation between changes in language due to purely
linguistic factors and those due to extra-linguistic phenomena. Developments
the Perfect in a Selection of Old English Texts 87
in society and culture can obviously affect the lexicon, but there could be more
subtle influences too. Elsness (1997: 255) notes that the temporal reference ex-
pressed by language might be skewed, since many early texts that survive deal
with events situated in a reasonably distant past, while many contemporary
texts are often interested in the more immediate context, including the imme-
diate and recent past. Such shifts in focus and subject matter could to a certain
extent influence the way verb tenses and other temporal indicators are used
in a passage (cf. Macleod 2012: 72–73). For instance, the Old English samples
to be presented below are dominated by pluperfects, while my Middle English
samples show a number of present as well as past perfects. This must surely
be an epiphenomenon due to the nature of the selected material – the plu-
perfect is to be expected in annals dedicated to historical events such as the
Anglo-Saxon Chronicles.
Placing these concerns about the nature of the evidence on a more solid em-
pirical footing, Szmrecsanyi (2016) provides empirical confirmation that there
are limitations associated with inferring grammatical change from variable
text frequencies in historical corpus data. Based on a case study, Szmrecsanyi
(2016) verifies that fluctuations in the frequencies of grammatical variants in
real time may not only reflect genuine grammatical changes in progress, but
can also be conditioned by other, extraneous ‘environmental’ changes, such as
a change in content or textual subject matter, as conjectured above.
Thinking along similar lines, Mustanoja (1960: 505) too highlights certain
pitfalls associated with our historical sources:
Figures based on two late ME literary texts […] suggest that even at the
end of the period the general frequency of the perfect is many times
smaller than that of the preterite. The notion obtained from these figures
is perhaps not quite accurate, for there is reason to assume that the fre-
quency of the perfect is somewhat higher in everyday speech, where the
speaker naturally tends to look at events from the angle of the moment
of speaking.
There are several varieties subsumed under Old English and even more
under the designation Middle English, so it cannot be assumed that there
is necessarily a direct developmental relation between the languages rep-
resented in the texts […]. Still it is clear that in some general sense a type
of diachronic relation exists.
Despite all the aforementioned potential pitfalls and caveats to be wary of,
Elsness (1997: 255) also concludes that broad diachronic comparisons are
nevertheless worthwhile endeavours capable of yielding meaningful, valid re-
sults. Issues related to corpus design and compilation are further discussed in
Jacobs and Jucker (1995: 6ff.), in Chankova (2008: 16–18), and in many other
sources dealing with the diachronic description of English. Conscious of the
limitations of our data, we can still draw some non-trivial conclusions and gain
insights from a corpus study. The choice of samples will be essential in over-
coming at least some of the limitations, at least in part.
As has emerged already in the previous section, it has been noted by schol-
ars (e.g. Zimmermann 1968: 108–158, quoted in Fischer 1992: 258 and Macleod
the Perfect in a Selection of Old English Texts 89
2012: 230), that the Middle English present and past perfect appears more fre-
quently in colloquial style, while it is rare in purely narrative contexts; it is
also frequent in instructional genres. In the Early Middle English works that
Zimmermann examined (Ancrene Wisse, the ‘Katherine Group’, Layamon/
Laȝamon’s Brut and The Owl and the Nightingale), the perfect is found almost
exclusively in direct speech (cf. Macleod 2012: 229–230, for Old English). This
observation highlights the link between the perfect and the colloquial register.
Therefore, the investigator should seek texts which approximate the spoken
usage of the time.
Analysing be- and have-perfects, Rydén and Brorström (1987) and Rydén
(1991: 343) focused on two types of text, private letters and comedies, since
they believed that those genres would reflect the ‘frontier’ uses. However, some
of the texts Rydén and Brorström examined turned out to be conservative rath-
er than innovative and progressive (Rydén 1991: 343). Overall, comedies proved
to be ‘more genuine reflexes of non-conservative language’ (Rydén 1991: 344;
see Rydén and Brorström 1987: 197, 199, 200–201). This finding is not surprising,
as private letters, depending on stylistic consciousness and levels of literacy,
might be formal as well as informal, though of course in principle they are ex-
pected to be less formal than official letters. As Rydén (1991: 344) recapitulates:
On the whole, our texts can be said to represent informal written lan-
guage, with some speech-based elements, rather than spontaneous spo-
ken discourse, a fact which should be kept in mind when it comes to
assessing the texts (or rather the grammars as extracted from the texts)
as ‘spearheads’ of the paradigmatic development.
Following the experience and knowhow obtained from earlier research, here
I will also be examining genres which have proved to be reasonably reliable
reflections of genuine contemporary usage. My mediaeval and Early Modern
samples will be based on comedies and comic tales, as well as on private letters.
While there are a number of electronic corpora available, synchronic as
well as diachronic, I have instead opted for a more traditional philological ap-
proach, investigating a selection of texts. There are several reasons why I have
chosen not to avail myself of an authoritative and truly invaluable database
such as the Helsinki Corpus (or a later version). First, I am looking for ambigu-
ity, which has been intentionally left out or unwittingly neglected in earlier
empirical research (e.g. Rydén and Brorström 1987: 24, 32). Moreover, ambi-
guity is difficult to code or tag (or sometimes even to detect), so reading the
actual texts in their entirety might work better. Close reading will also make it
easier to immerse oneself in the broader context, which often turns out to be
relevant to the linguistic analysis. As Macleod (2012: 241) admonishes,
90 chapter 4
[i]f research upon a language is based on data that do not reflect actual
usage, but rather a statistical composite reflecting the practices of no
individual speaker, the conclusions reached by such means necessarily
have a less stable foundation than might be wished.
Second, since many previous studies have been based on the same material,
especially the Helsinki Corpus, this might skew the data and the findings;
therefore, I have decided to put together a slightly different, tailor-made sam-
ple which will complement the earlier larger-scale studies. Thus breadth of
coverage and size have been sacrificed for the gains of depth and close reading.
Below I will first examine a number of entries from the Anglo-Saxon and
the Peterborough Chronicles, tracing the use of the perfect in the light of mor-
phosyntactic agreement, ambiguity and semantic interpretation. I will then
move on to the perfect and passive auxiliaries as used in Chaucer, 15th-century
drama, Shakespeare and Restoration Comedy, as well as the correspondence of
the Pastons and of Samuel Pepys (Chapter 5).
this A version is the oldest, it is not the author’s original and other versions
sometimes preserve better readings. One of these is the so-called Cotton man-
uscript, Tiberius A.vi (London, British Library, Cotton Tiberius A.vi), referred
to as Manuscript B. Manuscript B was copied by a single scribe from a lost
source close to Manuscript A in the late tenth century (between 977 and 1000)
and goes on until the year 977, after which it was not continued. Although this
is disputed, the manuscript is traditionally associated with Abingdon (http://
asc.jebbo.co.uk/intro.html, accessed on 19 Sept 2016).1 The A version will be
consulted for the years between 700 and 950, as it appears in digitised format
on the following website: http://asc.jebbo.co.uk/a/a-L.html (accessed on 19
Sept 2016). In principle, I will be exclusively analysing the A text, only checking
other manuscripts occasionally if and when necessary, for instance if there are
pertinent divergent readings noted in the editions used.
The other manuscript of interest here is most famously known as the
Peterborough Chronicle. It is also designated as Manuscript E (Oxford, Bodleian
Library, MS. Laud 636), and was most probably initially copied in 1121 or 1122
at Peterborough and maintained in various hands until 1154 (see Plummer and
Earle 1892: xii, Irvine 2004). This is a representative of the so-called ‘Northern
Recension’ of the Chronicle, and the transcript I use is the one which appears
at http://asc.jebbo.co.uk/e/e-L.html (accessed on 19 Sept 2016). For both
Manuscripts A and E, the online publications were verified with a more au-
thoritative traditional source, Plummer and Earle’s (1892) printed parallel
edition (which was fed into the Helsinki Corpus too). In order to ensure ac-
curate interpretation, I relied on the translations from Whitelock et al. (1961).
1 See however Plummer and Earle (1892: x–xi), who point out a Canterbury connection, as they
do for MS A.
92 chapter 4
Have-perfects 1 1 0 2 6.7%
with transitive
verbs
Have-perfects with 0 0 0 0 0%
intransitive verbs
Be-perfects with 0 0 1 1 3.3%
intransitive verbs
Progressives 0 0 3? 3 10%
Passives with beon 3 1 12 16 53.3%
Passives with 0 0 2 2 6.7%
weorðan
Ambiguous 0 0 5 5 16.7%
passives/statives
with beon
Ambiguous 0 0 0 0 0%
passives/statives
with weorðan
Ambiguous 0 0 1 1 3.3%
be-perfects
Hardly any perfects were found in this section of the Chronicle. As noted in the
general discussion of corpus planning, such a genre with exact dates from the
reasonably distant past presupposes fewer perfects, since reference is made to
past time without much relevance to the immediate context in the present (i.e.
the scribe or the author’s here and now; cf. Macleod 2012: 72–73). Hence, the
few perfects I register are all pluperfects. In addition, most of this manuscript’s
entries for the eighth century are rather brief and matter-of-fact,2 and most of
2 An especially short, vivid and poignant example comes from the entry for 761:
(i) Her wæs se micla winter. [761]
‘Here [i.e. in this year] was the big winter.’ (translation mine)
the Perfect in a Selection of Old English Texts 93
the perfects I did harvest (two out of three) come from an exceptionally exten-
sive narrative for the year 755. Of course, the scarcity of the perfects might in
part be due to the early date of composition of the originals from which the
Parker scribe copied. There is ample evidence that this text is too early for the
perfect to have come into its own (this would happen centuries later):
The pluperfect is not used where it would be expected today – instead, the
Anglo-Saxon in (1) has the simple preterite (for more details, see Mitchell
1985: §611ff., §§634–644, Macleod 2012, Los 2015: 73–74); here and in many
other places elsewhere, only context or lexemes like ær ‘ere, earlier, previ-
ously’ are relied on to signal past in the past, as in (2) (on ær, see Mitchell
1985: §§1112–1115).
3 Following the actual scribal practice, the electronic edition used as the source of the ex-
amples distinguishes two forms of the OE character æsc: ę and æ (cf. Irvine 2004: civ).
4 The Appendices of this book have been made available online. They can be accessed through
a QR code which can be found in the References section of this book.
5 On the relative order of elements in perfect and passive constructions, see Mitchell
(1985: §703ff., §§757–758), and the discussion in Chapter 3. Word order might also affect
agreement options, as is evident from Mitchell (1985: §706, §710). Kilpiö (2007: 329–330, 2013:
108) demonstrates some intriguing correlations between word order and agreement in the
perfect – placing the object before the participle decisively favours agreement, while plac-
ing the object after the participle disfavours agreement (cf. the Dictionary of Old English at
http://tapor.library.utoronto.ca/doe, accessed on 27 Feb 2017).
94 chapter 4
Of the two have-perfects with transitive verbs, both from the account for 755,
example (3) shows the agreement expected of a participial object comple-
ment in the accusative masculine singular (-ne in the strong declension),
while no such agreement is present in (4); for it, neuter plural should be in-
dicated with -u on the past participle belocen, but the -u is not there, so at
least as far as the have-perfects here are concerned, agreement morphology
is dropped even without cases of ambiguous zero marking. Neither does the
context in (3) strictly suggest anteriority in the past (cf. (12) below, again with
oþ ‘until’).
By contrast, the only unambiguous be-perfect with a mutative intransitive
verb, (5), is of the type for which it is impossible to tell whether it does or does
not agree, since the singular morphological marker would be zero anyway. In
terms of idiom, was… agan could be a common set phrase or formula, as could
many other examples in this chapter, e.g. wæl + habban/beon/weorðan + ofslæ-
gen, to ærcebiscepe + gehalgod, gefeoht + gefohten, geweorc + geworht, etc. (see
Danchev 1969: §59ff., Smith 2001, and earlier chapters for the resilience and
survival of frequent archaic set phrases with a high degree of entrenchment).
the Perfect in a Selection of Old English Texts 95
Agreement here might have defaulted to neuter singular (with a zero expo-
nent on agan) because of the numeral + noun subject xliii wintra. Mitchell
(1985: §563) notes that when a finite verb precedes a numeral acting as subject,
the verb may be singular or plural. Wintra is a masculine noun, most probably
appearing in the partitive genitive plural because it follows a numeral, so it is
not nominative and therefore cannot trigger agreement. Since the finite auxil-
iary was is singular, the participle has no option of being in the plural (though
see the next paragraph).
A comparison to the (proto-)passive would be especially instructive. There
are many more prospective passives than perfects in the sample, which might
conceivably suggest that the passive had reached a more advanced stage of
grammaticalisation by this time; on the other hand, the greater number might
be due to the subject matter of this type of narrative. Three of the passives
show agreement, although it is rather defective in one of them, given in (6) (cf.
Chapter 3, Section 3.1.2):
Interestingly enough, at this early stage, non-zero agreement markers are not
lost in the perfect and passive on any significant scale (a total of two instanc-
es), even though there are quite a few cases with expected zero morphemes,
especially in the passive (see Table 1 above and Appendix 2.1).8
6 The invariable relativiser þe presumably takes its gender and number features from its ante-
cedent geferan. Note plural wærun.
7 The attributive adjective wunderleca takes the traditional strong nominative feminine
plural -a, whereas the -e on the predicative participle might be influenced by the masculine
nominative plural. It is widely attested as an alternative in the feminine as well, however (see
Chapter 3, Table 2). Might this mean that the two are treated differently? These attributive
and predicative elements are certainly given different markers, suggesting that they were
probably not perceived as equivalent (more on this to follow). On whether gesewene could
be an adjective, ‘visible’, see Mitchell (1985: §766).
8 The Appendices of this book have been made available online. They can be accessed through
a QR code which can be found in the References section of this book.
the Perfect in a Selection of Old English Texts 97
Very often it was difficult to tell whether something was a genuine passive
structure with dynamic meaning or a stative copular clause with linking be
(Rydén and Brorström 1987: 27 admit similar difficulties):
The example in (10) was regarded as dynamic because of the adjunct of fre-
quency oft, though it could be stative nonetheless (cf. e.g. ‘He was often sad/
in a bad mood/incapable/out of action’, see also Mitchell 1985: §§766–767). By
contrast, the following were treated as ambiguously stative:
In (12), adjectival status for the participle might be indicated by the modifying
adverb swiþe ‘very, exceedingly’ (see Mitchell 1985: §741). In addition, (10) and
(12) show very clearly the flexibility of word order pointed out already.
Lack of overt morphology where it is expected did not always ensure a pas-
sive reading, as in the following potentially copular clause, in which the parti-
ciple (subject complement?) is missing the traditional -u morpheme to mark
nominative feminine singular agreement with the subject:
98 chapter 4
After some deliberation, (13) was put down as a passive, though for many
of these examples it was not at all easy to decide with any amount of con-
fidence. This particular decision was prompted by the dynamic semantics of
give, not by the missing agreement morphology.10 It will emerge below that,
in all the periods studied, the grey area between genuine passive and a stative
copular clause remains uncurbed (cf. Mitchell 1985: §735). By contrast, only
one example was judged to be ambiguously perfect or passive/stative (see
Appendix 2.1,11 section on ambiguity between be-perfect and passive/stative) –
it involves the verb for ‘gather’; ‘the synod was gathered/collected’ can be inter-
preted as ‘someone had gathered/collected the synod’ (hence passive) or ‘the
synod had gathered/collected’ (perfect; more on this to follow).
The other OE proto-passive auxiliary, weorðan ‘become’, does not seem to
suffer from such a ‘functional disadvantage’ to the same extent; weorðan is nev-
ertheless used much less frequently than beon ‘be’ here (only twice), despite its
apparent ‘functional superiority’.12
9 Feminine and neuter -u endings could be dropped because participial stems might be
treated as heavy/long (-u is generally lost after heavy/long syllables). However, the past
participle suffixes (strong -en and weak -ed/-od/-ad/-d/-t – see Mitchell 1985: §983) did
not contain long vowels and did not create clusters of consonants (in my material here),
which is what makes a syllable heavy. Note Mitchell’s examples with feminine singular -u
in the next section.
10 Cf. observations above and the discussion of this sentence in Mitchell (1985: §839), where
it is also treated as a bona fide passive.
11 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
12 Weorðan however is not completely immune from ambiguity between a passive and a
copular reading – see Mitchell (1985: §735, §798) and the discussion in Chapter 3, Section
3.4.3.
the Perfect in a Selection of Old English Texts 99
Quite apart from perfects and passives, the texts turn up a causative con-
struction with a now-lost verb which meant ‘to command’ – OE (ge-)hatan,
preterite singular het.13 This was later supplanted by causative have (among
other options), as indicated by the Modern English translation in (15). First
introduced in Chapter 3, the topic of the increased functional load of have will
be taken up again in subsequent sections and chapters.
Mitchell (1985: §33, §738) however adds that the form without -u is more com-
mon, as in (18):
In fact, there are no feminine singular forms with -u in the entries of the
Anglo-Saxon Chronicle between 734 and 1001 (Mitchell 1985: §33 fn. 7; cf. previ-
ous section too).
In the plural, the adjective or participle can either remain uninflected or
receive a generalised -e for all genders. Mitchell (1985: §34, §760) observes that
feminine plurals in -a and neuter plurals in -u are rare. Uninflected plurals do
occur but in general appear to be the exception rather than the rule. According
to some sources, only one indisputably uninflected plural is attested in the
Anglo-Saxon Chronicle between 734 and 1001 (Mitchell 1985: §34, with refer-
ences, though see the next sections, where I report more and return to this
example):
14 Below is a late example from MS E with plural agreement which is not included in my
sample (see Mitchell 1985: §718):
(i) Đas þing we habbað be him ge-writ-en-e, ægðer ge
these thing(nt)[acc.pl] we have about him pref-writt-en-pl both
god-e ge yfel-e
good-pl and evil-pl
‘These things we have written about him, both the good and the bad.’ [Chronicle, MS
E, 1086, quoted in Mitchell (1985: 288), gloss and translation mine]
102 chapter 4
The perfect is still extremely rare. Even in clauses like (20), where we would
use it today alongside the passive, Old English only has a preterite passive:
Of the two have-perfects from this sample, again one shows agreement with an
accusative masculine singular object, (21), while the other one does not, (22)
(appearing further down).
It should be noted that the (proto-)perfect with have and an agreeing parti-
cipial object complement closely parallels in morphology and syntax the sec-
ond clause, which exhibits an agreeing accusative modifier (ungecyndne) of
the direct object (cyning), and has the indisputably lexical verb underfon ‘to
receive, get, take’, which appears in the preterite plural – underfengon (the past
participle of underfon is underfangen). It is thus safe to conclude that have is
not fully grammaticalised here – due to the parallel to the second clause and
the adjectival agreement. At the same time, it is clear that the people referred
to as ‘they’ no longer literally ‘have’ Osbert as a king; therefore, the interpreta-
tion should be perfect-like, despite the agreement and the parallel with the
next clause (readers will remember the relevant discussion from Chapter 3;
also see Macleod 2012: 120). The meaning of possession has been bleached in
favour of abstract anteriority.15
15 Yet Mitchell (1985: §728) counters, not entirely convincingly, that the original sense of this
construction was ‘They had (=held) their king Osbryht (having been) rejected’.
104 chapter 4
The other have-perfect from the sample ‘should’ also supply its participle
(afliemed) with the accusative masculine singular agreement morpheme -ne,
but it is now missing:
In the first place, it is worth pointing out that a perfect is here used with ær
‘previously, earlier’, unlike what we saw above (cf. Mitchell 1985: §638, with ref-
erences). Attention should also be drawn to the word order, with separation
of the auxiliary and the participle; this, according to Denison (see Chapter 3,
Section 3.1.2), should suggest that (22) is not a real perfect yet. Nevertheless,
there is certainly no agreement, so this is obviously a transition stage and there
is a clash between the various criteria. As demonstrated above, the universal
applicability of criteria such as presence or absence of agreement is at best du-
bious. Both examples arguably show semantic bleaching of have, whereby the
object is not around and cannot be had or possessed, but the agreement does
not seem to be affected by that. Probably due to chance, there are no cases in
this sample of perfects with expected zero morphology – the alleged original
foothold for the loss of agreement.
There are six progressive-like combinations. They are important for two rea-
sons. Firstly, some of them show that -ende is simply the unchangeable present
participle ending, even when the subject is singular, as in (23):
16 However, see Mitchell (1985: §101, §974), who notes that both present and past participles
can be declined strong or weak and provides examples of inflected present participles.
the Perfect in a Selection of Old English Texts 105
whereas there are no traces of be-perfects here. This might have implications
for functional-load explanations. In any case, be indeed seems to have been
extremely rare in the perfect from the very outset (though have-perfects are
not faring much better in terms of statistics at this stage either).
By contrast, passives with be are relatively numerous. Four of them show
agreement with a plural subject:
Eorlas here is certainly nominative plural. It is not in the partitive genitive plu-
ral, which might be expected in the context of numerals. The auxiliary in (25)
is missing from the second clause, but could arguably be recovered from the
previous clause, so both occurrences were regarded as be-passives (the first
one ambiguously stative, however):
17 On passives introduced by ‘expletive’ þær, see Mitchell (1985: §754). Normally, examples
with (of-/ge-)slægen ‘slain’ were treated as bona fide passives, but those containing ex-
pletive þær and wæl were thought to be potentially stative. This is probably a slippery
distinction.
18 Although they share an auxiliary, these were counted as separate instances because of
the different morphological agreement options on the participles. However, whenever I
encountered the same auxiliary with, say, two zero-marked participles, this was counted
only once, as in the following:
(i) 7 þær wearþ monig mon ofslægen 7
and there became[sg] many man(m)[nom.sg] slain &
a-druncen on gehwęþere hond [853]
pref-drunk on either hand
‘and many men [lit. many a man] on both sides were killed and drowned there’
106 chapter 4
Essentially, what we get across the board is zero marking in the singular and -e
in the plural; despite the numerical advantage of the legitimate zero-marked
forms, this has not led to the loss of morphology where it is expected, as ad-
ditionally seen from a passive with weorðan from the same year (cf. Section
4.4.2.2):
This is significant because there are more instances of lack of plural agreement
in the Chronicle than the single one quoted in Mitchell (1985: §34, with refer-
ences, see Section 4.4.2.2 above). The finite verb wurdon is inflected for the
plural, but the participle gefohten is not. Importantly, however, example (28) is
still exceptional and we once again see a robust tendency for preservation of
agreement where it should be overt, regardless of the many cases of expected
zero exponence.19
Similarly to be above, there are analogous cases of ellipsis with weorðan:
In (29), weorðan is first used as a singular lexical verb; it is then only implied in
the second clause, where it should be plural and combined with an agreeing
past participle. Likewise, fela þusenda ofslægenra (lit.) ‘many of thousands of
killed’ in (30) implicitly goes with earlier wearþ, which should be construed
as lexical and existential in this stretch of the sentence, as opposed to the
rest, i.e. ‘there were many thousands of casualties’ [lexical] v. ‘were slain/re-
pelled’ [auxiliary(-like)]. This sharing of forms (one overt, the other(s) ellipti-
cal) blurs the lines between auxiliary and non-auxiliary uses of the same verb.
19 Gefeoht is a heavy-stem neuter noun which is zero-marked for the nominative plural in
(28) above. The partitive genitive could have been used instead because of the presence
of a numeral, but the genitive plural should take the morpheme -a. Compare (24) above,
in which eorlas follows a numeral but is overtly nominative plural. On verbs with ‘cognate’
objects/passive subjects such as gefohten and gefeoht, see Mitchell (1985: §754). In addi-
tion, zero marking is legitimate in the nominative neuter plural of the strong adjectival
paradigm, but again only for long/heavy stems.
108 chapter 4
Furthermore, when the auxiliary is missing like that, it is sometimes not clear
which form should be supplied, as in (30), where gefliemde ‘repelled, put to
flight’ was taken to belong with the weorðan-passives despite the presence of
progressive wæron in a later clause.
(31) also beautifully demonstrates the happy coexistence of zero forms in the
singular and -e in the plural:
The ellipsis in the second clause is not unproblematic, however. As it is, it was
taken to imply a form of weorðan for the gap, but another manuscript inserts
wæron in front of to gehalgode (see Plummer and Earle 1892: 80–81 fn. 5). In
sum, beon and weorðan are interchangeable in the passive (cf. the earlier dis-
cussion and Mitchell 1985: §786). In view of the solid numbers of weorðan and
its greater clarity/functional ‘streamlining’ (debatable though it is), it is myste-
rious why it disappeared (especially for a functionalist).
the Perfect in a Selection of Old English Texts 109
However, I would claim that weorðan gave the clause a more dynamic reading,
unlike ambiguously stative counterparts with beon (but recall the discussion
above and Mitchell’s objections to the rigidity of this distinction):
It is worth stressing again that this ambiguity between passive and stative
seems to have persisted throughout the history of the language, ‘blind’ to any
functional considerations and ultimately unresolved by the apparent func-
tional advantage of weorðan. In addition, future work will need to assess the
influence of frequent formulae like the one in (34) or geweorc geworht, among
others (cf. Danchev 1969 and the earlier discussion).
In conclusion, Table 2 confirms that the perfect is very rare and agreement
is either randomly present or absent from a clause where it ought to be overt.
There are no cases of expected zero morphology. Semantics seems not to mat-
ter, in that an unequivocally perfect reading can be enforced in both agreeing
and non-agreeing contexts. Passives and statives with beon and weorðan are at-
tested in greater numbers and suggest that loss of non-zero agreement cannot
be tied to ‘rampant’ zero exponence. Although legitimate zero exponence in
the singular is predominant, this has hardly resulted in loss of plural -e.
110 chapter 4
and fewer bona fide passives than in earlier samples (see Plummer and Earle
1892: 85 fn. 9, 99 fn. 2, 102 fn. 1, 103 fn. 1, where changes of hand are indicated).
The genre and subject-matter is still indisputably the same, however, so the
observed differences must for the most part be genuinely linguistic. Although
most of the earlier entries are much shorter, there is no reason why perfects
should not have been used so much, for instance. It is significant that, by the
tenth century, the perfect is employed (instead of the earlier simple preterite)
to indicate anteriority in the context of adverbs like ær, (35), or even without
needing their temporal adverbial reinforcement, (36).
As with the other samples so far, the years given after each example are as
they appear in Plummer and Earle (1892), and the translations are again from
Whitelock et al. (1961). At first glance, Table 3a summarising the data may look
rather messy, but once the material is examined more carefully, some interest-
ing patterns emerge.
It is immediately striking that agreement has virtually disappeared from the
perfect, but not from the passive. There are only two unambiguous perfects
that still retain agreement, both of them from the early part of the period, the
890s.21 The first one is already familiar from the literature review in Chapter 3:
20 In line with the remarks on style, note also the non-use of a passive and the preference
instead for an impersonal construction with mon ‘one’ (I return to this issue later in this
section).
21 It should be stressed that overt agreement is lost equally early, however (observe the dates
of the entries in this section).
112 chapter 4
As already noted in the earlier discussion, this example supplies solid confir-
mation that overt agreement is not incompatible with a perfect reading. Both
objects are masculine, accusative and singular, and both participles indicate
overt masculine accusative singular agreement. However, the term of service is
already up, i.e. there is none left, and neither do they have any food, so the only
plausible reading is (at least partially) perfect, with bleached lexical seman-
tics. (37) stresses the resultativeness of the events and may or may not further
indicate anteriority – i.e. it might simply mean that at that past moment the
service was up/complete and the supplies were depleted, without bringing to
the fore the temporal precedence. The comparable PP adjunct in (38) below
suggests that the giving of oaths had a result relevant ‘in this year’, i.e. similar
resultativeness without emphasis on anteriority. (37) was the only agreeing
have-perfect. In ten attestations, overt agreement was expected but was not
there. This is especially noticeable in the plural, as in (38) and (39).
The semantics here is again doubtless perfect. If you give someone your hostag-
es, you no longer have them, but as we saw above, this perfect reading does not
have to preclude overt agreement. In (38), geseald, with its plural morpheme -e
missing, coexists in the same sentence with another zero-marked participle,
geworht, whose zero morphology is however legitimate (see below). According
to the traditional hypothesis proposed in Chapter 3, this might have been
the reason why agreement was ultimately lost from the perfect. ‘Bare’ forms
sneaked in in contexts like this. However, the picture is actually more complex
when different facets of it are examined. Below is another example of missing
overt agreement:
It is important to point out that agreement and lack thereof can be found in
very close proximity, as in (44):
This sentence also involves ‘provisions’ being eaten and therefore impossible to
literally have in one’s possession (this time the unfortunate horses). The pred-
icative participle gewægde in the first clause does agree with its subject, and
the attributive adjectival modifier miclne shows accusative masculine singular
agreement with the head noun dæl. One might expect to find the same -ne
morpheme on freten (as in the analogous example in (37) above). Coupled
with the semantics of the sentence, the lack of agreement in (44) might thus
testify to the more advanced grammaticalisation of what is already morphing
into a perfect, in comparison with the predicative construction from the first
clause, and also compared to proper adjectival modifiers like attributive mi-
clne, to which the participle used to be more similar before it ceased to agree.
22 It should additionally be noted that there is no perfect passive in this context yet. Due to
the meaning of the verb gesamnod ‘collected, assembled’, this sentence can be perfect or
passive.
the Perfect in a Selection of Old English Texts 115
The presence of the clear and unambiguous -ne ending on the attributive mod-
ifier invites the conclusion that the agreement was lost due to the more highly
grammaticalised nature of the perfect, not because people simply forgot the
rules and markers of agreement, and perhaps not so much because of zero
morphology or ambiguity elsewhere. As is plain to see, this speaker (writer)
obviously still uses the -ne suffix, and also extensive plural agreement, but not
in the (proto-)perfect. At this stage, it should be reiterated that the absence of
agreement is especially striking in the plural:
All of these examples could technically have a marker for plurality on the parti-
ciple but they do not. The first one, (45), is the clearest: the object is accusative
masculine plural; morphologically, the form ærendracan cannot be genitive
plural (terminating in -a), despite the preceding numeral which might be used
with a partitive genitive (cf. scipa in (46)); with a great degree of confidence,
asend can thus be expected to receive a plural -e in order to match acc.pl
ærendracan. In (46), agreement might have defaulted because this time scipa
is certainly in the genitive plural after the numeral. In (47), the invariable rel-
ativiser þe is ‘entitled’ to pick up the masculine gender and the plural num-
ber of the antecedent wicinga, and the participle aspanen might have copied
those features (see below for such examples). And yet, the adjectival agree-
ment markers are not there, most likely pointing to the verbal status of those
participles.
116 chapter 4
Accusative neuter plural -u is expected only with short/light stems, but this
one, befæst, is long/heavy due to the consonant cluster, so zero is customary for
it anyway (at least in attested, as opposed to prehistoric/reconstructed, OE).
In addition, the zero plural of heavy neuter a-stems like wīf might have further
reinforced zero marking (as noted above). And yet, this is not all there is to it,
as becomes evident from (51):
In (51), eall is also a heavy stem, but it is marked for the plural with -e.23 True,
this is not the ‘standard’ accusative neuter plural -u, but it is plural nevertheless
(unless it is an adverb, which seems unlikely). The heavy-stemmed forworht,
by contrast, is not inflected, so it cannot be just the type of stem or analogy on
zero-marked plural nouns that matters (cf. the -u on geatu in (51), as opposed
to zero-marked plural wīf in (50)). Things are more complex and it is not easy to
23 If the double consonant had by then been degeminated, eall might perhaps have been
treated as a light syllable.
118 chapter 4
say why ealle is inflected in (51), but forworht is not, since both of them contain
a consonant cluster. It is perhaps reasonable to suggest that this is not to do
with the influence of ‘expected zero morphology’, because these expectations
seem to have been overridden. It appears that ealle agrees in (51), despite its
heavy stem, because of the type of construction it is in, i.e. it is in an attribu-
tive plural environment. If it attributively modifies a neuter singular noun/pro-
noun, as in (52)–(53) below, it remains (legitimately!) uninflected, but this has
not prevented it from agreeing in the plural, and neither has its stem type. By
contrast, forworht in (51) might be uninflected due to its predicative position
and/or due to the fact that grammaticalising the perfect was already underway.
Other reasons to expect no overt agreement are clausal objects, (54), or no ob-
jects at all, (55), both of which are attested (once each) in the sample:
24 Land in (52) could conceivably be a zero plural as well, and eall might serve as the head of
the NP in (53).
the Perfect in a Selection of Old English Texts 119
(55) shows that have can already occur with an intransitive verb (gewicod).
No agreement can be triggered here because there is no accusative object. It is
also curious that this appears in the same sentence as a zero-inflected transi-
tive past participle (ongunnen in the context of geweorc).
The situation of have-perfects is in some ways similar to that of the other
perfect (proto-)auxiliaries in OE – beon and weorðan. Only one instance of
agreement (in the plural) was recorded, given in (56).
Here too, agreement is much more frequently missing from the perfect, as
in (57), indicating that grammaticalisation must have been underway.
Once more, it should be noted that there is agreement on attributive oþre, but
not on the participle (as was the case with the have-perfect).25 Another ques-
tion that arises is whether it is possible to give (57) a passive or stative interpre-
tation. Judging by Bosworth and Toller’s dictionary entry and examples quoted
below, the strong verb a-cwelan (pptcp a-cwolen/a-cwelen) is intransitive,
meaning ‘to die, perish’, a translation equivalent of Latin mori ‘die’ (a deponent
verb, i.e. passive in form but active in meaning):
25
O E oþer is always declined strong, even after a demonstrative (see Mitchell 1985: §508, §568,
and below). It is, strictly speaking, used (semi-)independently here, without a following
noun.
120 chapter 4
No transitive uses were found in Bosworth and Toller, and (60) is an active
intransitive clause in the preterite with a structure otherwise similar to that
of (57), so (57) must be a perfect. Mitchell (1985: §795) supplies the follow-
ing potentially passive example: … mid hungre wurdon acwealde ‘(lit.) with
hunger became dead/killed’, where the participle is weak and belongs to a
separate lexeme. There is indeed a transitive verb a-cwellan, ‘to kill’, which is
weak, with a past tense a-cwealde, and past participle a-cweald (see http://
bosworth.ff.cuni.cz/000223, accessed on 14 Jan 2017). Thus, pairs of verbal
lexemes (strong intransitives alongside weak transitives) serve to differentiate
between be-perfects and passives.
There was another interesting be-perfect for which agreement had failed:
As the translation indicates, the original does not necessarily possess the
meaning of ‘past in the past’ – it may still be viewed as preterite be with a
participial subject complement (i.e. an adjective), as it were painting a picture
of what could be seen at that past moment. Yet again, the attributive adjective
Deniscan agrees (in the weak declension), but the predicative participle aseten
does not bear nom.nt.pl -u (this state of affairs might be similar to what
we find in Modern German, for which see Eisenberg 1994: 366). Crucially, this
sentence from the Chronicle goes on as follows:
the Perfect in a Selection of Old English Texts 121
This is the entry from Manuscript A. Manuscript D of the same text, how-
ever, has gefaren hæfde (see Plummer and Earle 1892: 87 fn. 1; cf. Mitchell
1985: §740). Significantly, MS. D is another representative of the so called
26 Here I give a taste of the intricacies of classification and analysis, this time regarding faran
‘travel’. Bosworth and Toller list the following forms and definititions for the instransi-
tive strong verb faran (to farenne; ic fare, ðu farest/færest/færst/færsþ, he fareþ/færeþ/færþ,
present plural faraþ; past singular for, past plural foron; past participle faren): “A word
expressing every kind of going from one place to another, hence ‘to go, proceed, travel,
march, sail; īre, vādĕre, incēdĕre, transīre, migrāre, nāvīgāre’ (http://bosworth.ff.cuni.
cz/010118, accessed on 17 November 2016)”
By contrast, the weak counterpart feran has transitive (alongside intransitive) uses, as
cited in Bosworth and Toller (cf. also Mitchell 1985: §735):
feran, to ferenne; past singular ferde, past plural ferdon; past participle fered [ fer ‘a
journey’]
‘to go, make a journey, set out, travel, march, sail; īre, ĭter făcĕre, proficisci, transīre,
migrāre, nāvĭgāre’
A transitive collocation involves mycelne siþfæt feran ‘go so great a journey’ (http://bo-
sworth.ff.cuni.cz/010498, accessed on 17 November 2016). Finally, there is a third entry
from Bosworth and Toller (cf. Mitchell 1985: §735): ge-fered (adj./ptcp.) ‘associated, band-
ed together’ (http://bosworth.ff.cuni.cz/047936, accessed on 17 November 2016).
According to Mitchell (1985: §781), geferede can be active or passive in Beowulf 361 Her
syndon geferede, feorran cumene … ‘(lit.) Here are/have? travelled, far come …’, though
it is hard for me to get how it can be passive. Then Mitchell goes on to supply an ex-
ample of intransitive and transitive gefaren too, so figuring out valency is often far from
straightforward:
(i) he hæfde on þæm emnete gefaren
‘he had travelled on the level ground’ [intransitive, Orosius 186.22, quoted in Mitchell
(1985: §781), translation mine]
(ii) Ða he ða þreo burga gefaren hæfde … [þa þreo burga as a quasi-adverbial?]
‘When he had travelled/visited the three cities’ [transitive, LS 34.19, quoted in
Mitchell (1985: §871), translation mine]
the Perfect in a Selection of Old English Texts 123
Probably due to the stylistic preferences of the new scribes from this period,
(proto-)progressives, numerous in the previous century, are now not used at
all. Similarly, passives proper are much rarer too. Instead, those scribes seem to
favour the formally active construction with indefinite/impersonal man/mon
‘one’, (67) (cf. Irvine 2004: clviii; see also examples earlier in this section; the
uses of OE man/mon are dealt with in Mitchell 1985: §§363–377, §747).
As mentioned above, despite the low numbers, the agreement trends from the
earlier entries are in evidence in this section too. Two be-passives show plural
agreement, one of them is given in (68):
The same scenario is replicated with passive weorðan. The singulars (a total
of five) have the expected zero marking, as in (72), and the plurals (a total of
three) take overt -e, as in (73)–(74), with no defaults in overt exponence.
27 This type of clause was treated as a dynamic passive, as opposed to a copular stative, be-
cause of the meaning contribution of the adjunct of place, either a PP or an AdvP. In (70),
this is unlikely to be the (potentially) expletive þær which was encountered above in the
context of wæl – see esp. example (25) in Section 4.4.2.3.
126 chapter 4
Finally, there were many more cases of what were judged to be ambiguous pas-
sives or statives. This type of ambiguity remains rife throughout the periods
the Perfect in a Selection of Old English Texts 127
studied. There were three agreement categories for those sentences too: five
of them showed overt agreement, all in the plural; zero morphology was nor-
mal for another five, which had masculine or neuter singular subjects; there
were also three breakdowns in agreement (see Appendix 2.128 and Table 3b).
However, all three of them contain feminine singular subjects, as in (77).
28 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
29 On the non-occurrence of the perfect passive in such contexts, see Mitchell (1985: §§782–
785), the discussion above, and Chapter 3.
128 chapter 4
marking, while zero is found in all the other nominative slots. By contrast, zero
is legitimate only with neuters in the accusative singular, and only with heavy
neuters in the nominative/accusative plural of the same paradigm.
For both have- and be-perfects, there are many examples where zero mor-
phology is indeed expected – the overwhelming majority with be-perfects, and
in the case of have-perfects, as many as the examples in which overt morphol-
ogy should be present but is nevertheless missing. However, if that is believed
to be the reason why agreement disappeared across the board in the perfect
(and this sample does demonstrate virtually no agreement at all, with two ex-
ceptional survivors), some questions still await answers. In a single sentence,
the Perfect in a Selection of Old English Texts 129
In both copulars and proto-passives there is an overt copula linking the subject
and the (original) subject complement, indicating the relationship between
them, whereas the link between an object and an object complement (as in the
proto-have-perfect) is not overtly expressed, making it easier to reanalyse the
object complement as a non-agreeing verbal element. In other words, the pas-
sive makes explicit the relationship between the active object and the comple-
ment participle, which is implicit in SVOC O structures, including those with
have – cf. I had him bound v. He was bound (by me). Admittedly, this scenario
is also based on a kind of marking (instantiated in the overt/covert copula),
but this kind of marking is much more ubiquitous and with arguably more far-
reaching consequences than non-explicit morphology in certain slots of the
paradigm. Thus, zero morphology and its impact appear to have been over-
rated in the specialist literature and there are other plausible reconstructions
of how things must have proceeded.
As pointed out above, the confusion of endings is evident from cases like to
scipon, which should actually be to scipum (dative plural). However, this late
entry contains overt accusative feminine singular agreement on the participle
asmeade, which might be deemed strange and unexpected, especially in the
singular, considering the date of this text. Perhaps the -e was somehow still felt
to be a more distinctive ending, and ealle might suggest adjectival status for
the participle too, which arguably increases the likelihood of agreement. Ealle
could be an adjective or perhaps an adverb (see the previous section and the
conclusions in Chapter 9). In addition, burh is a zero-marked feminine singular
noun, but there is an -e on the adjective/participle, so any purported analogy
on zero-marked nouns has failed to operate here (cf. Section 4.4.2.2, and the
opposite scenario in (79) – -e on the noun and zero on the participle).
The partially conservative character of the manuscript might have to do with
the fact that this kind of language was probably transmitted as a learned writ-
ten register influenced by earlier exemplars rather than being a pure reflection
of the naturally acquired spoken language. Or rather, the two were most likely
mixed – the earlier written documents serving as conservative bookish models
and the actual spoken language at the time (see Irvine 2004 for textual trans-
mission). Still, the consistency in agreement patterns is striking, even more so
against the background of general ‘confusion’ elsewhere.
Not surprisingly, cases in which agreement has failed in the have-perfect
are more numerous (as before) – four of the six transitive have-perfects here
are not inflected where they could have been. In (79), the accusative feminine
singular -e is missing from the predicative participle gemarcod, but it is pres-
ent on the attributive modifier ælce, perhaps suggesting that, unlike ælce (and
maybe asmeade above), gemarcod was not felt to be a modifier related to the
direct object to the same extent.
the Perfect in a Selection of Old English Texts 133
The antecedent of the invariable relative particle þe is the ‘churl Hugh’, so gesett
could legitimately have hosted the accusative masculine singular ending -ne.
But it does not. Interestingly enough, a little later in the same entry, an adjecti-
val object complement does agree in the accusative masculine singular:
Therefore, it should be concluded that the scribe or author fails to use the
agreement markers in the perfect because he does not feel that those parti-
ciples modify the object, not because he has forgotten the appropriate mor-
phemes, since he uses them in comparable constructions elsewhere.
The most striking departure from the state of affairs in the passive is that
perfects do not host plural agreement in this sample either (though see Section
4.4.2.2 above for an example from MS. E, 1086; cf. Kilpiö 2007: 332). In (83) and
(84), the objects are plural but the participles have no overt plural morphology
(acc.pl -e). The word order is also noteworthy.
(84) Hi heafd-on þa ofer-gan East Engla .i. and East Seaxe .ii.
they had-pl then pref-gone East Angles and East Saxons
7 Middel Seaxe .iii. 7 Oxenafordscire .iiii. … [1011]
and Middle Saxons and Oxfordshire
‘They had then overrun: (i) East Anglia, (ii) Essex, (iii) Middlesex, (iv)
Oxfordshire …’
Even if the objects in (84) were all singular, there could still have been plural
agreement because of the list of conjoined place names. In a passive clause
below, (87), a conjunction of two singular nouns indeed triggers resolved plu-
ral agreement. The marking on East Engla is not clear, so the zero morphology
on the participle might be legitimate agreement with the nearest conjunct if
East Engla was treated as a neuter singular perhaps, or this could be a geni-
tive plural noun (characteristically terminating in -a), though there is hardly
any justification for having a genitive here. East Engla is far more likely to be
an accusative plural direct object, like those with the component Seaxe, and
the Perfect in a Selection of Old English Texts 135
agreement has simply broken down. The following example from Bosworth
and Toller, again from the Chronicle, including MS. E, suggests that both types
of name are actually grammatically plural, triggering plural verb agreement
(on plural names like Seaxe, see Prokosch 2009[1939]: 246):
Whatever the correct analysis may be, it is important that there could have
been plural agreement in (84), even if all the enumerated objects were sin-
gular. By contrast, zero morphology is legitimate in only one transitive have-
perfect, (86). Though legitimate zero morphology is widespread in passives,
plural agreement is still healthy there, so the loss of plural agreement in the
perfect cannot be attributed to this lone instance of expected lack of overt
marking.
There is arguably no real accusative object here, at least not a noun; mæst
might be taken to be a genderless object (or an adverb serving as adjunct, or
a determiner used independently in the neuter singular), so agreement has
defaulted legitimately anyway. As an aside, an alternative modern translation
might involve did in place of had done, laying no emphasis on the anteriority
and acting as a reminder that the perfect has not yet attained its modern state.
As before, passives show robust agreement in the plural and zero in the sin-
gular. The examples in (87) and (88) below demonstrate that zero morphology
and overt agreement can coexist happily side by side in the same sentence.
Moreover, they confirm that overt marking survives until quite late, as late as
1012.
136 chapter 4
It should be reiterated that, in the passive, agreement (again!) only fails in the
feminine singular (expected nom.f.sg -u), never in the plural:
30 Unlike some of those above, this example is treated as a dynamic passive because of the
meaning contribution of the PP adjunct of time.
the Perfect in a Selection of Old English Texts 137
31 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
32 As already noted above, some researchers argue that analogous forms, such as wearð ges-
icclod ‘became sickened’, could be potentially adjectival (see Mitchell 1985: §735, and ex-
ample (97) below).
138 chapter 4
Things are essentially the same with examples judged (admittedly, sometimes
arbitrarily) to be ambiguous between passive and stative. (95) once more
shows plural agreement (on druncene) and expected zero morphology (on sin-
gular gebroht) side by side in the same sentence. Therefore, it can be concluded
that the agreement system which operated in the previous annals is still intact
at this late date, ‘oblivious’ to any influence coming from legitimate zero mark-
ing or to other large-scale changes in the morphology of Late Old English.
Adjectival status for druncene is suggested by the modifier swiðe and by the
argument structure of the verb – (95) does not come from ‘X drank them’. The
same reasoning applies to the combination swiðe gehindred ‘much hindered’
from another example for 1003 (see Appendix 2.2).34 Swiðe likewise points
to adjectival status for the participle in the only combination with weorðan
which was judged to be ambiguous:
33 The adverbial clause of reason which contains gebroht is treated as a bona fide passive. It
remains to determine whether the presence or absence of the perfective prefix ge- on the
participles is significant.
34 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
the Perfect in a Selection of Old English Texts 139
Due to another type of cue, this time lexicographic, the example in (97) below
was deemed ambiguously stative too. Bosworth and Toller’s dictionary entry
gives both a dynamic and a stative translation: ge-siclian (past -ode, past
participle -od) is defined either as ‘to be taken sick or ill’ (dynamic), or ‘to be
infirm’ (stative) (http://bosworth.ff.cuni.cz/016005, accessed on 20 Oct 2016).35
The clause in (98), by contrast, was assigned to this category a little more
arbitrarily:
35 Cf. the note above regarding the treatment of wearð gesicclod ‘became sickened’ as adjec-
tival in Mitchell (1985: §735).
140 chapter 4
The first of the examples with ‘licensed’ zero morphology is especially curious:
Indeed, (104) probably features the intransitive perfect use of this verb, as
suggested by Whitelock et al.’s (1961) translation and since this is an example
Bosworth and Toller cite to illustrate intransitive ge-wendan. The usage of wen-
don ut in (103) also favours a perfect interpretation, but the potential for am-
biguity is still there nevertheless (despite the undeniable disambiguation in
context). The final sentence, coupled with the other ones adduced here, might
suggest that, after all, OE morphology did not keep ambiguity between perfect
and passive readings at bay as much as has been claimed.
36 Alternatively, it could be argued that unfriðflota got transferred to the feminine gender
class because of influence from semantically related feminines like scipfyrd. This explana-
tion is less attractive.
the Perfect in a Selection of Old English Texts 143
37 The absence of unequivocal be-perfects from this stretch of MS. E is probably just a fluke.
It would be illuminating to draw a parallel to ‘proto’-progressives. More plentiful in the
early entries, progressives were absent from the previous sample. However, they have now
made a comeback (there are three of them in the annals under investigation). So their
absence from the earlier annals must be a coincidence and can scarcely be taken to indi-
cate that the progressive had disappeared. The same applies to the observed fluctuations
between bona fide passives and competing constructions with man/mon ‘one’. Thus it is
not advisable to make too much of the non-occurrence of clear be-perfects here.
144 chapter 4
The explanation might actually lie elsewhere. Van Gelderen (2018: 97) notes
that ‘the language of the Peterborough Chronicle is traditionally seen as repre-
senting the change from Old to Middle English and is often seen as represen-
tative of changes contact with Scandinavian brought’ – see for instance (86),
where the preposition wið is already used with its Old-Norse-influenced sense
of ‘with’, rather than the earlier ‘against’ (cf. (96) and the earlier annals). It is
insightful to note that both the Peterborough Chronicle and Secunda Pastorum
are of Northern provenance, as is MS. D discussed in the previous section. All
of them exhibit displacement of perfect be by have, so the advanced loss of
perfect be might in fact be associated with contact with Old Norse speakers
in the north of England, as suggested in Chapter 3, Section 3.5.5. What these
Chronicle entries reveal might be the earliest seeds of this development. In ad-
dition to documenting the Viking raids and the ensuing clashes and atrocities,
this section of the Chronicle could also be a testament to the Viking influence
on the incipient retreat of be from the perfect. This is probably a contact-
induced change whereby speakers started to rely on the default auxiliary in
bilingual interactions, reinforced by the greater frequency of have-perfects
in the language of the Scandinavian invaders. This process is reflected in the
consistently low numbers of perfect be in my northern samples, with reliance
placed on have instead, and will be further traced in the Middle English survey
to follow in the next chapter.
4.5 Conclusion
For instance, the stative meaning of possession does not rule out absence of
overt agreement, and conversely, a dynamic perfect reading does not rule out
presence of agreement. And yet, there is a robust tendency throughout, even in
the late entries already on the cusp between Old and Middle English, for agree-
ment to appear more in non-grammaticalised adjectival constructions, often
side by side with non-agreeing perfects. I therefore believe that, rather than
acting as a trigger for grammaticalisation and reanalysis, disposing of agree-
ment is an outward sign and a consequence of ongoing grammaticalisation.
Scribes do not appear to have simply forgotten the requisite endings – instead,
they use them surprisingly consistently in constructions which are perceived
as genuine modifiers of nouns and drop them in perfects which they arguably
regard as more verbal in character.
One of the objectives was to test empirically the role of legitimate zero ex-
ponence as a bridging context which supposedly enabled reanalysis and sub-
sequent loss of agreement across the board. Both have- and be-perfects were
examined and comparison was sought to passive or potentially stative con-
structions with beon ‘be’ and weorðan ‘become’. The material was divided into
three categories – examples where overt agreement is present, examples where
overt agreement is missing, examples where zero marking is expected anyway.
I confirm that there is a substantial amount of expected zero morphology
in the perfect, as well as in beon- and weorðan-passives. However, while the
prevalence of expected zero exponence may appear to have led to an almost
complete loss of agreement in the perfect, particularly evident in the plural,
this is certainly not the case in the passive. In both types of passive, with beon
and weorðan, zero morphology is used in the singular (legitimately with the
masculine and neuter, not so with the feminine), whereas -e appears consis-
tently in the plural (unlike the perfect). These findings call into question the
claims that ambiguous cases of expected zero morphology were responsible
for losing agreement across the board, and point to different rates at which the
grammaticalisation of these constructions proceeded.
As for the ambivalence between be-perfects and passives, a characteristic
of OE is that its verbal derivational morphology can often (but not always)
disambiguate a given verb as transitive or intransitive, unlike later periods,
where context and semantics assume a more important role. The rates of am-
biguity between perfect, passive and stative or between passive and stative are
for the most part comparable to those from Middle and Modern English (see
Chapter 5), though a tendency for the former to gradually creep up with the
gradual spread of the perfect makes itself felt. These issues will be addressed
again in the next chapter, as will be the clues for potential Scandinavian in-
fluence on the loss of perfect be discernible in the northern recension of the
Chronicle.
chapter 5
Picking up where the previous chapter left off, Chapter 5 charts the further
development of the perfect from Middle into Modern English. As agreement
on the participle is no longer at issue, the focus is on the competition between
the two auxiliaries and the role ambiguity had to play in the displacement of
be, as compared to some of the other explanations outlined in Chapters 3 and
4. The material comes from some of Chaucer’s comic tales, a 15th-century mys-
tery play, comedy from the Renaissance and the Restoration period, as well as
the correspondence of the Paston family and Samuel Pepys. The chapter con-
cludes with an epilogue featuring commentary on Victorian English, and the
next chapter draws some parallels to German.
5.1 Introduction
My main concern here will be to concentrate on the ambivalence and the inter-
actions of the various auxiliaries, also trying to find more evidence for a contact
scenario. The diverse selection of passages provides a series of vivid snapshots
of how the perfect construction is gradually edging towards the state at which
it finds itself today, in addition to being an eloquent testimony to the multiple
waves of grammaticalisation discussed in Chapter 3. Section 5.2 examines ma-
terial from the 14th century (Chaucer), Section 5.3 is about a 15th-century play
from the north which might confirm the influence of Scandinavian, Section
5.4 tackles a 16th-century play by Shakespeare, while Section 5.5 analyses
Restoration comedy from the 17th century. To bolster the database, in Section
5.6 I supply data from the Late Middle English correspondence of the Paston
family, while in Section 5.7 I do the same for the 17th-century correspondence
of Samuel Pepys. The Epilogue (5.8) adduces material as late as Dickens in
order to consolidate the discussion of English and to convey one final time
the message that the functionalist story from earlier chapters crumbles when
subjected to close scrutiny. The changes involved must have been much more
mechanistic and/or contact-driven than driven by functional considerations.
The tales are often quite suited to their tellers and they seem like dramatic
soliloquies, which makes them similar to comedy on the stage. Benson (1987a:
7) sums up the fabliau style as simple, vigorous and straightforward; fabliaux
deal with the present time, relying on realistic settings in familiar places; the
148 chapter 5
table 1 Perfect, passive and modal auxiliaries in the Miller’s and Reeve’s Prologues and
Tales
(1) Whan that the Knyght had thus his tale ytoold [3109]
Here I also included typically transitive verbs with implied objects which were
not overtly expressed, (2),1 as well as verbs with prepositional complements
(a.k.a. prepositional objects), (3), and the copula be, (4).
As noted by Mustanoja (1960: 501), the verb be has always formed its pres-
ent and past perfect by means of the auxiliary have (cf. Mitchell 1985: §1099
and Macleod 2012: 120–122 for OE; an example involving heafde gebeon from
the Anglo-Saxon Chronicle entry for 1096 is provided in the Dictionary of Old
English, at http://tapor.library.utoronto.ca/doe, accessed on 27 Feb 2017).
A question arises whether instances in which the auxiliary and the main
verb are separated might not be carriers of stative meaning, rather than being
genuine perfects. It was concluded that they were simply ME perfects (though
see discussion and caveats in Chapter 3, esp. Sections 3.1.2, 3.1.4, and Chapter
4, esp. Section 4.4.2.1). The free word order in the perfect, including different
permutations of the relative positions of the auxiliary, the main verb and the
object(s), as in (5)–(8), is most likely due to the poetic form and the less strict
word-order constraints during this period, less strict than ModE that is (cf.
Kilpiö 2007: 335–336 for the flexibility of the word order in perfects from ME
prose texts too). Moreover, the same flexibility is evident in the passive, (9),
where adjacency of the finite and non-finite verb is certainly not a must.
(5) He auntred hym [‘took a risk’], and has his nedes sped [‘accomplished his
purpose’] [4205]
(6) This joly lyf han [‘have’] thise two clerkes lad
Til that the thridde cok bigan to synge [4232–4233]
(9) Ther was hir whete and eek [‘also’] hir malt ygrounde [3991]
In general, there are now much clearer cues of anteriority such as til ‘till’, (6),
or after, (10), compared with the ubiquity of clauses with þa ‘then/when’ in the
Anglo-Saxon Chronicle, which do not lay particular emphasis on anteriority. By
Middle English, the semantics of the perfect has edged closer to what we cur-
rently have in Modern English. The more abstract meaning of modality/coun-
terfactuality, illustrated in (12)–(13), further testifies to even more advanced
grammaticalisation.
A substantial number of intransitives – 14 – formed their perfect with have.
Those included verbs like crow, itch, run, misgo, creep, wake (‘stay awake’). Gräf
(1888: 69) confirms that misgo, for instance, is only used with have in Chaucer’s
work and I return to the likes of wake below. The following are typical exam-
ples of intransitive have-perfects:
It is striking that the intransitives that use be and have are normally in com-
plementary distribution. This is consistent with the constructionist view (see
Chapters 2, 3 and 4), and a similar observation regarding other aspects of gram-
mar is also made by Danchev (1969). Each verb prefers one or the other aux-
iliary, but does not tend to occur with both. Run, for instance, prefers have,
whereas go (but not misgo ‘go astray’) generally takes be. In the only case where
go combines with have in my sample, have seems to indicate a hypothetical
context, suggesting that the action was not accomplished (see discussion in
Chapter 3):
In at least one example, (13), have was probably used with an intransitive due
to the non-finite (and hypothetical/counter-factual?) environment (the read-
er is again referred to the preceding discussion, esp. Chapter 3, Sections 3.3
and 3.4.1).
The perfect from Middle into Modern English 151
(14) A litel after cokkes hadde ycrowe [or ‘were(n) ycrowe’?] [3358]
(15) Thise sely [‘innocent, hapless, foolish’] clerkes han [‘have’] ful faste yronne
[or ‘are(n)/be(e)n yronne’]
Toward the fen, bothe Aleyn and eek [‘also’] John. [4090–4091]
(16) I hadde almoost mysgoon [‘gone astray’] [or ‘was … mysgoon’]; [4218]
(18) And ever sithe hath so the tappe yronne [‘is… yronne’ potentially passive?],
Til that almoost al empty is the tonne [‘barrel, cask’] [3893–3894]
What might have disambiguated this as non-passive is the syntactic and se-
mantic context, with ever sithe ‘ever since’ favouring a perfect interpretation.
In any case, out of 14 examples of have used with intransitives, substituting
be for have would have created potential misunderstanding in only three or
four at most. Apart from run in (18), such might be the case with had waked
2 Judging by the OED and the Online Middle English Dictionary entries, the verb for ‘crow’ only
takes have in the perfect (https://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=
31869142&egdisplay=compact&egs=31875371&egs=31876210, accessed on 10 Jan 2017). By con-
trast, there are attested combinations of ‘be’ and ‘run’: see Gräf (1888: 69) for examples from
Chaucer, as well as phrases like ben runnen ‘to have run or gone’ from the Middle English
Dictionary (https://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=159568672&e
gdisplay=compact&egs=159590540, accessd on 10 Jan 2017).
152 chapter 5
‘had stayed awake’ in (19) – if it was not for the adjunct al nyght ‘all night’, was
waked could conceivably have been interpreted as the ambivalent ‘was awo-
ken/woken up’ (Rydén and Brorstöm 1987: 49 provide later examples of be- and
have-perfects with related (a)wake(n)). However, the adjunct and the context
would have supplied ample disambiguation anyway and there are moreover
clear signs that had here does not serve any disambiguation purposes at all.
Contrary to what one would expect if disambiguation was paramount, Gräf
(1888: 65) notes that when ME wake means ‘awake, wake up’, it combines ex-
clusively with be in Chaucer’s oeuvre to form its perfect (But whan that he was
waked out of his sleep), while it takes have in the non-mutative sense of ‘stay
awake’, as in (19).
(19) For she was falle aslepe a lite wight [‘a short time’]
With John the clerk, that waked hadde al nyght [4283–4284]
A third context with potential for confusion might involve the verb for ‘work’,
as in (20).
3 It is interesting to observe that Chaucer’s representation of a Northern dialect features have-
as well as be-perfects with intransitives, among the latter (i) and (ii). This state of affairs
should be compared to the texts of actual Northern provenance – the Peterborough Chronicle
and Secunda Pastorum. Inconsistencies in Chaucer’s use of Northern dialectal forms in this
tale are noted in Crystal (2004: 165–167).
(i) And forthy is I come, and eek [‘also’] Alayn [4031]
(ii) What, whilk [‘which’, Northern dialect] way is he geen [‘gone’, Northern dialect]? [4078]
The perfect from Middle into Modern English 153
Apart from fall with a spatial sense where it indicates motion, it turns up in
a copular sense with the be-perfect too, as in (23) (cf. Chapter 2, Section 2.2,
for remarks on pathways of pragmatic inferencing during grammaticalisation
which lead to greater abstraction).
(23) For she was falle aslepe a lite wight [‘a short time’]
With John the clerk, that waked hadde al nyght [4283–4284]
Similarly, (25) was interpreted as a clear be-perfect, not a passive (cf. Gräf 1888:
76). What disambiguates it is the temporal subordinate clause (with syn ‘since’)
and the adverb meaning ‘hence’, which provide a typical perfect context, as
confirmed in (26) and in the earlier discussion.
(27) is also clearly not a transitive grow, so this copular use of grow can form
nothing but a be-perfect (or act as an adjective, cf. Gräf 1888: 64):
154 chapter 5
(27) This wenche [‘girl’] thikke and wel ygrowen was [3973]
4 There is reasonably solid evidence that (28) is perfect: firstly, have is also attested in this con-
text, as in (i); secondly, the adjectival construction goes with a nominative subject, as in (ii).
And yet, an impersonal adjectival construction with an accusative experiencer, creative word
play and departure from general norms cannot be ruled out a priori (Gräf 1888: 77 indeed
treats wo-bigon as mostly adjectival, including in (28)).
(i) Gyle haþ bi-gon hire so, heo graunteþ al his wille. [(c1390) Chaucer CT.ML.(Manly-Rickert),
https://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=14857611&egdisplay
=compact&egs=14863637&egs=14862294&egs=14870945, accessed on 10 Jan 2017]
(ii) That sche was wo bego withal. [(c1395) Chaucer CT.WB.(Manly-Rickert), https://quod
.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=14857611&egdisplay=compact&
egs=14863637&egs=14862294&egs=14870945, accessed on 10 Jan 2017]
5 The Appendices of this book have been made available online. They can be accessed through
a QR code which can be found in the References section of this book.
The perfect from Middle into Modern English 155
(31) and (32) show coordination with an adjective and a noun phrase,
respectively:
(31) His hoote love was coold and al yqueynt [‘quenched’] [3754]
It is worth asking whether a perfect reading might not be present in the fol-
lowing type of construction, treated as ambiguously stative or passive, but not
perfect:
Crucially, (33) is different from, say, I’m done/finished, where the subject is an
animate agent and where the sense is ‘I have done/finished [whatever I had
to do/finish]’. This, however, is not the case with This tale is done, where *This
tale has done would be ungrammatical, or rather #This tale has done [some-
thing] is semantically/pragmatically anomalous (cf. however, This tale/game
is/has begun, to which I return below). Certainly, even at a conservative es-
timate, and even if other scholars dispute some of the choices made as to
category-assignment, the current sample demonstrates that there is much
156 chapter 5
more ambiguity between a passive and a stative reading than there is between
a passive and a be-perfect.
Before moving on to what was considered genuine ambiguities between
passive and be-perfect, several additional borderline cases should be men-
tioned. (34) is regarded as passive (or potentially stative) (cf. Gräf 1888: 63 for
similar intuitions):
While thy wit has overcome [something] would be grammatically and semanti-
cally well-formed, it simply makes no sense in this context, where the person
referred to as a ‘fool’ has (ostensibly) ‘taken leave of their senses’. (35) was like-
wise treated as passive (though some scholars might suggest otherwise, e.g.
Gräf 1888: 72–73; also see the Middle English Dictionary entry, which indirectly
corroborates my analysis, https://quod.lib.umich.edu/cgi/m/mec/med-idx?ty
pe=byte&byte=46261898&egdisplay=compact&egs=46337024&egs=46277303,
accessed on 16 Jan 2018):
(35) Now are we dryve til hethyng [‘to contempt’] and til scorn. [4110]
The context indicates passive here, because the two students have been fooled
by the cunning miller; i.e. they do not get themselves into that mess; rather, it
is the miller that drives them to contempt and to scorn. In addition, though not
an argument-clincher, a parallel passive construction appears in the following
line (repeated from (29) above):
(37) The sely [‘innocent, foolish’] tonge may wel rynge and chymbe [‘chime’]
Of wrecchednesse [‘wretchedness’] that passed is ful yoore [‘long ago’].
[3896–3897]
Section 3.1.4, Gräf 1888: 20ff.).6 Mustanoja (1960: 504) confirms that in Middle
English, the perfect could occur with adverbs of past time, as in adventures that
whilom han bifalle from the General Prologue to the Canterbury Tales. In (30)
above, the act of entering (ben entred ‘are/have entered’) took place on that ilke
nyght ‘on that same night’. The following analogous case from my corpus also
shows that an indisputable have-perfect (hath take) can certainly be used with
specific past time reference, as indicated by the ‘when’ temporal subordinate
clause:
(38) And whan the millere saugh [‘saw’] that they were gon
He half a busshel of hir flour hath take [4092–4093]
Be-perfects were compatible with specific past time reference as late as the
17th and 18th centuries, as shown in the following example from Pepys’s Diary
(cf. Rydén and Brorström 1987: 162 for late-18th-century attestations).7
(39) This day I have news ... that the fleet is sailed yesterday from Harwich
to the coast of Holland, to see what the Duch will do. [Samuel Pepys, Diary,
22 April 1665, quoted in Nakamura (1987: 40)]
The remaining ambiguous cases, which could be perfects or passives, are then
relatively few – initially, there appear to be six of them, as noted in Table 1. The
following are the most convincing ones:
6 It is unlikely that Chaucer has opted for a historic-present-tense passive to report a past event
in (37) due to what he uses in the surrounding context, as in the following preterite passives,
just a few lines above and below the example in (37):
(i) For sikerly, whan I was bore, anon … [3891]
(ii) This dronke Millere hath ytoold us heer
How that bigyled was a carpenteer [3913–3914]
7 (Have-)perfects can occur with specific past time adverbials to this day, as in the following
example from the conversation section of a modern corpus (see Quirk et al. 1985: 195 n. [a],
Michaelis 1994, Miller 2000, 2004a, Bowie et al. 2013, Brown and Miller 2016: 229, 245–247):
(i) I think I’ve had a meal there years ago. [from Biber et al. (2002: 369)]
158 chapter 5
Has the game or tale begun/ended or has it been begun/ended (cf. Chapter 4,
Section 4.4.2.5 for OE parallels)? Has the character’s heart broken or has it been
broken? Damagingly for functionalists, such supposedly ill-suited examples as
the Campaign is begun, Christmas is ended or it is broke into factions and partys,
are rather long-lived and are attested as late as the 18th and 19th centuries (see
Rydén and Brorström 1987: 54–57, 81, where they are treated as perfect only).
Turn is also discussed at length in Rydén and Brorström (1987: 175–180), with
an analogous ‘ill-adapted’ example as late as the 1830s from Darwin himself: his
mind is now turned to exertion … (Darwin).
In fact, the above number might have to be revised downwards; if turn in
(43) is conceived of as an intransitive verb in the perfect, i.e. ‘all his fantasy
had turned to learn astrology’, the sentence does not seem particularly well-
formed, since it hardly warrants a pluperfect time frame (though it cannot be
completely ruled out). As it is, was turned is probably adjectival and means
‘was inclined/prone’. The example in (44), and another identical one, were like-
wise put down as ambiguous perfects/passives, though they are most likely
just copular clauses comparable to today’s ‘we are agreed/accorded’.
Gräf (1888: 77, 81) believes that be(e)n accorded and ysworn is simply a per-
fect. ME accorden can mean ‘to agree, to become reconciled’ or ‘to reconcile,
make friends of (persons)’, so it has both transitive and intransitive uses (see
the Middle English Dictionary). In addition to the phrase been accorded, the
participle also appears with another link verb: fallen accorded ‘to become rec-
onciled’, as in (45), which suggests copular status for both (44) and (45).
(45) But at the laste, with muchel care and wo, We fille acorded by vs seluen
two. He yaf me al the brydel in myn hond. [(c1395) Chaucer CT.Mch.
(Manly-Rickert); Source: https://quod.lib.umich.edu/cgi/m/mec/med-id
x?type=byte&byte=1169670&egdisplay=compact&egs=1178495, accessed
on 10 Jan 2017]
Taking out the two instances of accorden and the one with turn would leave
only three genuinely ambiguous be-perfects or passives, but I keep them
The perfect from Middle into Modern English 159
8 The Appendices of this book have been made available online. They can be accessed through
a QR code which can be found in the References section of this book.
160 chapter 5
[…] they herde a belle clynke’. However, Gräf (1888: 92) provides Chaucerian
examples with past participles after see.
It is important to reiterate the main points that have emerged in this section.
Firstly, the majority of intransitive verbs which combine with have instead of
be do not seem to do so in order to avoid ambiguity. Secondly, although ambi-
guity between perfect and passive might at first sight appear to be widespread,
the syntactic and semantic context significantly reduces the number of un-
clear cases. Thirdly, ‘undesirably’ unclear constructions persist for an astound-
ingly long time. Having thus analysed Chaucer, I now move on to a text from
the following century.
Table 2 Perfect, passive and modal auxiliaries in the Second Shepherds’ Play (Secunda
Pastorum)
As with Chaucer, here I included verbs with unexpressed objects which how-
ever are normally transitive:
(50) Have done! [i.e. ‘Get it over with! Finish up!’] [663]
162 chapter 5
More dynamic is run in (54), which is used with have by Chaucer too.
The most interesting use involved a construction with have + gone in (55),
which did not express any hypothetical meaning (cf. esp. Chaucer’s examples
(12) and (13)). Be could have been employed instead without any ambiguity
in (55).
Notably, this occurs side by side with a question by another shepherd, who
asks: Is he commen?, see (56). It is puzzling that be was not selected for (55),
and ambiguity cannot have been a determining factor. The combination
of have + gone with a non-hypothetical meaning is innovative compared to
Chaucer, and even to the later texts below. It is tentatively suggested that this
might be due to the northern provenance of the Second Shepherds’ Play. It is
well known that the northern dialect of Middle English is more progressive in
terms of grammar, and this is where most grammatical innovations originate
before they make their way south (e.g. changes in verbal morphology, as well
as loss of inflection or noun gender; see Danchev 1997, Irvine 2004: cliv, Brinton
and Arnovick 2011: 288ff., 296ff., 311, Dolberg 2012, 2018, forthcoming, Lutz 2012:
514, among many others). It seems that be-perfects first started to be lost in
9 This number includes three instances of the past participle farne, probably from ex-
tended senses of the OE verb faran, ‘go, travel’, ME faren, today’s fare. See Appendix 4
and the Middle English Dictionary entry (https://quod.lib.umich.edu/cgi/m/mec/med
-idx?type=id&id=MED15284, accessed on 11 Jan 2017).
The perfect from Middle into Modern English 163
the north too. This coincides with the hypothesis involving language contact
with Scandinavian as a contributing factor to the loss (see Chapter 3), as well
as with the data from MSS. D and E of the Anglo-Saxon Chronicle presented in
Chapter 4. Contact with Old Norse in the north of England is highly likely to
have resulted in more and more preference being given to the prominent have
auxiliary at the expense of marginal be. Therefore northern sources from the
Danelaw area, including Secunda Pastorum, show innovative instances of sub-
stituting have for be with common intransitives such as go and fare(n) ‘travel’,
in addition to exhibiting consistently low rates of surviving be-perfects com-
pared to more conservative material from the south.
In line with what was said above, the use of be with intransitives in this
sample is severely curtailed. There are only 4 instances in the entire play, and
those are with the usual last survivors, come and go. Come and go carried on
combining with be as late as the Victorian period (though see the discussion of
Dickens’ language in Chapter 3 above and the Epilogue below; cf. Rydén and
Brorström 1987: 61ff.). Again this 15th-century text proves to be quite ahead of
its time.
(59) Thay ar gone full clene, that have thay lorne. [‘they have lost the chance’]
[694, translation from Walker 2000: 56]
For (56), Walker (2000: 46) gives the translation ‘Is he here?’, which might sug-
gest subject-complement status for the participle, but this might be anachro-
nistic. (56) is still best regarded as a perfect, though perhaps already undergoing
some fossilisation into a fixed expression. Another important observation
is that not many of the intransitives which combine with have instead of be
would have been potentially ambiguous if used with be – they would be either
completely inappropriate with it (?*so long am I nappyd; see (51)), or absolutely
unambiguous, as in (60), where, incidentally, the pluperfect serves a modal
function of indicating non-fulfilment.
(60) I wald [‘wish’] I had ryn [‘run’] to [‘until’] I had lost hir [108]
[OR: ?‘I was/were ryn’ – still unambiguous because of context?]
164 chapter 5
However, the clauses in (61), semantically a ‘future perfect’, and (62), where
Present-Day English could resort to a perfect progressive (‘has been lying’),
might create confusion if be were substituted in them, and I will return to such
cases below. There were only about three such potentially unclear instances
out of the eleven have-perfects with intransitives.
More evidence will be adduced below that the advanced loss of be-perfects
and the widespread use of have with intransitives in Secunda Pastorum does
not seem to have been prompted primarily by the avoidance of ambiguity. In
the light of this, language contact emerges as a more credible explanation.
Contrary to what would be expected if this displacement was mostly due to
a drive to achieve a one-to-one correspondence between form and function/
meaning, have, in addition to virtually taking over the role of perfect auxiliary
across the board, starts to appear in contexts suggestive of some modal colour-
ing, as already hinted in the discussion in Chapter 3 and illustrated here with
authentic material in context. There are two or three potential examples in
the text of the incipient recruitment of have for expressing modal (or at least
delexical) meanings:
(64) I had lever [‘rather’] be dede [‘dead’] or [‘than’] she had any dyseasse.
[486]
Admittedly none of them is indisputably modal. The first one, in (63), is partic-
ularly unclear and should be discarded. It appears to showcase delexical have,
i.e. ‘have/take hold of my hand’ (Steven Kaye, p.c.). (64) is the most convincing.
Here, the context suggests the translation ‘I would rather be dead [now] than
10 The distinction between intransitive lie and transitive lay must have helped, though it has
been lost in many dialects (since?). Van Gelderen (2014: 130) mentions confusion of lie
and lay in Middle English. I treat (62) as potentially ambiguous in order to be generous to
the hypothesis that is argued against.
The perfect from Middle into Modern English 165
For (65), Walker (2000: 56) supplies the reading ‘We have nothing to lose’, so it
is not at all certain if have has a modal colouring, but an interpretation where-
by it deontically expresses ‘We shouldn’t lose it’ is not that far-fetched, though
probably not the best reading in context. Nevertheless, the meaning of have,
on either interpretation, is slowly drifting towards modality.
As before, it was very difficult to tease apart bona fide passives from copular
statives. In context, a dynamic reading is favoured for (67) because of the verb
yede ‘went, walked’ in the subordinate clause, and because the entire play is
about the act of stealing the sheep (compare a similar example in the discus-
sion of Chaucer’s tales above).
Similarly, though slain might be paraphrased with the adjective ‘dead’, I still
included (68) in the passive section, so that ambiguous cases did not multiply
exponentially. This choice is in no way indisputable or foolproof, but it was
thought safer to err on the side of caution, so that the number of potential
ambiguities would not be artificially inflated to prove my point.
11
M E lever(e) goes back to a comparative form of OE leof ‘desirable, beloved, dear, precious’,
related to OE lufu ‘love’, from PIE root *leubh- ‘love’, with cognates in German lieb, Gothic
liufs, Old Icelandic ljūfr, Latin libīdō ‘desire, pleasure’, Old Church Slavonic/Old Bulgarian
ljubъ ‘pleasant’ (Łęcki 2010: 69–70). It is this word, in addition to the present-time refer-
ence and the simple preterite in the second clause introduced by or in (64), that prevents
a perfect reading of ‘I had rather been dead’/‘I would rather have been dead’, which is less
likely here.
166 chapter 5
Even with this conservative approach, most of the ambiguity again involved
genuine passives, 36 in number, vs. potential statives, 17. By contrast, there
were hardly any ambiguous be-perfects (see below). Stative copular status was
indicated by adjectival, (69), or nominal paraphrase, (70), or by coordination
with nouns or adjectives, (71); so for (71), coordination with adjectives is the
main reason why it was not counted as a genuine passive.
(70) These men that ar wed have not all thare [‘their’] wyll [73] [‘are (part/
members of) a couple’]
The example in (72) was treated as passive, but this choice needs to be justified:
Only a passive reading seems viable in this context. ‘With you will I have left’
does not make much sense at all, because of the anomalous temporal orienta-
tion and the meaning which this interpretation would entail; the context and
the preposition with here suggest that the speaker and addressee(s) will/wish
to remain together, not that the speaker will/wishes to leave a place or a person
in the addressee’s company (though cf. the section on Chaucer and Gräf 1888:
79–80, who would perhaps admit a perfect reading in the sense of ‘remain’).
Once more, when all things have been considered, the result is a negligible
number of only two ambiguous be-perfects, (73) and (74):
(cf. Chaucer in the previous section). What is more, the verb run appears three
times in the perfect elsewhere in the play, all of them with have and none with
be, as in (75), making it unlikely that (73) is a be-perfect.
(75) Ye have ryn in the myre, and ar weytt yit [‘still wet’] [494]
are from Evans and Tobin (1997). The following website was also used: http://
shakespeare.mit.edu/taming_shrew/full.html (accessed on 28 July 2016). Only
the first three acts were examined (as they appear in Evans and Tobin 1997),
because otherwise this sample would have grown disproportionately big com-
pared to the others. The first three acts amount to approximately 13,247 words,
making this the largest sample analysed so far in this chapter. Importantly, this
text is not included in the Helsinki Corpus.
(76) Seeing too much sadness hath congeal’d your blood [Induction, Scene II, l.
132]
table 3 Perfect, passive, progressive and modal auxiliaries in the first three acts of
Shakespeare’s The Taming of the Shrew
(78) And therefore has he closely mew’d her up [‘placed her in close
confinement’],
Because [‘so that’] she will not be annoy’d with suitors. [I.i.183–184]
Throughout the periods studied so far in this chapter, word order proves of
little value in indicating the extent to which the perfect is grammaticalised.
Another typical feature of Shakespeare’s English is that a relative clause might
be missing its subject relativiser:
Other examples include woo [somebody], (82), and tune [something], (83).
12 H
ave applies to a (potentially) transitive and to an intransitive verb in this example. There
is hardly any possibility of finding be, however, as it is doubtful that sleep would normally
combine with it anyway; as confirmed by (52) above, as well as by the OED and the follow-
ing examples from the Middle English Dictionary, sleep usually takes have:
(i) Þo hi hadde alle islepe [Ashm: islept; vrr. slepe, slepen] ynouȝ, sone hi gonne arise &
wende to here schip. [a1325(c1280) SLeg.Pass. (Pep 2344), https://quod.lib.umich.edu/
cgi/m/mec/med-idx?type=byte&byte=182957819&egdisplay=compact&egs=1829883
22&egs=182999307, accessed on 12 Jan 2017]
(ii) Þat niȝt he hadde litel yslape. [c1350(a1333) Shoreham Poems (Add 17376), https://
quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byte=182957819&egdisplay=
compact&egs=182988322&egs=182999307, accessed on 12 Jan 2017]
(iii) When thei hadde scleped & saw tyme, Thei ros vp be-fore the prime. [c1440(a1401) Life
Bridlington in NM 71 (Yale 331), https://quod.lib.umich.edu/cgi/m/mec/med-idx?
type=byte&byte=182957819&egdisplay=compact&egs=182988322&egs=182999307,
accessed on 12 Jan 2017]
170 chapter 5
(83) His lecture will be done ere you have tun’d. [III.i.23]
Line 25, only two lines down from sentence (83), contains the words tune your
instrument, thereby confirming the transitive nature of this verb. Verbs with
finite, (84)–(85), and non-finite, (86), clausal complements were treated as
transitive too, even when incompatible with NP direct objects.
(87) You have but jested with me all this while [II.i.20]
(90) approximates a set expression, along the lines of ‘finish up, get it over
with’, like modern ‘have done with it’, but it was nevertheless included in the
count:
Have got was accepted as a bona fide perfect construction too – (91)–(92).
Judging by the context, it is unlikely that it had by that stage been reinterpret-
ed as a possessive stative equivalent of ‘have/possess/own’ (at least in these
examples).
(93) Persuade him that he hath been lunatic [Induction, Scene I, l. 63]
[copular be]
(94) We have not yet been seen in any house [I.i.199] [perfect passive]
The most surprising finding, compared to the texts from the 14th and the 15th
centuries, is the virtual absence of intransitives which form their perfect with
have, despite the larger size of this sample. There were only two such instances:
(96) features a conjoined structure where the auxiliary have applies to two
past participles, the first taken to be potentially transitive and the second a
non-mutative intransitive (as already mentioned above). (97) can be explained
away by reminding readers that have tends to combine with intransitives of
172 chapter 5
motion such as come in hypothetical contexts, i.e. ‘then G. wouldn’t have come
by the worst’ or ‘if then G. had not come by the worst’ (see Chapter 3, esp.
Sections 3.3 and 3.4.1, as well as Rissanen 1999: 229–231 for Early Modern condi-
tionals). In the light of this, it is remarkable that Shakespeare prefers be to form
the perfects of (mutative) intransitives, a total of 17, including a surprisingly
conservative counter-factual conditional clause with the be-perfect of come,
(98). This goes against the tendencies noted in Chapter 3 that hypothetical
contexts and the past tense both favour have over be.
Regarding (102), only a perfect reading, i.e. ‘until the sun have/hath/has set’
makes sense; (102) cannot be the passive of an active *Someone (has) set the
sun (cf. (74) from the previous section for another treatment of set).
Likewise, the passive would make no sense for (103). What is unequivocally
meant here is ‘My master has grown/become quarrelsome’ (cf. *Someone grew/
has grown my master quarrelsome).
There were numerous instances of be welcome, but they were discarded as ad-
jectival and not recorded, apart from the following one, in which the dialogue
might indicate that contemporary native speakers could perhaps feel a certain
link between welcome as an adjective and a perfect of having/being [well] come:
(107) For me, I’ll not be gone till I please myself. [III.ii.212]
Admittedly, let’s have gone and I must have gone would sound markedly anom-
alous if inserted above, so be gone is probably a set phrase already, though
13 Etymologically, welcome as a verb derives from Old English wilcumian ‘to welcome, greet
gladly’, ultimately from wilcuma (n.) ‘a welcome guest’, literally ‘will-comer’, i.e. ‘one
whose coming suits another’s will or desire’, from willa ‘will, desire, pleasure’ + cuma
‘comer, guest’, related to cuman ‘come’ (see the OED). The first element of this compound
word was subsequently altered to the adverb wel(l) and the second element cuma came
to be identified with the imperative or infinitive of the verb come, under the influence
of Old French and Latin. This folk etymology must have served as the popular basis of
Shakespeare’s word play and confirms that the general English-speaking public did per-
ceive a connection between the compound welcome and its components well + come.
174 chapter 5
I will not have gone till I please myself is somewhat more convincing. Rydén
and Brorström (1987: 104 fn. 4) exclude constructions like imperative be gone or
I must be gone ‘I must go’ from the category of genuine perfects since there is
no opposition with have. The salient semantic component in be gone must be
the result, whereas have gone would probably highlight the anteriority.
Importantly for present purposes, this snapshot of the history of English
vividly demonstrates the proliferation of functions of both auxiliaries under
investigation. Already present here are modal and causative uses of have, as
well as modal and progressive be. The modal construction had lever/rather,
previously encountered in the 15th-century Townley play, finds continuation
in (108):
For both of these, Evans and Tobin (1997: 151, 159) provide the translations
‘what is it to you’ or ‘it’s no concern of yours’. No matter whether they repre-
sent deontic obligation, as it would appear to a modern reader, or whether
they stand for ‘it’s none of your business’, as per the translations provided in
the explanatory notes, modal overtones are present anyway. Note also the in-
terchangeability of have and be in the originals and the translations, once more
showing how one generic verb can easily substitute another. (111), on the other
hand, is only faintly modal, mostly because of the imperative mood, and might
need to be discarded:
As indicated elsewhere in the play (see e.g. I.ii.194), to’t stands for ‘to it’. Have
to it does not exist in English now, but have at it does, in the meaning ‘get on
with it’, ‘get down to it’, ‘take it away!’ (Steven Kaye, p.c.).
(112) is likewise not unproblematic, but it might illustrate the rise of caus-
ative have:
The perfect from Middle into Modern English 175
In Shakespeare’s English, verbs of motion are often omitted under certain cir-
cumstances, as amply illustrated elsewhere in the play (e.g. I’ll in to counsel
them [Induction, Scene I, l. 136], i.e. ‘I’ll go in to counsel them’). So a verb of mo-
tion might be reinserted in (112) above too, something along the lines of ‘we’ll
have thee taken to a couch’. Be that as it may, the following are less contro-
versial causatives, demonstrating how have has been branching out into new
(quasi-)auxiliary territories:
(113) I’ll have them [‘books’] very fairly [‘handsomely’] bound [‘I desire them
to be, i.e. see that they are’] [I.ii.145]
14 It should be noted, however, that despite the emergence of the progressive, there are
many instances in which Shakespeare uses a simple form where a progressive would be
required today (see Rissanen 1999: 216, Beal 2004: 78ff., and example (129) below). (Proto-)
progressives were mentioned in the sections on the Anglo-Saxon Chronicle in Chapter 4.
On the (disputed) continuity between the OE progressive and the one that I now record,
consult Chapter 4, Fischer and van der Wurff (2006: 135ff.), Brinton and Arnovick (2011:
229, 373).
176 chapter 5
There are two instances of be, (119)–(120), in which it combines with a to-
infinitive and has undeniable modal overtones, just like have above. The modal
be-to construction expresses futurity combined with deontic colouring, as it
still does today. (120) can be paraphrased as follows: ‘I need to/should/have to
get a man’, once more showing the synonymy of modal be and have. The shift
to deontic and future meanings was illustrated in Chapter 3, Section 3.5.2.3.
The passive still accounts for the greatest share of auxiliary uses of be – as
many as 57 dynamic passive clauses. (121) and (122) attest to the freedom with
which Shakespeare could employ conversion and coin passivisable verbs out
of other parts of speech, even proper nouns, (122) (see Crystal 2004: 332–333).
Apart from the perfect passives noted above, there were also non-finite pas-
sives: with a present participle, (123), and a to-infinitive, (124), including a
passive bare infinitive after a modal, (125).
Although such complex combinations are overall on the increase, there are
nevertheless instances where we would now expect, say, a perfect passive, but
we find a simple form instead, so grammaticalisation has not quite progressed
to its modern state (see Chapter 3):
(126) So workmanly the blood and tears are drawn. [Induction, Scene II, l. 60]
[instead of ‘have been drawn’, in a painting]
More convincingly, (129) is best read dynamically as: ‘To me she is getting/
being married’, since they haven’t got married yet at this stage in the play. Note
incidentally that now we would use the progressive to convey the meaning of
arrangement for the future, as pointed out earlier. Also compare the previous
samples, where other category-assignment decisions were made for the same
participle due to context. This vindicates taking the approach outlined at the
beginning of Chapter 4.
Again due to context, it is clear that (130) is passive too – the gentleman has
been grieved by his shrewish daughter, and grieve is not really mutative. ‘Had a
gentleman thus grieved’ would not fit in context, and also perhaps temporally,
as the past perfect might introduce anomalous time reference.
(133) But art thou not advis’d [‘aware’], he took some care [I.i.186]
[‘Aren’t you aware/don’t you know?]
(135) But be thou arm’d for some unhappy [‘hateful’] words. [II.i.139]
[‘prepared, ready’]
Similarly, a perfect reading (‘His lecture will have done’) is ruled out for (137),
for which there is a ready adjectival/adverbial paraphrase – ‘His lecture will be
over’ (cf. (134) above and Chaucer in Section 5.2.2):
15 There were other combinations of be + done here which were treated as genuinely passive
because no such adjectival paraphrase was available – see Appendix 5.
The perfect from Middle into Modern English 179
(137) His lecture will be done ere you have tun’d. [III.i.23]
Secunda Pastorum, where be-perfects are severely curtailed, and then returns
with Shakespeare, who uses be-perfects more freely.16 In effect, be-perfects
tend to be accompanied by ambiguity. If this is so, though, Shakespeare does
not seem to resolve uncertainty of interpretation by resorting to have-perfects
with intransitives, as the earlier authors had done, including Chaucer, whose
sample contains more mutative have-perfects (disambiguating at least to a
certain extent). Those are virtually absent here and might suggest that a cor-
relation between unclear be-perfects and the rise of have-perfects with mu-
tatives is at best tenuous. Lack of clarity might just appear to thrive where
be-perfects thrive themselves. Secunda Pastorum could indicate that eliminat-
ing be-perfects eliminates ambiguity. So does ambiguity ‘thrive’ in Shakespeare
because he does not rely on have-perfects with mutative intransitives (instead
of be) and the be-perfect has not been suppressed to a sufficient degree?
In fact, thriving might be an overstatement – out of the eight instances in
the first three acts of The Taming of the Shrew, not that numerous as they are,
four are only partially ambiguous, (139)–(142):
None of these would make a very convincing passive today – compare the
modern ill-formedness of ?*Someone resolved/agreed me, though resolve had
transitive senses such as ‘inform’ or ‘satisfy’ in Shakespeare, and Wycherley’s
Country Wife does have Thou hast stared upon her enough to resolve me [II.i,
p. 172] (see Appendix 6,17 as well as the OED for earlier transitive uses of agree
16 Here, the temporal dimension should be compounded with the dimension of regional
variation – Chaucer and Shakespeare seem to represent conservative southern usage,
whereas Secunda Pastorum is innovative for its time because it originated in the north. By
the Restoration period, the change has presumably made its way south too.
17 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
The perfect from Middle into Modern English 181
and resolve with human objects). Resolved can take a to-infinitive clausal
complement, as in (139), so its perfect should be built with have. In addition,
a perfect reading would be strange in context for I am agreed. I have agreed
would indicate some sort of temporal precedence, which is hardly appropri-
ate, though it should not be ruled out completely (i.e. (142) might focus on
the present state).18 Despite the theoretical ambivalence potential, examples
(139)–(142) should perhaps be analysed as mostly adjectival. This is confirmed
by Shakespeare’s usage elsewhere. As evident from (85) above, when he forms
the legitimate perfect of the verb agree, he selects have, as we would today:
we have ‘greed so well together... What is more, both adjectival constructions,
with agreed and resolved, survive intact to this day, making it hardly likely that
they were ever thought of as real be-perfects, or at least they conceivably be-
came ‘fossils’ like be gone. The adjectival status of PDE resolved, for instance,
is confirmed by compatibility with modifiers like very in (143). In any case, the
survival of these expressions makes it unlikely that ambivalence was ever an
issue that needed to be fixed.
(143) Well, the president is very resolved, looking for a – a way to have a diplo-
matic solution [COCA, 1990s]
This leaves only four indisputable cases which could be perfect, as well as
passive – (138) above and (144)–(146) below:
(145) My lord, ‘tis but begun. [i.e. ‘the play is but begun’] [I.i.252]
(146) Petruchio:
Myself am mov’d to woo thee for my wife.
Katherine:
Mov’d! in good time [‘indeed, forsooth’]! Let him that mov’d you hither
Remove you hence. I knew you at the first
You were a moveable. [II.i.194–197]
18 Nakamura (1987: 34–35) treats they are not yet agreed to impeach him and the Duke of York
and Duchess and my Lady Castlemaine are now all agreed in a strict league from Pepys’s
Diary as be-perfects. Crucially, Nakamura often fails to take into account potential ambi-
guity, e.g. phrases like [X] is altered, the war is begun, or the river is frozen are regarded as
perfect only (see e.g. Nakamura 1987: 35ff.).
182 chapter 5
(146), for instance, could be an intended perfect (‘I have moved to woo thee …’)
which is then turned into a pun by Katherine, who takes it to be a passive in-
stead. Rydén and Brorström (1987: 147–148) note that this verb tends to favour
have ‘obviously’ in order to avoid ‘interpretative difficulties’, though they quote
ambiguous examples from the 19th century. As pointed out above, we know for
a fact that Shakespeare was fond of such linguistic play and loved puns based
on the availability of more than one reading – allowing multiple interpreta-
tions appears to be favoured rather than disfavoured in this text. It turns out
that ambivalence is embraced and cherished, rather than avoided. Even so,
upon consideration ambiguity seems hardly more widespread here than it was
in the previous samples (considering their sizes), as highlighted already. Once
again, it is worth stressing that such ‘interpretative difficulties’ persisted well
into the modern period (cf. also Chaucer’s earlier use of begin), with similar
examples of be changed from the 18th and 19th centuries provided by Rydén
and Brorström (1987: 58–61). Despite its supposedly ‘inferior fitness’, Rydén
and Brorström (1987: 59) note that be is still clearly favoured for the perfect of
change as late as the 19th century.
In the same vein, there were only two instances of lack of clarity as to the
choice of be or have due to a missing auxiliary – (147) and (148). The second
of those, (148), is rather dubious, as it is surrounded by unequivocal passives
or adjectives – the unfortunate horse in question is troubled, infested, defiled
with and full of various diseases (see further Section 5.6.2.2).
Finally, there was only a single contracted ’s in the excerpt, (149). In the light of
all the evidence presented here, we can be fairly confident that is should be re-
covered in both instances involving come, (147) and (149), as far as Shakespeare’s
usage is concerned. Moreover, there are indications that ’s can only stand for is
in pre-18th-century English more widely (Rydén and Brorström 1987: 32 fn. 10).
In conclusion, the three text samples investigated so far in this chapter dem-
onstrate that cases of ambiguity in context are not as many as might have
been expected – they consistently do not tend to exceed the 5% mark in all
The perfect from Middle into Modern English 183
table 4 Perfect, passive, progressive and modal auxiliaries in the first two acts of
Wycherley’s The Country Wife
(150) and he’s a sign of a man, you know, since he came out of France. [I.i,
p. 162] [instead of ‘has been a sign of a man’]
(151) and till now I never thought I should have envied you or any man about
to marry [II.i, p. 172]
Similarly to the texts examined earlier, the present simple might still be used
in the passive, as in (152), in order to avoid a combination of a perfect passive,
despite the meanings of anteriority and current relevance (cf. Chapter 3).
(152) The writings are drawn, sir, settlements made [II.i, p. 174] [instead of
‘have been drawn’]
(154) Madam, you would not have been so little generous as to have told him.
[II.i. p. 175]
(155) Thou hast stared upon her enough to resolve me. [II.i, p. 172]
(157) I have left at Whitehall an еarl to dine with you [I.i, p. 162]
(158) your appearance at the play yesterday has, I hope, hardened you for the
future [I.i, p. 158]
Another feature that sets this 17th-century drama sample apart is the fact that
for the first time the material yields the perfect of auxiliary or substitute do:
(159) for he would but ruin you, as he has done hundreds [II.i, p. 171]
(161) What dost thou blush at nine and forty, for having been seen with a
wench? [I.i, p. 166]
The perfect from Middle into Modern English 187
In the light of the comments regarding word order, (163) was deemed ambigu-
ous. It comes from one of the few lines of verse in the text, so the alignment in
which the object intervenes between the auxiliary and the main verb might be
due to metric or rhyming considerations:
(163) Who for his business from his wife will run,
Takes the best care, to have her business done. [II.i, p. 185]
Therefore, this could be a genuine perfect with a word order that is not found
elsewhere in the first two acts of this play (of the type characteristic of the ear-
lier verse samples). Alternatively, it could be a causative or quasi-passive con-
struction. In support of this reading, the context implies that someone else will
do the wife’s ‘business’ instead of the husband. Crucially, this deviance in my
drama database confirms that, even at this late stage, it is not only be-perfects
that were plagued by uncertainty of interpretation (for the ambiguity of have-
perfects, cf. the sections above and the previous chapter on OE).
In general, here and in the other samples, transitives by far outnumber
intransitives, which might have contributed to the demise of be-perfects, as
already discussed in the previous sections and chapters. It is interesting to
ponder whether people talk more about events involving more than one par-
ticipant. There were only four properly intransitive have-perfects. The first
three, (164)–(166), are of verbs which are not typically mutative. What is more,
the hypothetical (and non-finite, after the modal) context in (165) and (166)
favours the use of have, as noted in Chapter 3.
(165) but I would have ventured for all that. [II.i, p. 168]
(166) and I would as soon have lain with him, as have named him.
[II.i, p. 185]
The sentence in (167) shows that have can displace be even with verbs such
as go, though I found only one instance of this, as opposed to many more in
which go combines with be (see below and compare similar examples from
the earlier case studies of Chaucer and Secunda Pastorum, so this can hardly
be called a precedent).
188 chapter 5
(167) I’ve gone too far, and slipped before I was aware [II.i, p. 171]
There were still quite a few be-perfects, three times more than intransitive per-
fects with have. No fewer than four of those contain the phrase let’s be gone, as
in (168)–(169).
(168) Pray, Sir Jasper, let us be gone from this rude fellow. [I.i, p. 155]
(169) Why do we stay? Let’s be gone; I can’t indure the sight of him. [I.i, 156]
No anteriority seems to be implied here and by this stage be gone is in all prob-
ability a fossilised fixed phrase which is no longer a genuine perfect formed on
productive syntactic principles. Indeed, it was often written as a single word
in this period (Steven Kaye, p.c.). Apart from the four instances of let’s be gone,
two similar examples might at first suggest that the perfect here is not that
moribund, since it is not invariably preceded by let’s, as in (170) (cf. imperative
have done [with words] from example (90), Section 5.4.2).
Nevertheless, speakers may have felt that be gone was no longer a bona fide
perfect, but a fixed phrase which is hard to decompose as a sequence of verbs,
making it a copular adjectival construction with a stative meaning, or even
a compound verb. This much is suggested by substituting get for be in lines
which appear in close proximity, giving the clause a more dynamic feel:
(171) and (172) have an adjectival participial object complement, which might
hint that be gone was already adjectival in status too, with the participle acting
as subject complement (cf. previous section). Again, no anteriority is indicated
by be gone in (173), and the modal must arguably shows that be gone behaves
like one (compound?) verb.
(173) Come, is your ladyship reconciled to him yet? Have you agreed on
matters? For I must be gone to Whitehall. [II.i, p. 184]
The perfect from Middle into Modern English 189
As expected, there are also more typical be-perfects with come and go,
the mainstay of this type of construction which can be found as late as the
19th century (cf. Chapter 3 and the Epilogue below):
(174) I can be sure she that shews an aversion to me loves the sport, as those
women that are gone, whom I warrant to be right [I.i, p. 157]
(175) We are come to wait upon her to the new play. [II.i, p. 177]
(176) She has locked the door and is gone abroad. [II.i, p. 177]
(177) Fie, fie upon ‘em! They are come to think cross breeding for themselves
best, as well as for their dogs and horses. [II.i, p. 178]
(178) and it is come to that pass, we dare not take your words, no more
than your tailor’s, without some staid servant of yours be bound with you
[II.i, p. 184]
Finally, it should be noted that (179) can be nothing but a perfect of copular
grow, equivalent to ‘has become’. A passive reading is easily ruled out in con-
text (as in the previous periods).
(179) So the little fellow is grown a private person with her. [II.i, p. 179]
(181) Come, Horner does not come [II.i, p. 176] [instead of ‘is not coming’]
Modal have is represented by had better, (182), and a similar construction with
as good in (183).
(182) and he had better employ her, than let her employ herself [I.i, p. 156]
(183) Well, I see one had as good go to law without a witness, as break a jest
without a laughter on one’s side. [I.i, p. 162]
(184) Sir Jasper: Well, well, in short I have business at Whitehall, and
cannot go to the play with you, therefore would have you go –
Lady Fidget: With those two to a play? [II.i, p. 180]
(185) and (186) also contain some tinge of (semantic) modality in them, most
plausibly volition. These instances of have do not take another verb and are
best treated as more or less delexicalised, albeit without being accorded aux-
iliary status.
(185) So, so, why, ‘tis as I’d have it. [II.i, p. 185]
(188) Shy husbands and keepers like old rooks [‘swindlers’] are not to be
cheated, but by a new unpractised trick [I.i, p. 154]
(189) Do not teach my wife where the men are to be found. [II.i, p. 169]
(190) Why, the next thing that is to be heard is, thou’rt a cuckold. [I.i, p. 163]
The perfect from Middle into Modern English 191
(191) This is one of those, my pretty rogue, that are to dance at your wedding
tomorrow [II.i, p. 172]
Even if the construction is taken to express pure futurity, without deontic over-
tones, futurity is still in some sense related to modality, since prediction can
hardly be divorced from expressing the speaker/writer’s attitude to the propo-
sition, rather than merely reporting a fact.
Once again, it was often exceedingly difficult to judge whether something
was a dynamic passive or a clause that is indeterminately passive or copular,
so the decisions made are sometimes arbitrary and open to debate. I finally
settled on 50 dynamic passives vs. 15 instances of ambiguity (see Appendix 6).19
Very frequently, the same verb or construction might be classified differently.
For (192)–(196), the context imposes a dynamic passive reading, where there is
a single occurrence of an activity on a particular occasion:
(194) … wife, who before would not be acquainted with me out of jealousy
[I.i, p. 157]
(195) And so go to her, begin thy new employment. Wheedle her, jest with her,
and be better acquainted one with another. [II.i, p. 183]
(194), for instance, was taken to mean that the wife refused to be introduced
to the speaker (the punctual action), rather than not wanting to be his ac-
quaintance (the state). For (196), a stative ‘I shall be free of …’ does not sound
convincing in context. By contrast, (197)–(199) were treated as ambiguously
stative, since no comparable dynamic readings were unequivocally imposed
by the context, and adjectival paraphrases were readily available, as indicated
in the brackets after each example sentence:
19 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
192 chapter 5
(197) I’m satisfied you are of the society of the wits and railleurs [‘banter-
ers’] [II.i, p. 173] [= ‘I’m willing to believe that you are …’ or, less likely,
‘I’m happy/glad that …’]
(198) Nay, by this means you may be the more acquainted with the husbands,
but the less with the wives. [I.i, p. 157] [= ‘more familiar’]
(199) And next to the pleasure of making a new mistress is that of being rid of
an old one [I.i, p. 157] [= ‘of being free of’]
(200) was treated as closer to the adjectival end of the scale mainly because of
the adverb well (though compare better in (195)), and the available paraphrase
with the adjective popular:
(200) I have given you security already to save you harmless, my late
reputation being so well known in the world, madam. [II.i, p. 184] [= ‘so
popular in the world’]
(201) a woman is hated by ‘em as much for loving her husband as for loving
their money [I.i, p. 155]
(202) I think that was prettily said of me, ha, Harcourt? [II.i, p. 173]
(204) and the crime’s the less when ‘tis not known. [II.i, p. 179]
(205) Well, that’s spoken again like a man of honour [II.i, p. 184]
The perfect from Middle into Modern English 193
(207) we are your friends, and will not take it ill to be left, I do assure you.
[I.i, p. 162]
(208) Hold, do not rail at him, for since he is like [‘likely’?] to be my husband,
I am resolved to like him. Nay, I think I am obliged to tell him you are not
his friend. [II.i, p. 175]
As emerged from the discussion in Section 5.4 above, example (208) should
perhaps be removed from the count, since it is most probably adjectival.
Resolved is clearly parallel to the more convincingly adjectival obliged, though
Wycherley uses it with an object too (to resolve sb., see Appendix 6),20 so (208)
might be passive. As it is, resolved also takes a complement – the to-infinitive
clause. So, were this to be a perfect, in all likelihood it should be formed with
have anyway. As in the previous sections, this was still put down as ambiguous,
despite the remote possibility, in order to ensure that my data is not skewed in
favour of the hypothesis I am trying to prove.
There were only two instances of missing auxiliaries in a non-conjoined
environment:
(209) for you look upon a friend married as one gone into a monastery, that is,
dead to the world [II.i, p. 173]
20 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
194 chapter 5
(210) my wife has just now the smallpox come out upon her [II.i, p. 177]
Based on the material above, the chances are that these ought to be construed
as combining with be, though there was a minority example of have gone too.
There were another two cases of ambiguity due to contraction of the auxiliary,
(211)–(212):
(211) the vizard-masks [‘prostitutes’], you know, never pitty a man when all’s
gone [I.i, p. 158]
(212) He’s come newly to town it seems, and has not heard how things are with
him. [I.i, p. 165]
Elsewhere in this portion of the text, ’s only instantiates different uses of is,
never of has (cf. the comment about ’s in pre-18th-century English made in the
previous section and Rydén and Brorström 1987: 32). I have not found a single
instance of ’s in the first two acts of the play where it unambiguously stands for
has; it could conceivably stand for has only in the two examples in (211)–(212),
but in all the other numerous cases, it certainly represents is, mostly the copula
or, on a few occasions, the progressive or passive auxiliary (see (180) and (205)
above). Secondly, as has been pointed out already, come and go tend to almost
exclusively combine with be to form their perfect. All of these factors contrib-
ute to disambiguation in the context of the discourse and favour be over have.
Needless to say, the loss of be as perfect auxiliary has of course in no way af-
fected the continued usage of such contracted or elliptical constructions: The
late visitors were Aunt and Uncle Israel Fish, come straight from the opera, still
in their evening clothes … [2003, The Great Husband Hunt by Laurie Graham].
It probably never mattered to speakers whether there was only one option for
spelling out the ‘complete’ version (with have), or more than one (be or have).
Finally, it should be mentioned that some combinations were excluded
from all the counts. For instance, do not be frightened does not belong in any of
my categories because the context rules out even a passive reading (?do not be
frightened by anyone). Similarly, (213) is not included in any of the categories,
as it is neither a passive derivable from Someone mistook the man, nor a perfect
like A man has often mistaken. It is adjectival only (be resolved from (208) above
probably belongs here too!). The same applies to (214).
(214) to try whether I was concerned enough for you [II.i, p. 176]
[= ‘was anxious’]
In conclusion, most of the ambiguity again lies in the grey area between genu-
ine passive vs. stative copular clauses. In the 17th century too, ambiguous be-
perfects seem to be statistically negligible, with the number of unclear cases
reduced even further by context. The relative proportion of contracted and
missing auxiliaries has risen, but the overall picture is the same, with indica-
tions that contraction should actually favour be. What is more, there is some
uncertainty attending the supposedly clearer and functionally superior have-
perfect, a residue of ambivalence paralleled in Old English (see Chapter 4).
Against this backdrop of consistently underwhelming amounts of ambiguity
of perfect be, The Country Wife offers one more sample which demonstrates
the proliferation of the functions of both auxiliaries under investigation. In line
with the findings of previous scholars, I document four centuries of reasonable
stability in the relative shares of the major functions of the two auxiliaries,
with the role of ambiguity being at best indecisive at all times and with little
perceptible fluctuation or indisputable directionality in one way or another:
4.2% of unclear be-perfects in Chaucer (from all sources combined), 1.9% in
Secunda Pastorum, 5.5% in Shakespeare, and 3.3% in Restoration comedy.
table 5 Perfect, passive, progressive and modal auxiliaries in the letters of William
Paston I
In these non-verse texts, the auxiliary and the main verb in have-perfect
constructions tend to stay together and not to be separated by the object.
There are a couple of exceptions in which nought comes between the auxiliary
and the lexical verb, but it quickly becomes clear that nought has by this stage
been decategorialised as a marker of negation, no longer the original reinforc-
ing noun/pronoun which meant ‘nothing, not a bit’. The object in (215) below is
probably the preposed more or perhaps this is the have-[something]-to-do-with
pattern with the object missing and more simply an adjunct, while trespass in
(216) is intransitive, with the PP headed by against possibly acting as a prepo-
sitional complement.
(215) and more I haue nought hadde to do with þe seyd John [PROBABLY TO
MASTER JOHN URRY: DRAFT 1425, 11]
(216) I haue nought trespassed a-geyn noon of these iij [TO WILLIAM
WORSTEDE, JOHN LONGHAM, AND PIERS SHELTON 1426, 03, 01]
Apart from those, there was only one more example of unusual word order in
a non-relative and non-interrogative clause:
The object hem ‘them’ is preposed in (217), whereas a potential reflexive pro-
nominal object is altogether missing in (218) below:
(218) And also þe Priour of þe seyd hous [unclear] hath resigned in-to yowre
worthy handes by certeins notables and resonables causes [TO THE
VICAR OF THE ABBOT OF CLUNY: DRAFT 1430, 04?]
An example of a verb with a PP complement which was grouped with the tran-
sitives is given in (219), whereas (220) shows the perfect of causative do, haue +
doon, followed by another past participle, examyned: a causative construction
which was unambiguous but was outlived by the likes of causative have, pre-
sented in (221) (see Denison 1993: Section 10.2.2.5).
(219) and þat þe same John, atte reuerence of yowr right worthy persone,
hath cesed of his sute of certeins processes ageyns me [PROBABLY TO
MASTER JOHN URRY: DRAFT 1425, 11]21
21 It cannot be ruled out that this is actually a phrasal verb, i.e. cease off.
198 chapter 5
(220) I haue, after þe aduys of yowr lettre, doon dwely examyned þe jnstru-
ment by þe wysest I coude fynde here [TO WILLIAM WORSTEDE, JOHN
LONGHAM, AND PIERS SHELTON 1426, 03, 01]
(221) and how sone I myght haue the seid stoon caried to Moneslé a-forn seid
[To JOHN STANFORD 1425]
(221) is most likely causative and not perfect in the light of the observations
about word order made above. Crucially, neither of the two intransitive have-
perfects, with trespass in (216) above and the one in (222) below, would have
been in competition with be.22
(222) monkes clad and vn-professyd, þat have abedyn þere wythow [unclear] t
abbyte ix or x yeere, [TO THE VICAR OF THE ABBOT OF CLUNY: DRAFT
1430, 04?]
(224) ne yet ne dar not, rydyn ne goo abowte swyche occupacion as he arn vsed
and disposed [MEMORANDUM TO ARBITRATORS 1426–7]
They are used and disposed would make no sense as passives. The anteriority
entailed by the perfect would not fit here either, as the reference is to the pres-
ent time. Used and disposed should therefore be regarded as adjectival. The
remaining two cases are equally suspect:
(225) And here aunswere is þat al þis processe, þough it were in dede proceded
as þe instrument specifieth, is not suffisant in þe lawe of Holy Cherche
[TO WILLIAM WORSTEDE, JOHN LONGHAM, AND PIERS SHELTON
1426, 03, 01]
22 If the PP headed by against in (216) is a genuine complement, this example had better be
transferred to the group of transitives.
The perfect from Middle into Modern English 199
Proceed does not normally take an object, but the dictionary entry below might
suggest that (225) and (226) are impersonal passives, which means that they are
not strictly ambiguous. According to the electronic Middle English Dictionary
(http://quod.lib.umich.edu/cgi/m/mec/med-idx?type=id&id=MED34752, ac-
cessed on 23 August 2016), in its most common sense, proceed means ‘to go or
move in a certain direction’ or ‘to continue in an action or process’ – a typical
mutative intransitive. The dictionary then gives the passive-looking phrases:
be hit proceded ‘let (the incision, ligature, ?procedure) be continued’, as well as
hit is proceded ‘it (a ligature) is continued’; that ... hit be proceded ‘that ... it (a
procedure) be carried on’. That this type of proceed is most probably passive is
indicated by the Latin original supplied in (227). Latin procedatur is a 3rd per-
son singular passive (in the present subjunctive) ending in -tur.
As confirmed by the same dictionary, it is proceeded can also mean ‘it is argued’,
including in a legal sense, which is probably the sense intended in William
Paston’s correspondence. Indeed, the dictionary explains the self-same exam-
ple in (226), ‘it is proceeded against the said Adam’, as follows: ben proceded ‘of
legal actions: to be carried out, put into effect; -- also impers.’ (http://quod.lib.
umich.edu/cgi/m/mec/med-idx?type=id&id=MED34752&egs=all&egdisplay=
open, accessed on 24 August 2016). So, upon careful examination, even those
cases of possible uncertainty turn out to be spurious, but they are in this cat-
egory due to the potential of proceed to build perfects and to passivise.
Finally, below are the two instances of ambiguity due to a missing auxiliary
(with the participles (?) passed and n-less falle[n]), neither of which is very
certain:
(228) A prest of Norffolk þat spak with yow in Jull or August last passed told me
þat … [TO AN UNIDENTIFIED LAWYER IN ROME: DRAFT 1425, 11, 05]
200 chapter 5
(229) where þe seyd William þe tyme of þe seyd enformacion was, wyth Ser John
Jermy, knyght, and othre of þe counseill of þe seyd Duk of Norffolk in hys
lordshipes in Norffolk and Suffolk þanne to hym falle by þe deth of þe right
worthy and noble lady hys modyr [MEMORANDUM TO ARBITRATORS
1426–7]
In sum, this correspondence from the early 15th century confirms my findings
above in terms of the relative share of the uses of be and have, as well as the in-
significant part that ambiguity seems to play (cf. esp. the similar percentages in
the contemporary tables for Chaucer and the Second Shepherds’ Play). I now
move on to the correspondence of William’s grandson.
table 6 Perfect, passive, progressive and modal auxiliaries in the letters of William
Paston III
relative shares of the auxiliary uses of the two verbs under investigation. The
proportions are strikingly similar to everything analysed so far. It is once again
easy to discern the proliferation of functions mentioned above (also compared
to the correspondence of the earlier generation of Pastons), though the hege-
mony of perfect have and passive be remains undeniable. It should be noted
that ‘hot news’ contexts typical of letters might favour the use of the present
perfect, but its numbers do not seem to be significantly greater than they are
in other types of comparable text. There are 35 have-perfects of transitive or
copular verbs here, including auxiliary be. The perfect auxiliary and the past
participle are not normally separated by the object, as in (230), with only one
exception, (231):
(230) letynge yow wete that I haue reseuyd of Alwedyr a lettyr and a nobyll in
gowlde þer-in. [TO JOHN PASTON III 1478, 11, 07]
(231) Also ȝe sende me worde in the letter of xij li. fyggys and viij li. reysons.
I haue them nott delyuered, but I dowte not I schal haue [TO JOHN
PASTON III 1479, 02, 23]
The example in (231) might be a curious stative, i.e. ‘I don’t have them in my
possession yet, in a delivered state’ or a have-based passive; it also illustrates
an auxiliary in an elliptical construction. A PP complement (as opposed to a
direct object) is illustrated in (232):
(232) and they ferde as thow they wolde not haue medylde wyth them
[TO JOHN PASTON III 1488, 05, 13]
In the earlier discussion (see Sections 5.4.2 and 5.5.2), I cast doubt on the vi-
ability of some be-perfects due to anomalous temporal reference in a context
where no anteriority is entailed, e.g. I must be gone [to Whitehall] (attested in
the 16th and 17th centuries), with be gone behaving like a single compound
verb. Here I find confirmation that time-reference issues can arise with have-
perfects too (already mentioned in passing while discussing example (154)
from Section 5.5.2 above). According to one possible reading of (233) below,
he sent his lordship a sum of money via a servant [in order] to have excused
him[self], but the perfect is not strictly necessary; [in order] to excuse him[self ]
would do just as well.
(233) He sent my lorde be a seruaunt of hys xl li. to haue excusyid hym, and it
wolde not be takyn; [TO JOHN PASTON III 1489, 05, 06–10]
202 chapter 5
(234) in hope þat they schuld haue ben lycensyd to haue gone ouer
[TO JOHN PASTON III 1488, 05, 13]
It turns out that both auxiliaries face questions of defective temporal mark-
ing. However, while must be gone has lost all traces of anterior meaning and is
hardly a perfect any more, seemingly redundant have-perfects still signal
unreality or anteriority, albeit vacuously, arguably retaining more of their
functionality.
There were six have-perfects with intransitive verbs, demonstrating the dis-
placement of be by have in this domain. Four of them are with the verb go,
(234)–(237), and one with come, (238) – those could hardly have caused much
ambiguity if constructed with be. There is also one with sail, (239).
(235) where as it was seyde that my lord Woddevyle and other schulde haue
gone ouer in-to Breten to haue eyded the Duke of Breten [TO JOHN
PASTON III 1488, 05, 13]
(236) and soo whan he was countyrmaundyd thos that resortyd thedyr
to haue gon ouer wyth hym taryde there styll [TO JOHN PASTON III
1488, 05, 13]
(237) And also I had went to haue had folkys a mette wyth me at Hedyngham
whyche dednott. [TO JOHN PASTON III 1492, 02, 18]
(238) how that I was sore disesyd, notwythestondyng I was welewyllyd to haue
come to fulfyll my promesse [TO JOHN PASTON III 1492, 02, 18]
(239) And they had nott seylyd not paste vj legys butt they aspied a Frencheman
[TO JOHN PASTON III 1488, 05, 13]
As pointed out already (see Chapter 3) and aptly illustrated by the material
here, contexts of modality and non-finiteness favour have over be. The same
abnormal temporal reference is detectable in ‘resorted to have gone over’ in
(236), and in most of the others. (237) and (239) have a past perfect, another
context which paved the way for the displacement. It is possible, however, that
The perfect from Middle into Modern English 203
went might be a form of the verb wenen ‘to believe, opine, suppose’ (see http://
quod.lib.umich.edu/cgi/m/mec/med-idx?type=id&id=MED52207, accessed
on 26 August 2016). This much is suggested by the fact that the past participle
of go used by William Paston III is otherwise gone, not went.
There were four unambiguous be-perfects, three of them with the resilient
verb come:
(240) there was ij c of them that gete them in-to a Breten schyppe þe whyche
was late come ouer wyth salte [TO JOHN PASTON III 1488, 05, 13]
(241) we wryte vnto you of them as thay [i.e. ‘tidings’] be comen to oure know-
lage, and as foloueth. [TO JOHN PASTON III, SENDING COPY OF A
LETTER FROM HENRY VII TO THE EARL OF OXFORD 1489, 04]
(242) And where as ye desyre that I schulde sende yow worde of suche tydyng
as Phylyp Lewes and Wyndesor bryngythe fro the corte, they be come
thens bothe [TO JOHN PASTON III 1489, 05, 06–10]
(243) and Phylyp Lewes is redyn ageyn to the Kyng, and schall brynge wyth
hym money for all ther wagys that schall be in my lordys retynew
[TO JOHN PASTON III 1489, 05, 06–10]
It is worth noting that (240) takes be even in the pluperfect. (241) uses come in a
metaphorical, non-spatial sense. Crucially, (243) has a less usual verb – to ride,
pointing to a certain productivity of the be-perfect. And yet, it might again be
tentatively suggested that the advanced rate of displacement here might be
due to geography and (earlier) contact. The Pastons come from Norfolk, which
is an area settled by Viking invaders – note the use of ON they, them, their in
(240)–(243) (though, of course, those Scandinavian forms had been diffused
to most areas by the 15th c. and the earlier generation of Pastons has OE he
‘they?’ and hem, e.g. in (224) and (217), respectively). Perhaps more tellingly, the
ON preposition fro ‘from’ shows up in (242), though alongside the Southern/
Midland -eth ending for the third person singular and plural present. Like the
Peterborough Chronicle and Secunda Pastorum, the Pastons’ periphrastic out-
put appears innovative in comparison with later texts such as The Country Wife
(cf. Rainer 1989 for the innovative use of the perfect in letters from the north).
This fits the contact scenario outlined in earlier chapters, with the Pastons
being on the border of the northern territory, which seems to be reflected in
their somewhat mixed language. The data on the rates of be-perfect loss in my
samples consistently points to a diffusion of this change from north to south.
204 chapter 5
Perfects aside, arguably there were four examples of nascent causative have:
(244) And where as ȝe reqwere to haue it grauntyd wyth-owt fyne or fee, I wold
ȝe had it wyth the payment of þe fyne or fee [TO JOHN PASTON III 1487,
00, 307]
(245) And he had many lordys and gentylmen to aunswere for hys trowthe
and hys demenynge afore-tyme [TO JOHN PASTON III 1487, 00, 307]
(246) Also he schewyd me that Herry Wyott wholde fynde the mene to haue yow
condemnyd and recouer þe obligacion of xl li. ageyns yow. [TO JOHN
PASTON III 1488, 05, 13]
(247) And also I had went to haue had folkys a mette wyth me at Hedyngham
whyche dednott. [TO JOHN PASTON III 1492, 02, 18]
(248) And if Bayard be on-solde I pray yow late hym be made fatte ageyns the
Kynge come in-to the contré, what so euer I pay for the kepyng of hym.
[TO THE BAILIFF OF MAUTBY 1487]
The functions of be have been enriched with the advent of the progressive, of
which there was no trace in the correspondence of William’s grandfather, but
there are now four occurrences. And yet, the example in (249) demonstrates
that both passive progressives and passive perfects are still unusual:
The perfect from Middle into Modern English 205
(249) hys tentys and alys [alleys? balustrades?] be a-makyng [instead of later
‘are being made’] faste and many of them be made [instead of later
‘have been made’]. [TO JOHN PASTON III 1492, 02, 18]23
There was, however, a progressive following modal will, (250), and just one
instance of a modal be structure, (251).
(250) soo that be lykelyod hys grace wolbe goyng sone vpon Ester. [TO JOHN
PASTON III 1492, 02, 18]
(251) ye be xxti s. in hys dette, for a monthe was to pay for when [deleted in
MS] he had mony laste [TO JOHN PASTON III 1478, 11, 07]
(252) he aunswerde me that he wott well ther schall be non grauntyd of a gode
whyle [TO JOHN PASTON III 1487, 00, 307]
(253) for I tolde hym how that ȝe were begylyd whan ȝe were at London.
[TO JOHN PASTON III 1487, 00, 307]
(254) that þe towne schalbe dronkyn drye as Yorke was [unclear] whan the
Kynge was there. [TO JOHN PASTON III 1487, 00, 307]
23 Although ‘be made’ can be paraphrased with ‘be ready’, I still regarded this as a bona fide
passive, in view of the ‘perfect passive context’ and, most of all, in order to keep the cat-
egory of genuine passives as inclusive as possible.
206 chapter 5
(255) The garnison of the towne of Concarnewe, which is oon of the grettest
strenghes of all Bretayn, was besieged in like wyse and drevyn to that ne-
cessité that thei wyth-in offerid [TO JOHN PASTON III, SENDING COPY
OF A LETTER FROM HENRY VII TO THE EARL OF OXFORD 1489, 04]
In (256), the by-phrase disambiguates one of the instances (‘is agreed’), where-
as comynd might be a present participle (‘coming’?) and therefore progressive,
or it might be passive (‘communed’?). A passive is suggested by the use of -ing
for the present participle elsewhere and by the preposition at, as opposed to to.
Communication and discussion meanings are also supported by the context.
(256) Syr Herry Heydon schewyd me that it is agreyd be Syr Edmond Bedyngfeld
that the mater betwyx hym and my brodyr Yeluerton schalbe comynd at
Norwyche, and there a dyreccion to be takyn in the same mater mete for
them bothe. [TO JOHN PASTON III 1492, 02, 18]
Again, most of the ambiguity had to do with genuine passives vs. copular
statives – 16 such cases (see Appendix 7 for examples).24 There were only three
potentially ambiguous be-perfects. The first is only marginally appropriate as
an illustration of ambivalence:
(257) latynge yow wete that, as for John Petyrsons mater, as ȝett it is nott
spedde. [TO JOHN PASTON III 1487, 00, 307]
The Middle English Dictionary gives the following explanation of the central
meaning of the verb: ‘to achieve one’s goal, accomplish one’s purpose; fulfill
one’s expectation; also, refl. be successful’ (http://quod.lib.umich.edu/cgi/m/
mec/med-idx?type=byte&byte=189697457&egdisplay=open&egs=189714944,
accessed on 25 August 2016). There are plenty of examples there in which speed
forms its perfect with have in this first sense (cf. (261) below too). Speed also
takes have in its second (intransitive) sense of ‘fare, get along’, as confirmed by
the dictionary corpus:
24 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
The perfect from Middle into Modern English 207
(258) How þou hast sped, sey me now. [a1400 (c1303) Mannyng HS (Hrl 1701)
7798; http://quod.lib.umich.edu/cgi/m/mec/med-idx?type=byte&byt
e=189697457&egdisplay=open&egs=189714944, accessed on 25 August
2016]
(259) Somme wer glad þat Iason hath sped wel. [c1425 (a1420) Lydg. TB
(Aug A.4) 1.3513; http://quod.lib.umich.edu/cgi/m/mec/med-idx?type
=byte&byte=189697457&egdisplay=open&egs=189714944, accessed on
25 August 2016]
(260) By þat tyme I sall wete þe at saye How þat I hafe spedde. [c1440 (a1400)
Eglam. (Thrn) 630; http://quod.lib.umich.edu/cgi/m/mec/med-idx?
type=byte&byte=189697457&egdisplay=open&egs=189714944,
accessed on 25 August 2016]
This verb is transitive in a third sense of ‘to grant, fulfill (a prophecy, etc.)’, as
in (261)–(263). In (261), mater acts as an object (cf. (257)). Speed is also attested
in the passive phrase he shal be sped ‘his affairs shall be concluded’, while
(262)–(263) are passives with inanimate subjects and are structurally analo-
gous to (257).
(261) The legatis took her leue, as þouȝ þei had sped a grete mater.
[(a1464) Capgr. Chron. (Cmb Gg.4.12) 143/14; from the Middle English
Dictionary]
(262) Youre desyre shall be spedde. [(a1470) Malory Wks. (Win-C) 413/29;
from the Middle English Dictionary]
(263) This prophecye is now spad. [?a1475 Ludus C. (Vsp D.8) 148/58; from the
Middle English Dictionary]
This means that (257) must be passive, with (261) representing its active
counterpart. Nakamura (1987: 41–42, 48) confirms that the mutative intransi-
tive verb speed only occurs with perfect have in Samuel Pepys’s Diary. I found
no be-perfects for the intransitive senses listed in the second edition of the
OED, and this item is given as ambiguous here just because it has both tran-
sitive and intransitive uses. Taking all of the evidence into account, this first
case of ambiguity is highly suspect, as is the same verb in earlier samples (see
Section 5.4.2). The other two cases of ambiguity in the current correspondence
sample, (264)–(265), are much more convincing:
208 chapter 5
(264) As for my lord Treserer, he was not wyth the Kyng of all the counsell tyme,
þe whyche was endyd on the iijde day of Marche. [TO JOHN PASTON III
1487, 00, 307]
(265) my cosyn John Heydon tolde me that þe Prior of Waburnes horse was ri-
ally amendyd, and that þe Abott of Seynt Benetys schewed hym ther was
a bay hors off a parsons nygh onto Seynt Benetis [TO JOHN PASTON III
1492, 02, 18]
The likes of (264) have been resurfacing continuously ever since the survey of
Chaucer (and even earlier); so if this pattern was so detrimental to commu-
nication, it should have been discarded by now, rather than coming up again
and again. (265) is an ambiguous be-perfect because amend can have a transi-
tive meaning, ‘to restore (sb.) to health, make well; cure (a disease, etc.)’, or an
intransitive one, ‘to recover from illness, get well’; the ME phrase ben amended
is explained as ‘be cured, be well’ (https://quod.lib.umich.edu/cgi/m/mec/
med-idx?size=First+100&type=orths&q1=amenden&rgxp=constrained,
accessed on 17 Jan 2017).
More radically, both purported illustrations of ambiguity caused by auxil-
iary ellipsis should arguably be removed. In the first one, (266), come is not
terribly clear and against probably has the meaning of ‘in preparation for/in
relation to/regarding’ (cf. a parallel etymology for vis-á-vis), so a perfect might
make no sense, though (266) could potentially be read as ‘made fat in prepara-
tion for the King [who is/has] come into the country’ (or ‘against/in prepara-
tion for the King’s coming/arrival?’).
(266) And if Bayard be on-solde I pray yow late hym be made fatte ageyns the
Kynge come in-to the contré, what so euer I pay for the kepyng of hym.
[TO THE BAILIFF OF MAUTBY 1487]
As regards (267), all the signs, especially the final PP adjunct of time, point to
preterite forms with specific past time reference:
(267) The Lord Malpertuis, now late wyth vs in ambassade from oure dere
cousine the Duchesse of Bretayne, shippid at oure porte of Dortmouth
and arriued at Saynt Powle de Lyon in Bretayn on Palmesonday at iiij
aftere noone [TO JOHN PASTON III, SENDING COPY OF A LETTER
FROM HENRY VII TO THE EARL OF OXFORD 1489, 04]
The perfect from Middle into Modern English 209
Finally, in (268), there is no compelling reason (perhaps apart from the ad-
verb last) to consider the adjective (?) past as verbal (i.e. standing for ‘passed’),
and past has accordingly not been counted. Elsewhere in the letters, this type
of past corresponds to the clearer adjective next (but compare a different de-
cision regarding passed in (228), where the (unreliable?) spelling is arguably
‘more verbal’, though hardly making that token of passed more convincing as
an actual verbal form – probably they should both be excluded).
(268) Howbeyt they haue made proclamacions in þe cuntrey to mete wyth oder
of ther affynyté as on Tuesday last past [TO JOHN PASTON III 1489, 05,
06–10]
For the sake of comparison, in (269) I provide an example from Pepys’s Diary,
where past appears as a verb or adjective in a perfect or adjectival construc-
tion (see also Rydén and Brorström 1987: 133–140). It is now time to move on to
Pepys himself.
(269) We fearing the canonicall hour would be past before we got thither,
did with a great deal of unwillingness send away the Licence and
wedding-ring. [Samuel Pepys, Diary, 31 July 1665, quoted in Nakamura
(1987: 39)]
been rather innovative in at least some of his usage (though perhaps not of the
perfect) compared to more ‘serious’ genres (Nakamura 1987: 26).
However, it has to be noted that the Diary was originally written in short-
hand and it was only later decoded by scholars. There arise some non-trivial
issues pertaining to this – Pepys’s shorthand is not as precise as a modern
counterpart would be, which can sometimes result in ‘a degree of ambiguity as
to meaning’ (Sim 1997: viii; see also Nakamura 1987: 33 for issues to do with lack
of clarity in perfect constructions due to the original shorthand). Furthermore,
the Diary has already been the subject of pertinent research into the use of
have-perfects with mutative verbs.
Investigating the Diary and confirming earlier studies, Nakamura (1987: 23,
49, cf. Rydén and Brorström 1987: 10) has found little evidence for significant
change in the formation of the perfect of mutative verbs for almost four cen-
turies from Chaucer’s time to the end of the 17th century; consequently, the
switchover should be dated to the 18th century.25 In Nakamura’s publication
readers can find valuable data for a range of periods comparable to mine, in-
cluding statistics on Chaucer, Shakespeare and the Pastons. Nakamura (1987:
48) concludes that Pepys’s usage of the perfect with mutatives is more or less
the same as that during the periods from Chaucer to Shakespeare – with the
exception of the verbs walk, proceed, escape, speed, ride, pass and meet, the be-
perfect is preferred. This ascertainment of stability lasting for several centuries
has been corroborated here (in the samples examined so far). In addition, I
have discovered no great fluctuations in ambiguity.
In view of the original shorthand and the previous research on the Diary, it
is more expedient to instead focus on some of Pepys’s correspondence, which
will additionally make it easier to compare this Restoration correspondence
sample with the results obtained from the Pastons, as well as with Nakamura’s
findings based on the Diary.
25 Due to space and time constraints, confirmation with 18th-century material will have to
await future investigation; see especially the book-length study by Rydén and Brorström
(1987).
The perfect from Middle into Modern English 211
Table 7 Perfect, passive, progressive and modal auxiliaries in the letters of Samuel Pepys
and John Evelyn
sample is smaller, approximately 3,651 words, but it yielded about the same
amount of material due to the more advanced development and diversifica-
tion of the auxiliaries under investigation. The figures in Table 7 are again simi-
lar to everything seen so far.
There was a total of 39 have-perfects with transitive verbs and copular or
auxiliary be. In none of these was the object inserted between the auxiliary
have and the past participle. Again, a minority contained prepositional objects
or were used without an overt object, as in (270):
(270) Take notice also, I pray, how few have miscarried [Most of those who
dyed perished for want of covering] [E to P, Saturday 14 October 1665]
Miscarry can in fact be an intransitive verb here, which in ME and EModE had
senses like ‘suffer harm’, ‘come to grief’, ‘perish, die’, or (later) ‘fail to reach its
proper destination’ (Rydén and Brorström 1987: 132, cf. the OED). There were
another three have-perfects with intransitives. One of them is with the raising
verb seem, which does not really qualify as mutative:
212 chapter 5
The other two, (272)–(273), featured come in the present as well as in the past
perfect, which would have been completely unambiguous with be.
(272) For finding divers Chyrurgeons, and Sick-persons at my dores who had
come from Several places with sad complaints that they could not procur
quarters for them [E to P, Tuesday 3 October 1665]
(273) What has come collateraly on you (not through my fault) ought not be
imputed to me [E to P, Tuesday 3 October 1665]
(274) (long before this as having prospect sufficient of what is befallen us)
[E to P, Tuesday 3 October 1665]
Although us may look like an object, (274) is probably the result of confusion
with mutative intransitive fall, of which befall is a derivative, or with the obso-
lete intransitive sense of befall as ‘fall’ (see OED entry for befall, as well as Petré
and Cuyckens 2008 for the properties of the prefix; cf. Rydén and Brorström
1987: 23–24, 53, 81–82, for such patterns involving ‘quasi-objects’ of verbs like
befall). (274) can only be a perfect and no passive, since no argument is missing
or has been promoted or demoted – there is a subject (thematically a force or
cause) and a quasi-object (patient/affected, or better a ‘deep-structure’ loca-
tive adverbial/adjunct or goal complement, i.e. ‘something has fallen/come
(up)on us’, as suggested by Rydén and Brorström 1987: 23, 81). Crucially, both
participants in the event are there. This can be compared to an unusual pas-
sive, in (275), which also looks like it has an extra object (again a thematic
locative/source), but the subject is actually a promoted patient/affected object
from the active and the agent is missing altogether. In the entry for the verb
discharge, the OED supplies more examples of the omission of the preposition
from, which was possible in earlier periods and resulted in a second ‘locative
quasi-object’ for discharge.
(276) a Shipp of ours (the little Guift) that in a Conflict with a Hollander on
the Irish Coast (wherein shoe though much over matched hath acquitted
her selfe very well) hath had severall Men wounded [P to E, Thursday
27 April 1665] [i.e. ‘several men were/got wounded’]
214 chapter 5
(277) at Chatham, and Gravesend, where I was threatnd to have our sick all
expos’d [E to P, Saturday 23 September 1665] [i.e. ‘that our sick would
all be/get exposed’]
(278) and these coming under my particular cognizance, as I have had (to
the greate increase of my Trouble) the Hospitals of London to look after
during the absence of my Brother Commissioners [E to P, Thursday
12 October 1665]
(279) By what I have sent you, you will have a Specimen of the Method
observed where I have any-thing to do [E to P, Saturday 14 October 1665]
The first one seems like a potential candidate for causative have, if observed is
construed as related to have, i.e. ‘have this observed, make sure it is observed’.
However, observed might also be a (reduced relative) post-modifier of Method,
i.e. you will receive a specimen of the method which is observed (the latter is
far more likely). Crucially, this once again shows have in an ambiguous context.
The second construction, with the clause-final infinitive, might conceivably
carry a modal meaning comparable to the one in (278). Alternatively, it is just
the have-something-to-do-(with) pattern.
The functions of be have been enriched with five instances of the progres-
sive and five more modal (deontic) uses with a to-infinitive (some of which
probably expressed futurity more than anything else – see Appendix 8).26 In
one of the examples of the progressive, the present participle was conjoined
with an adjective (albeit with an ultimately present participial etymology in
26 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
The perfect from Middle into Modern English 215
(280) I was forc’d to dispatch Warrants to the Connestables and other Officers
to be ayding and assistant to my Deputyes [E to P, Tuesday 3 October
1665]
(281) where there will be care taken for them [E to P, Saturday 23 September
1665]
About four such instances were found, which might in fact contain simply an
existential there construction with copular be and a participial post-modifier
to the noun. This much is suggested by (282), where the post-modifier is clearly
adjectival:
(282) there were none of those Vessells ready you were pleas’d to name [E to P,
Tuesday 3 October 1665]
Essentially, this reaffirms that ambiguity is inherent in the passive too, as also
demonstrated by the fifteen clauses which could be passive as well as stative.
(283) should be treated as exclusively passive, however:
(283) They all this while concealing their having pensions, enjoy the Kings
Super-allowance in the Hospitals, which formerly was not continud
[E to P, Thursday 12 October 1665]
Formerly and the context indicate past time which favours a preterite reading –
this non-specific, generic ‘beforeness’, however, is not the type of past time that
is wholly incompatible with the perfect. Be that as it may, continue probably
never combined with be to form its perfect. Even the intransitive senses of
continue in the second edition of the OED contain have-perfects only. Eight
of the fifteen clauses in my dataset which are passive or ambiguously stative
involve the formulaic construction be pleased, as in (282) or (284), which might
be merely adjectival:
216 chapter 5
(284) and what his Majestie shall be pleas’d to superadd to his other Commands
[E to P, Monday 1 May 1665]
(286) For prevention whereoff you may be pleas’d to order that none be
admittd from any our Infirmitories into Pensions [E to P, Thursday
12 October 1665] [= ‘it may please you to order …’?]
(288) and some of these concernd me as far as Deale and Sandwich, where we
are so overlayd, that they send them back upon us, and they perish in the
returne [E to P, Tuesday 3 October 1665]
Overlaid evidently stands for ‘overwhelmed [with sick and wounded soldiers
and prisoners]’. So might additionally suggest adjectival status for overlaid.
Cure is also normally transitive, so (289) is unlikely to be a perfect, not least
because of the hospital context too (though the OED lists an obsolete intransi-
tive sense of ‘be cured, get well again’, attested between the 16th and the 18th
centuries, which is marked as rare).
(289) First, Our men in the London Hospitals steale downe to Chatham before
they are Cur’d [i.e. ‘healthy again’] [E to P, Thursday 12 October 1665]
Even if (289) were treated as potentially perfect (though this would be hardly
justifiable), the statistics would remain more or less the same. There was no
The perfect from Middle into Modern English 217
(290) From a letter this day come to my hand from a Shipp of ours [P to E,
Thursday 27 April 1665]
(291) but how shall we (when recoverd) secure them from running away?
[E to P, Saturday 23 September 1665]
5.8 Epilogue
As noted throughout this chapter, although the unclear cases I have recorded
are not that numerous, at the same time the availability of multiple readings
does not seem to have been detrimental or undesirable in discourse. In order
to illustrate once again that speakers throughout the history of English could
tolerate and obliviously live with a large amount of ambiguity in close proxim-
ity, I give a final quotation from Shakespeare (cf. Chapter 3, Section 3.4.4):
(292) The bonds of heaven are slipp’d, dissolv’d, and loos’d [Shakespeare,
Troilus and Cressida V.ii, line 156]
(293) By that time I had made up my mind that the period was come when I
must tell my guardian what I knew. [1853, Charles Dickens, Bleak House,
Chapter XLIII]
27 A similar case is provided by the version of the Lord’s Prayer in Ayenbite of Inwyt, a Kentish
text from 1340. The same token of by ‘be’ applies to y-halȝed ‘hallowed’, an adjective or a
transitive past participle, cominde ‘coming’, a present participle, and y-worþe ‘become’, a
typically intransitive past particple with the prefix y- (from OE ge-), along with loss of
final -n:
(i) Vader oure þet art ine hevenes, y-halȝed by þi name, cominde þi riche, y-worþe þi wil …
[1340, Ayenbite of Inwyt, quoted in Baugh and Cable (2013: 417), emphasis mine]
The perfect from Middle into Modern English 219
At the same time, the narrator of Bleak House opts for non-finite have in
I should have come, as well as plurperfect she had come elsewhere (both from
the same chapter).28 Double marking is in evidence too: should have been gone
in another hour (ibid.), indicating that be + gone is a copular structure with a
participial adjective. Nevertheless, Chapter LVII has non-standard after they
was gone to bed, with a to-PP, said by a police officer and presumably suggest-
ing that (vestigial) be-perfects were still part of colloquial English, at least to
a certain extent. In addition, there is the following example with a missing
auxiliary:
(294) Some of these vessels were of grand size: one was a large Indiaman just
come home [1853, Charles Dickens, Bleak House, Chapter XLV]
The overall distribution of the be- and have-perfect with the verbs come and go
in the 19th century is discussed by Rydén and Brorström (1987: 66–71, 109–115),
including Dickens. Be come and have come receive an average of 21% v. 79%,
respectively, for all authors, but with only 4.6% for be come in Dickens’s let-
ters. Dickens likewise uses have gone (as opposed to be gone) 76% of the time
in his correspondence. By the Victorian period, ambiguity ‘should’ have argu-
ably disappeared with the almost complete disappearance of be as a perfect
auxiliary, especially in the language of a progressive user like Dickens (with
87% of have with intransitives in his correspondence, according to Rydén and
Brorström 1987: 197, 203), but there are instances in Dickens which confirm
that it persisted nevertheless. This makes it difficult to claim that the demise
of the be-perfect had to do with streamlining the system and getting rid of un-
clear cases. Are burst open, was passed and am changed in (295)–(297) below
can be perfect or passive/stative. It is also worth noting the neighbouring have-
perfect as well as the clear be-passive at the end of (295) – even in the 19th cen-
tury they all seem to coexist happily together, unperturbed by considerations
of what is ‘functionally desirable’. In the context of (296), we would probably
use had (been) passed today. The adverb in how changed might further suggest
adjectival status for the participle in (297).
28 It is noteworthy that Dickens also opts for had when a past event is clearly over and does
not continue into the past reference moment, as in (i). He was come would probably have
implied that he was still there at the time (cf. Smith 2001: 369; as well as the Modern
English resultative stative is gone, which can no longer indicate an anterior event that
does not go on into the present: *He is gone and then is back again vs. He has left and then
has come back again, Elliott 2001: 63).
(i) We agreed that he was either prevented from coming, or that he had come, and gone
away; [1853, Charles Dickens, Bleak House, Chapter LXI]
220 chapter 5
(295) then upon my honour, upon my life, upon my reputation and principles,
the floodgates of society are burst open, and the waters have – a – oblit-
erated the landmarks of the framework of the cohesion by which things
are held together! [1853, Charles Dickens, Bleak House, Chapter XL]
(296) I could not venture to approach her, or to communicate with her in writ-
ing, for my sense of the peril in which her life was passed was only to
be equalled by my fears of increasing it. [1853, Charles Dickens, Bleak
House, Chapter XLIII]29
(297) I know full well how changed I am [1853, Charles Dickens, Bleak House,
Chapter LXI]
29 Chaucer’s sentence containing is passed in Section 5.2.2 above was treated as an unambig-
uous perfect because of syntactic clues; no such clues are present in Dickens’s example.
The perfect from Middle into Modern English 221
5.9 Conclusion
This chapter has documented the competition between the two perfect auxil-
iaries in earlier English written samples which can be argued to approximate
the spoken medium, showing in ample detail how have gradually came to be
selected as the only auxiliary for the perfect, ousting be from that domain (but
only after centuries of ‘resistance’). The material from a collection of represen-
tative texts (14th to 17th century) demonstrates that ambiguity should not be
considered an important contributing factor in this process. Most instances
of replacement of be by have with intransitives would have been unambigu-
ous with be anyway. Furthermore, the syntactic and semantic context is often
sufficient to disambiguate, attenuating any prospective ‘communicative dam-
age’ arising from the availability of multiple interpretations (perfect, passive
or stative). There are even indications that the availability of multiple inter-
pretations was actually exploited as creatively desirable, rather than avoided.
Supposedly ill-suited hybrid perfects/passives such as X is begun or ended keep
turning up and survive until the 19th century, even after the almost total oblit-
eration of be as a perfect auxiliary. If perfect be was dispreferred because of its
inherent ambivalence, these should not have enjoyed such remarkable lon-
gevity. The greatest amount of ambiguity in all periods studied is between a
dynamic passive and a copular stative reading (e.g. PDE He was frightened),
by far outnumbering ambiguous perfects, but this has not harmed passive be’s
chances of survival. Neither do I record any imminent ‘relief’ from passive get.
In terms of frequency, the statistics confirm that have has always been pre-
dominant in the perfect and be in the passive. Moreover, perfects of transitive
verbs are much more common than intransitive ones.
The diachronic statistical trends are also problematic and inconclusive.
Ambiguous be-perfects account for 3.6% of all recorded auxiliaries in Chaucer’s
selected texts (plus 0.6% due to missing auxiliaries), 1.9% in Secunda Pastorum,
4% in Shakespeare (+ 1.5% due to missing or contracted auxiliaries) and 0.7%
in Restoration comedy (+ 2.6% due to missing or contracted auxiliaries), so
Shakespeare disrupts the pattern. No clear trend crystallises after collating
the percentages from ambiguous be-perfects and ambiguity due to contrac-
tion or omission of the auxiliary. If the categories are kept apart, ambiguous
222 chapter 5
In Chapters 2–5, I defended the idea that a good deal of language change is
simply random, in the sense of non-functionally motivated. Here, I return to
this claim by turning the spotlight on two Germanic varieties (English and
German) which took a very different course from each other, despite shar-
ing the same point of departure and despite their near-identical ‘grammati-
cal ecologies’. Chapter 3, Section 3.4.3, mentioned the loss of the Old English
weorðan-passive, which may have conspired to overload passive be and push be
out of its perfect role. It was stated there that this never happened in German,
hence both types of perfect survive. Things are hardly likely to have been as
simple as that (and it remains mysterious why OE weorðan disappeared but
German cognate werden did not).
In Section 6.1 below, I introduce the German auxiliary system, which is in-
deed very much reminiscent of that of earlier English outlined in Chapters 3–5.
I also emphasise that the potential for misunderstanding existed in German,
just as it did in English. In Section 6.2, I demonstrate that lack of clear mark-
ing is widespread in a German text from the 18th century comparable to the
English ones above. And yet, German has not experienced loss of perfect be,
although the same purported functional triggers were there. Granted that the
two languages started from a similar starting point, they must have come out
so differently with regard to their be-perfects either because of the essentially
arbitrary nature of language change and/or because of the different historical
pressures and contacts that they experienced.
6.1 Introduction
As noted in the opening paragraphs, this chapter aims to take a cursory glance
at the way German works in comparison with English. The two languages
are closely related – both of them belong to the West Germanic group of
Indo-European. Investigating a Modern German text within the same frame-
work as the English samples above can therefore provide an insight into how
things might have looked in English if it had preserved both types of perfect, as
well as its weorðan-passive.1
1 An overview of the Modern German temporal system can be found in Eisenberg (1994,
2004: 106ff.). A brief survey of the history of the German perfect is available in Rydén and
Modern German builds its perfect with haben ‘have’ or sein ‘be’ in combina-
tion with a past participle. In addition, the perfect has now taken over as a past
tense in that it is compatible with specific past time reference, such as gestern
‘yesterday’ (for more details, see Smirnickaja et al. 1977, Zifonun et al. 1997:
1686ff., 1702–1708, 1711, Atanasova and Sugareva 2004: 32–34, Eisenberg 2004:
108, 110–114, esp. 113, Macleod 2012: 241–242, Gillmann 2016). Haben is used
for the perfect of all transitive and reflexive verbs, including transitives with
implied objects, impersonal verbs (Es hat geblitzt und gedonnert. ‘There was
thunder and lightning’ – Atanasova and Sugareva 2004: 26), as well as modals.2
As shown in (1) and (2), haben has also been extended to intransitives which
do not express movement or a change of state, i.e. non-mutatives such as sit-
zen ‘sit’, stehen ‘stand’, liegen ‘lie’ etc., which were once compatible with either
auxiliary (see Rydén and Brorström 1987: 19, Eisenberg 2004: 108–109; as well
as Drinka 2017: 241, for exceptions in modern southern German, where be is
preferred for these verbs).
Brorström (1987: 18–19, with references), while Drinka (2017: Chapter 9) outlines the situa-
tion in Old High German and other early Germanic dialects. Petré (2013) is a comparative
diachronic study of the passive in English and German. Modern German (stative) passives
and perfects are compared to earlier English by McFadden and Alexiadou (2010).
2 As is the case in English, there are related stative forms with the past participle, such as (i):
(i) Das Pferd hat die Fesseln bandagiert. [from Zifonun et al. (1997: 1792)]
the horse has the pasterns bandaged
‘The horse has its pasterns bandaged’
These are not always distinguishable from the perfect and I return to them at the end of this
section (see Kostov 1982: 162ff.).
3 Unless otherwise specified, the glosses and the translations in this section are mine.
a Glance at German 225
Some verbs of motion can take haben if the emphasis falls on the activity
itself, including its duration, (5), rather than on the goal or destination, (6)
(Atanasova and Sugareva 2004: 27, Eisenberg 2004: 109; cf. Los 2015: 76–77 for
Dutch and earlier English, as well as Chapter 5, Section 5.5.2, example (167),
and elsewhere, for similar English material).
4 Werden is also an auxiliary for the future in German (see Zifonun et al. 1997: 1688–1689,
Atanasova and Sugareva 2004: 28, Eisenberg 2004: 107), in addition to being a lexical copu-
lar verb – wird geteilt ‘is/gets divided’ is a passive construction, whereas wird ärgerlich ‘gets
angry’ is copular (see Eisenberg 2004: 124–125).
a Glance at German 227
5 In addition, German seems to be recruiting more passive auxiliaries (see Zifonun et al. 1997:
1824–1829, Eisenberg 2004: 133–134, Askedal 2005), including bekommen and kriegen ‘to get,
receive’, which might be unexpected on a functionalist account (cf. Chapter 3). German al-
ready has two passives, so it hardly ‘needs’ another one. The new formations make it possible
to turn the dative indirect object or the sole dative complement into a subject in the passive,
as in (i):
(i) Renate krieg-t von Rainer ge-droh-t. [from Eisenberg (2004: 134)]
Renate get-pres.3sg by Rainer p ref-threaten-pptcp
‘Renate gets threatened by Rainer.’
228 chapter 6
the most beautiful stones gathered’ can get a perfect (active) reading, ‘They
have gathered the most beautiful stones here (themselves)’, as well as a passive
one – ‘The most beautiful stones have been/are gathered here for them (by
someone else)’ (Askedal 2005: 217–218). I set out to establish in my small-scale
corpus study below whether there is any ambiguity in a sample of (earlier)
Modern German, and if so, what it is.
The German play selected for this study is Minna von Barnhelm by Gotthold
Ephraim Lessing, first published in 1767. The original text used here is based
on the following electronic publication: http://www.gutenberg.org/cache/
epub/9187/pg9187.html (accessed on 12 Nov 2016). The idiomatic English
translations that appear below are those of Ernest Bell, available at http://
www.gutenberg.org/files/2663/2663-h/2663-h.htm (accessed on 12 Nov 2016),
whereas the glosses are mine.
In the preface to Bell’s translation (http://www.gutenberg.org/files/2663/
2663-h/2663-h.htm, accessed on 12 Nov 2016), the comedy is introduced as ‘the
beginning of German national drama’; it is characterised by a patriotic inter-
est in the historical background, as well as a ‘sympathetic treatment of the
German soldier and the German woman’; it has also been said to felicitously
blend ‘the amusing and the pathetic’, like its English counterparts above. In
that sense, it was deemed a fortunate choice for the purposes of comparison.
table 1 Perfect, passive, and modal auxiliaries in the first eight scenes of Minna von
Barnhelm
Due to space constraints, only the first eight scenes of Act I were examined,
a stretch of approximately 3,222 words. Nevertheless, this supplies enough ma-
terial to go by. The results are laid out in Table 1 above. For the most part they
are very similar to the ones obtained for English. By far the largest majority of
perfects are have-perfects with transitive verbs (in a broad sense). There is only
one intransitive have-perfect and only seven perfects with be. Again, mutative
intransitives are comparatively rare. Like English, German also demonstrates
the extension of the field of operation of both auxiliaries into the modal do-
main. Dynamic passives with werden are as many as stative passives with sein
‘be’, for which there are indications to assume adjectival status (see below).
The most palpable difference from the general situation in English is the total
absence of ambiguous be-perfects/statives and the large number of missing
auxiliaries, which however is in some respects similar to some of the later
English samples. Pepys’s correspondence in Chapter 5, Section 5.7.2, and The
Country Wife in Section 5.5.2, both from the 17th century, seem to have more
or less eliminated ambiguous be-perfects too, though neither of them has as
much ambiguity from ellipsis and contraction.
Starting with perfects, there are no major surprises, with the usual division
of labour between have (with transitives) and be (with intransitives), as illus-
trated in (10). As demonstrated by the material, the perfect is of course well
established by this late stage in the history of German.
(10) Was gilt’s, der Herr Major ist nicht nach Hause
what counts.it the sir Major be.pres.3sg not to house
ge-komm-en, und Er ha-t hier auf ihn
pref-come-pptcp and he have-pres.3sg here for him
ge-lauer-t? [I.ii]
pref-keep.watch-pptcp
‘What will you bet the Major has not returned home, and you have been
keeping watch for him?’
Strictly speaking, gelauert does not take a direct object, but a prepositional
complement with auf; I nevertheless group it with the transitives proper (as I
have done for English throughout). As noted in the introductory remarks, have
is used with modals, (11). Have also forms its own perfect, (12).
There is only one intransitive have-perfect, (13), with a verb that does not ex-
press motion or change of state and which also has transitive senses.
Be-perfects are comparatively rare, only seven. These include verbs for ‘come’
(kommen) and ‘fare/travel’ ( fahren), and/or their prefixed derivatives, as well
as the verb for ‘remain’ (bleiben) – see (10) above and (14)–(16) below. Bleiben,
(15), denotes lack of change of state. Arguably, the absence of change of state is
also a type of change of state (see Eisenberg 2004: 109).
The last of these, (16), has non-finite sein, and also a PP-dependent in the
clause (mit mir ‘with me’). This meaning of verfahren does not have to do with
a Glance at German 231
literal travelling and it can hardly be said to be mutative (or transitive) in this
sense.6 The verb sein ‘be’ also ‘self-sufficiently’ builds its own perfect, (17), tes-
tifying to complete grammaticalisation of sein as a tense/aspect auxiliary (un-
like OE weorðan in the combination wearð geworden ‘had happened’, English
be has always taken auxiliary have for its perfect; see Chapter 3, Section 3.4.4;
Chapter 5, example (4)).
(17) Ich wüßte mich auch nicht zu erinnern, daß er mir jemals
I knew me also neg to remember that he me ever
etwas schuldig ge-wes-en wäre. [I.vi]
something in.debt pref-be-pptcp were
‘nor can I remember that he ever did owe me anything’
In the modal realm, there are four instances of modal have, and one of modal
be. Have seems to be more wide-spread in this function. Zifonun et al. (1997:
1897ff.) discuss modal haben zu in German, where it can be synonymous with
müssen ‘must’, as in (18), or können ‘can’, perhaps as in (19).
Some readers might find the modal credentials of (19) a little dubious, but it
can be seen as modal. Zifonun et al. (1997: 1897) provide a similar paraphrase
with kann ‘can’: Sie hat immer etwas zu verschenken. ‘She always has something
to give away’ = Sie hat immer etwas, das sie verschenken kann. ‘She always has
something that she can give away’ (cf. the discussion of the grammaticalisa-
tion of English modal have to in Chapter 3). Modal be, on the other hand, has a
passive overtone, as in (20).
Eisenberg (2004: 131–132) talks about a modal infinitive and a modal passive,
including borderline cases, e.g. Karl ist leicht zu betrügen/Es ist leicht, Karl zu
betrügen ‘Karl is easy to deceive/It is easy to deceive Karl’, analogous to ex-
ample (20), which could be a modal passive or alternatively it could be seen as
an instance of ‘tough-movement’ (cf. Zifonun et al. 1997: 1897ff.). If modal, the
interpretation is one of possibility (see Zifonun et al. 1997: 1898).
By their very nature, werden-passives are usually dynamic, as in (21), while
those with sein normally express a state (see Section 6.1). There are five of each.
There are clear indications that the constructions with sein should be treated
as adjectival. Adjectival status is suggested by coordination with a genuine ad-
jective such as galant, (22), or by modifying adverbs such as stark, (23), gewiß,
(24), or sehr, (25). As was the case with English, this potential for ambiguity
does not seem to have bothered German speakers too much (additionally con-
firmed by the equal numbers of the stative and dynamic constructions).
The two most fascinating discoveries here are the total lack of ambiguous be-
perfects from the data and the numerous instances of auxiliary ellipsis – a sim-
ilar rise in the number of ellipted or contracted auxiliaries in the early modern
period was noted in Chapter 5 for English. The first of these discoveries might
prompt a functionalist to rejoice and attribute the preservation of be-perfects
to this lack of ambiguity, but the second finding is rather problematic. This
passage has many more missing auxiliaries than any other passage analysed
so far, a total of 15, half the number of transitive have-perfects, three times the
number of be-passives, and more than twice the number of be-perfects.
The first example of this kind, (26), can be interpreted in two ways due to
the missing auxiliary:
Selbst probably suggests a perfect: Haben Sie das selbst gemacht? ‘Have you
made it yourself?’ The passive/stative is not completely ruled out though – Ist
das selbst gemacht? ‘(lit.) Is that self-made?’ (28) is also noteworthy:
234 chapter 6
(28) Einem Manne wie meinem Herrn, der Jahr und Tag bei
a man like my master who year and day by
Ihm ge-wohn-t, von dem Er schon so manchen
him pref-live-pptcp from whom he already so many
schönen Taler ge-zog-en, der in seinem Leben keinen
nice thalers pref-draw-pptcp who in his life no
Heller schuldig ge-blieb-en ist [I.ii]
heller in.debt pref-remain-pptcp be.pres.3sg
‘a man like my master, who has lodged at your house above a year; from
whom you have had already so many shining thalers; who never owed a
heller in his life’
The ellipted auxiliaries with gewohnt and gezogen would have been forms of
haben ‘have’. The only one which is not omitted is the final one – a form of be
in geblieben ist. This however can hardly be said to have worked in favour of
displacing have in German, in the same way that similar arguments were made
for the displacement of be in English (see Chapter 3, Section 3.4.4).
Conversely, in (29), ist is again characteristically missing (from worden [ist]),
but a have-perfect is now present in an earlier clause – just the type of context
which can be said to favour displacing be, which however never happened in
German.
(29) Wenn ich sie nicht habe, so ist es ein Beweis, daß ich nie
if I it not have so is it a proof that I never
eine ge-hab-t hab-e, oder daß sie getilgt
one pref-have-pptcp have-pres.1sg or that it honoured
und von mir schon zurückgegeben word-en. [I.vi]
and by me already returned become-pptcp
‘If I have not got it, it is a proof that I never had it, or that it has been hon-
oured and already returned by me.’
The same kind of ellipsis can of course occur when the expected form is a form
of have, as in geteilet [hat] in (30), where both verbs essentially select the same
auxiliary:
a Glance at German 235
In (31), the perfect auxiliary can be recovered from the context – from the
stative/passive in the landlord’s turn. Importantly, this shows the multi-
functionality and unity of these words – copular, passive and perfect be are
perceived as the same lexeme (see (22) and the literature review in Chapter 3,
as well as the Epilogue to Chapter 5).
(31) Wirt: […] Das Zimmer ist doch sonst galant und
Landlord the room be.pres.3sg still otherwise gallant and
tapezier-t— Just: Ge-wes-en! [I.ii]
paper-pptcp Just pref-be-pptcp
‘LAND: […] The room is otherwise very nice, and is papered’ ‘JUST:
Has been!’
The next two examples, (32)–(33), are likewise missing their perfect have
(cf. gehabt habe in (29) above), while (34) is missing a form of perfect be:
Overall, there are many missing auxiliaries. (34) should have a perfect passive,
but the perfect auxiliary is not there (cf. (29) and the note on the avoidance
of perfect passives in Chapter 3). As mentioned already, both sein and haben
can be thus omitted: gefunden hat vs. worden ist. Morphologically, forms like
gefunden etc. can only be past participles because of the prefix ge- and the
suffix -en. Worden, albeit prefixless, is also certainly a past participle ending
in -en – -en cannot be a finite marker of preterite plural agreement as the sub-
ject welches is singular, and the ablaut vowel grade would also be amiss if this
were a preterite.
Finally, it is again not obvious whether the participle in (35) should be con-
strued as participating in a passive (ausgelegt ist/ausgelegt wurde/ausgelegt
worden ist ‘was/has been spent/laid out’) or perfect (ausgelegt hat ‘[my master]
has spent/laid out’).
Quite importantly, this example comes from a brief note written by one of the
characters. The same applies to another four. Even if these are discarded due
to the nature of the genre, the number of such ambiguity-producing omis-
sions remains astounding. As with the English material above, such ambigu-
ous contexts have persisted in German as well. A short random example from
the opening paragraphs of Chapter 3 of Remarque’s Im Westen nichts Neues
‘All Quiet on the Western Front’ (1928) is the one-word utterance Verstanden?
‘(lit.) Understood?’, ‘Got it?’ (addressed to a newly recruited young soldier in
WWI). It does not matter much whether this stands for Hast du das verstanden?
‘Have you understood that?’/i.e. ‘(Do) you get it?’, or Ist das verstanden? ‘Is that
understood/clear?’.
Although no other ambiguous cases were detected, the ambiguity fostered
by auxiliary ellipsis is still significant enough and provides solid grounds to
conclude that German is not devoid of uncertainty of interpretation and
a Glance at German 237
6.3 Conclusion
This short intermezzo between the discussion of English and the next two
chapters on Bulgarian has been intended to show that English and German
started out with the same system of perfects and passives, with parallel dis-
tributions of the various auxiliaries, as well as with the same potential for
misunderstanding. Like English, German has also seen a number of cases of
polygrammaticalisation of be and have, as well as the birth of new passive
constructions (coexisting with the original two – based on be and become).
Although I detected no ambiguous be-perfects in a randomly selected sample
of 18th-century German drama, interpretative difficulties due to omission of
the auxiliaries abounded. In all important respects, the picture is therefore
identical to the one in earlier English, including relative frequencies of the
auxiliaries. Nevertheless, German has not taken the same course, retaining
both flavours of passive and both flavours of perfect. Consequently, the (non-)
changes in these two closely related Germanic varieties cannot be due to in-
herent functional motivations, but are probably mostly arbitrary and/or partly
conditioned by external factors such as language contact.
In support of contact-induced change, Drinka (2017: Chapters 7 and 9)
notes a number of shared features characteristic of languages from the realm
of Charlemagne’s mediaeval empire, including the semantic transformation
of the perfect into a pure past, as well as an especially well developed and
maintained be/have dichotomy in the perfect. Being central members of the
Carolingian Sprachbund, German and French exhibit these developments,
whereas English, which sits on the periphery, does not. Thus, the preservation
of both auxiliaries in German is at least in part due to its core membership of
the Carolingian Sprachbund (additionally correlating with a plethora of other
linguistic and extra-linguistic factors attributable to the Sprachbund, e.g. the
reverence for and special use of Latin, for which see Drinka). While German
might have experienced areal pressure to maintain the dichotomy in the per-
fect, as a result of contact with Viking invaders English arguably came under
pressure to simplify the system by retaining only the more prominent of the
two auxiliaries, in line with analogous shifts in the other languages peripheral
to this Sprachbund.
chapter 7
It was stated in the introductory chapters that the choice of Bulgarian and
English for a comparative study has been prompted by the fact that they are
typical representatives of the Eastern and Western European methods of
building perfect constructions. Drinka (2017: Chapter 16) is a concise summary
of the east-west split in perfect formation, where readers can find more infor-
mation. In terms of structure, this split has to do with the different participial
profiles of western languages like the Germanic ones, which possess only one
all-purpose past participle, as opposed to eastern languages like those of the
Slavonic group, where two past participles are available – an active and a pas-
sive one. The existence of the two types of participle in Slavic is arguably what
enables be to function unproblematically as a universal perfect and passive
auxiliary, combining with a past active participle in the first case, and a past
passive one in the second. This is not the case in the west due to the more
limited participial inventory; hence have emerged as the dominant perfect
auxiliary, while be has always been the preferred choice for passives, with only
a secondary (or defunct) role in the perfect with intransitives, which do not
normally passivise.
Despite the traditional hegemony of be-perfects, Slavonic languages have
seen the birth of perfect-like resultatives with have too. Although functional
explanations analogous to those of English have been proposed for the evolu-
tion of have as a perfect auxiliary in Bulgarian, it will be argued below that
Bulgarian did not really ‘need’ a have-perfect (in line with the arguments from
the preceding paragraph); so its appearance must be due to the generic mean-
ing of have, a suitable and likely candidate for even greater generalisation and
hence reanalysis and grammaticalisation, most probably aided by language
contact, considering the rarity of have-perfects outside of Western Europe and
the geographical distribution outlined in the next paragraph. In addition, the
morphological agreement marking on Bulgarian participles is comparable to
that of Old English. Rather conveniently, Modern Bulgarian does not exhibit
the same type of ambiguity due to zero marking, which enables profitable ty-
pological comparison and theoretical generalisations as to the role of agree-
ment in reanalysis and grammaticalisation.
Bulgarian has an elaborate temporal and aspectual system, built around the lo-
cation of the event in time (tense), as well as its internal organisation (aspect),
incorporating distinctions such as (a-)telicity/boundedness (the presence/
240 chapter 7
For instance, the aorist of a perfective verb will usually express a finished ac-
tion in the past, as in (1). The imperfect tense of the same perfective verb will
not express action in progress in the past, but a series of complete activities,
often in a subordinate adverbial clause; it is the series as a whole for which
no temporal limits are supplied, not the individual events which make up the
series, as in (2).
There is also the category of the perfect, in all tenses and both in the perfec-
tive and imperfective aspect (see Comrie 1976: 62, 107–110, 126, Scatton 1993:
211–212). Most standard reference works on Modern Bulgarian treat the perfect
as a (type of) tense and describe its formation as a combination of a form of
the auxiliary съм/бъда / săm/băda ‘to be’ and the perfective past active parti-
ciple, which ends in -l, e.g. написа-л съм/napisa-l săm ‘I have written’, lit. ‘I am
having-written’. It is this dedicated active non-finite form that enabled ‘be’ to
function as a universal marker of the perfect, as well as the passive. The avail-
ability of this l-participle, as distinct from the past passive participle in -t or -n,
gave Slavonic languages the opportunity to create active periphrastic construc-
tions that were less easily crafted in the west, with its limited array of past
participles (see Drinka 2017: Chapter 16).2
2 On the Bulgarian perfect, consult Tilkov et al. (1983a: 318, 335, 348–350, 378–379), Nicolova
(2008: 268, 292–304, 313–316, 2017: 413ff.); also see Bojadžiev et al. (1999: 382ff., 389–392) and
242 chapter 7
The Bulgarian present perfect typically combines present and past mean-
ing, relating a past event to a present state, not unlike its English counterpart
(Comrie 1976: 52, 107). Like the English perfect, this is achieved by combin-
ing a present form of the auxiliary verb and a past (active) participle – the
present tense conveys the present relevance, while the past participle denotes
the past action. Like Germanic perfects with be, the Bulgarian construction
also originated in mutative intransitive contexts based on the /X is Y/ equation
schema with a subject complement (Kuteva 1998: 295). Notably, however, all
Slavonic varieties came to use be both for intransitive and for transitive verbs,
as in (5). The Germanic languages, including earlier stages of English, normally
show a division of labour between have and be, as discussed in earlier chapters.
As stated above, to a large extent this must be due to the different participial
profiles in the two language groups. Be could unproblematically serve as aux-
iliary for both the perfect and the passive in Slavic, since the perfect features
a specialised past active participle, while the passive has its own dedicated
non-finite form.
In the singular, the l-participle has endings for all three genders in order to
agree with the subject: masculine (купил/kupil – see (5)), feminine (купил-а/
kupil-a), and neuter (купил-о/kupil-o). The distinct gender inflections which
Old Bulgarian/Old Church Slavonic had in the plural (m.pl -и/-i, f.pl -ъi/-ъi,
nt.pl -а/-a), are now collapsed into one generalised plural form in -и/-i
(купил-и/kupil-i) (cf. (6)–(9) below).3
Old Bulgarian/Old Church Slavonic (OB/OCS) essentially had the same
perfect with бьiти/byti ‘to be’ plus the past active participle ending in -l. This
construction was not inherited from Proto-Indo-European, but appeared as a
Slavonic innovation from which the Modern Bulgarian forms are descended.4
Kucarov (2007: 249–253), where the perfect is rightly accorded its own category, separate
from tense.
3 See Nicolova (2008: 292), Duridanov et al. (1991: 299).
4 See Chapter 2, Section 2.2, Dobrev (1973), Duridanov et al. (1991: 299–301), Rikov (1993/1994),
Mirčev (2000: 117); cf. Huntley (1993: 152–153), Schenker (1993: 106). Doubts whether the OB/
OCS perfect had the same meaning as in Modern Bulgarian and occupied the same slot in the
system are expressed by Totomanova (2004), MacRobert (2013) and Ganeva (2015: 12 fn. 10).
The Perfect in the History and Structure of Bulgarian 243
As underscored above, in contrast to English, the verb ‘to be’ was the exclusive
marker of the perfect even in the earliest Old Bulgarian/Old Church Slavonic.
OB/OCS бьiти/byti was equally happy with intransitive verbs of motion, (6)–
(7), or with ordinary transitives, (8)–(9).
(6) ñѫ пришь-л-и
sǫ prišь-l-i
be.pres.3pl come-pptcp-nom.m.pl
‘they have come’ [original from Duridanov et al. (1991: 300)]5
As in earlier English, the perfect seems to have been interchangeable with the aorist in earlier
Bulgarian (cf. Chapter 3, the following chapter, Dobrev 1973 and Totomanova 2004). Although
they were not absolutely identical, often either of the two could appear in the same context
without any obvious semantic differentiation, and writers would frequently use them indis-
criminately, as will become evident in the samples from the next chapter.
5 Unless otherwise indicated, the transliterations, glosses and translations are mine.
6 It is noteworthy that the King James Bible uses a be-perfect here, but have in (9).
244 chapter 7
7 See Balan (1957: 23–24), Georgiev (1957: 43, 57), Tilkov et al. (1983a: 339–340), Minčeva (1987:
27–30, including notes on Greek and Romance influence on the development of the future
periphrases), Duridanov et al. (1991: 303–304), Huntley (1993: 154–155), Scatton (1993: 211),
Mirčev (2000: 117–118), Grković-Major (2011), who also cites abstract and delexical uses in
OB/OCS, such as ‘have peace’, ‘have faith’, ‘have pain’ (i.e. ‘be sick’). Načeva (1988) outlines the
wide range of meanings and functions of this verb in Modern Bulgarian.
The Perfect in the History and Structure of Bulgarian 245
7.4.1 Formation
Apart from the traditional be-perfect and the auxiliary functions of have dis-
cussed above, Modern Bulgarian has evolved another, quite distinct, perfect-
like construction which consists of a form of имам/imam ‘have’ and a passive
past participle of a transitive verb, which ends in -n or -t and agrees with the
object – (13a). This newcomer is practically interchangeable with the older be-
perfect with an active past participle – (13b).10
8 See Balan (1957: 23–24), Duridanov et al. (1991: 304), Huntley (1993: 154), Mirčev (2000:
118). Cf. Chapter 3, Section 3.5.2.3, as well as the grammaticalisation tendencies from
Chapter 2, Section 2.2.
9 Georgiev (1976: 310) treats this as a composite future tense.
10 See Georgiev (1957/1976: esp. 294, 300), Kuteva and Heine (2004: 43–44), and Nicolova
(2008: 270, 2017: 379–381); on forming the past passive participle, consult Tilkov et al.
(1983a: 380–382) and Danylenko (2005: 352–357). This construction has been around for
a while, especially in some dialects – see Elliott (2001), Xaralampiev (2001: 144–145), and
the sections below. Bybee and Dahl (1989: 70–73) supply similar examples from Czech,
Middle High German and Romance. Kuteva and Heine (2004) and Danylenko (2005) dis-
cuss analogous constructions in Northern Russian dialects and other languages. Chapter 3
246 chapter 7
This is different from (13), however, in that the subject (he) in (14) is not inter-
preted as the agent of the participial verbal element (left/having remained).
By contrast, if има/ima participates in an impersonal existential construction
functionally equivalent to English there is, the have-(proto-)perfect with an ac-
tive past participle in -l, (15a), has the same meaning as the established be-
perfect, (15b) (Bojadžiev 1968: 461–462).
presented a parallel state of affairs in OE. Have-perfects in Polish are mentioned in Łęcki
(2010: 168–169):
(i) Mam ten artykuł napisany. [Polish]
I.have this.m.sg article(m)[sg] written.m.sg
‘I have this article written.’/‘I have written this article.’ [from Łęcki (2010: 168)]
11 The word order of the auxiliary and the lexical verb is relatively free, though also governed
by the principles of clitic placement. See Elliott (2001: 102ff. and elsewhere), Nicolova
(2008: 292–294, 2017: 415–418).
The Perfect in the History and Structure of Bulgarian 247
b. постъпи-л-о е оплакван-е
postăpi-l-o e oplakvan-e
arrive-pptcp-nt.sg be.pres.3sg complaint(nt)-sg
‘A complaint has been received’ [from Bojadžiev (1968: 461–462)]
Extensive reports of the new perfect with имам/imam and the past passive
participle in the specialist literature date back to the 1930s and onwards, but
the pattern receives a mention even in the late 19th century.12 Georgiev (1976:
294) describes these combinations as typical of spoken, as opposed to written,
Bulgarian, in spite of their being frequent in officialese, reports, memos, etc.
They are also attested in the fiction of late 19th and early 20th-century writers,
such as Hristo Botev and Ivan Vazov, as in (16).
12 See Popova (1931), Balan (1957), Georgiev (1957/1976); cf. Kostadinova (2006: 153, 2009: 487,
with references) and Stoevsky (2006).
13 Impersonal existential constructions with има/ima ‘have’ take objects and there is no
overt subject. See Tilkov et al. (1983b: 89–90).
248 chapter 7
Like many of the other sentences above, (19) is interesting due to its word
order. The participle has moved next to the finite verb. The adjacency of the
two verbal forms can be taken to indicate a greater degree of ‘amalgamation’,
as discussed in the sections on OE in Chapter 3. More advanced grammati-
calisation could further be deduced from the future time context. Likewise
non-factual, a negative clause is provided in (20), although the subject of ‘have’
is not necessarily the implied agent of the participle (see Kostov 1982: 159).
(20) might show that semantic bleaching had commenced by 1863, because
no orders have been issued yet. However, one could still take ‘have’ here in
its original meaning of possession, albeit of something abstract, to the ex-
tent that orders can be given, taken or received: ‘I don’t have any orders yet’
(Steven Kaye, p.c.). In either case, ‘have’ is rather dynamic in the context of
receiving or having orders and is a delexical verb at the very least. It must also
be significant that ‘nothing’ and the participle are not adjacent as one object
constituent.
The Perfect in the History and Structure of Bulgarian 249
7.4.2 Agreement
Georgiev (1957/1976) has an extensive collection of examples of the fledgling
have-perfect, almost all of which show agreement between the object and the
participle. This is apparent in the sentences provided so far. One of the rare
exceptions with a breakdown in agreement is cited in (21), where the potential
controller (the direct object) is feminine singular, but the participle is neuter
(cf. Danylenko 2005 for East Slavonic analogues).
Sometimes the ‘object’ is not realised by a NP, which indicates further gram-
maticalisation, since lexical ‘have’ does not license such non-NP dependents.
The PP in (24) is less closely related to the participle than a direct object NP
would be; moreover, PPs themselves possess neither gender nor number. The
complement in (25) is a clause, again with no gender or number features,
which inevitably triggers default neuter singular morphology on the participle
(cf. Balan 1957: 23–24). It is interesting to point out that when the participle
fails to agree, it does not resort to an ambiguous zero marker (which was the
case in Old English). Instead, the non-agreeing option utilises the neuter sin-
gular, which is clearly and unambiguously indicated with -o.14
14 I will return to attested examples demonstrating lack of agreement in the established pas-
sive and be-perfect/renarrative (see Section 7.4.6).
The Perfect in the History and Structure of Bulgarian 251
By the same token, however, the analogous equivalence of (27a) and (27b)
could be taken to prove that the long-established be-perfect is nothing more
than a free syntactic combination with a subject complement which can be
either participial or adjectival. There is plural agreement in both cases.15
15 See Penčev (1968: 176), Hristov (2015, with references), as well as the discussion of the
ambiguities between passive and copulative be in English in preceding chapters.
252 chapter 7
In a similar vein, Georgiev (1957: 58) points out that, if such parallels are taken
seriously, then Bulgarian has not grammaticalised a passive either. In the fol-
lowing examples, the past participle also alternates with an adjective, and both
forms host agreement inflections:
The sentence in (32) is especially interesting, since the renarrated form with
the -l participle goes back to the original be-perfect (е имал/e imal), from
which the auxiliary has simply been dropped. Essentially, two diachronic lay-
ers or generations of perfect formations coexist side by side (see Chapter 2,
Section 2.2; more on this will follow too).
Just as in Old English, there are examples which show that the lexical mean-
ing of the verb for have has been bleached:
254 chapter 7
As Georgiev (1976: 299) notes, the coal is not there yet, so имаме/imame must
be an auxiliary verb. Possession in the strict physical sense is out of the ques-
tion. Имаме поръчани/Imame porăčani is more or less semantically equivalent
to the established be-perfect поръчали сме/porăčali sme.16 Such observations
lead Georgiev (1976: 300, 311) to conclude that the grammaticalisation process
is already at a rather advanced stage as of the 1950s, and that a new periphras-
tic form has arisen in Bulgarian. As already established, however, the idea of
‘having’ can be rather vague itself. Even in English, we could say We’ve got some
coal ordered, but it hasn’t arrived yet, with the tense showing that this is not the
same as the causative-like we had some coal ordered. Attested English examples
are: I’ve got 1500 Harry and Angela mugs due in this morning (attributed to a
souvenir merchant; http://newsthump.com/2017/11/28/confusion-in-germany-
at-news-that-prince-harry-will-marry-angela-merkel/, accessed on 29 Nov
2017) or She has a book coming out soon (from a linguistics talk). There is thus
clearly a sense in which the coal is theirs, they just do not have it in front of
them yet. The important point in (33) is not that the coal is not there yet. The
crucial thing is the reference to May, which does show that the construction
is used to talk about something done in the past, rather than just how things
stand now (I am grateful to Steven Kaye for alerting me to this).
16 Balan (1957: 25), however, describes the traditional be-form съм струпал (камъни)/săm
strupal (kamăni) ‘[I] have piled up (stones)’ as an active structure focusing on the agent,
with the have newcomer имам струпани (камъни)/imam strupani (kamăni) ‘[I] have
(stones) piled up’ as its passive counterpart which takes the focus away from the agent (cf.
Nicolova 2017: 380–381, as well as example (36) below). This is reminiscent of the distinc-
tion between the have-perfect and the ‘pseudo-passive’ with have in English, as well as the
distinction between dynamic and stative.
The Perfect in the History and Structure of Bulgarian 255
The second stage sets in when the verb имам/imam loses its lexical meaning
of possession, thereby turning into an auxiliary. The book in (35) has been or-
dered but is strictly speaking not yet in the speaker’s possession, as already dis-
cussed in relation to (33) above (cf. Stoevsky 2006: 145, 149, for similar Modern
Bulgarian examples of the type ‘have something confiscated’, as well as for ob-
jections to the view that these examples really represent desemantisation, in
line with the arguments concerning (33)).
The final stage is when the participle stops agreeing with the object, instead
appearing in the invariant neuter singular form. Such structures are also cited
in Balan (1957: 25) and illustrated once again in (36).
As is visible from the material, not all examples necessarily show all the gram-
maticalisation traits. Sentence (36), for instance, lacks agreement and the
256 chapter 7
participle is adjacent to the finite verb, but the possession meaning of first-
person имам/imam is arguably still there, so no semantic bleaching is in evi-
dence. Two further steps, which are yet to take place in Standard Bulgarian,
would be the extension of the construction to intransitive verbs and to inani-
mate subjects.17 The latter step might already have been initiated (at least in
embryonic form) – Stoevsky (2006: 148 fn. 1) notes Bulgarian examples of the
type ‘All computers have antivirus programs installed on them’, though agent-
hood is problematic here (cf. Kostov 2004: 128, Kostadinova 2006, 2009).
Georgiev (1976: 303) singles out the use of 3sg има/ima as an impersonal ex-
istential verb functionally equivalent to there is in English, (36)–(38), as a fac-
tor which contributed to the grammaticalisation of the new have-perfect, here
verging on a passive (though now this verb will have to serve more functions).18
17 See Friedman (1976: 97–98) and Elliott (2001) for a comparison between Bulgarian,
Macedonian and Serbian, regarding the progress each of them has made along this path.
Cf. Kostov (1982: 164–165), Bužarovska and Mitkovska (2010), as well as Kostov (2004: 127–
128) and Stoevsky (2006: 147–148, 150), who additionally deal with the extension of the
pattern from perfective to imperfective verbs.
18 On impersonal има нещо/ima nešto ‘there’s something’, see Balan (1957: 24). On the
links between possessivity, existentiality, and impersonal ‘have’, see Kostadinova (2006:
154–155) and the comments in Chapter 3 about the interchangeability of have and be.
Have-passives in Bulgarian are discussed by Nicolova (2017: 380–381).
The Perfect in the History and Structure of Bulgarian 257
Having thus laid out the stages along have’s grammaticalisation path,
Georgiev (1976: 300–301) suspects a functional trigger for these processes. He
attributes the rise of the new have-perfect to the fact that, between the 17th
and the 19th centuries, the old be-perfect started being employed to express
non-evidential, renarrated events, under the influence of Turkish.19 The same
pattern emerges here: the old be-perfect became excessively burdened with
expressing the current result of a past action, which some Slavic scholars
label ‘relativity’ or ‘time-correlation’, as well as with renarrated, non-witnessed
events. Therefore, a new construction had to come to the rescue and make the
functional load of be somewhat lighter, thereby avoiding potential ambiguity
between resultativeness and renarration.20
Mirčev (1976), on the other hand, objects to the conclusions and the chro-
nology proposed in Georgiev’s grammaticalisation story, providing Old and
Middle Bulgarian material which predates the 17th century and features the
same putative have-perfect constructions. Mirčev (1976: 315) claims that it can
hardly be supposed that those structures have undergone any development in
Bulgarian towards the formation of a new analytic complex, as he finds it hard
to detect any semantic differences between the Old Bulgarian/OCS phrase in
(39a), and its Modern Bulgarian translation in (39b) (cf. Stoevsky 2006). An at-
tested OB/OCS example is provided in (39c).
19 On renarration in Bulgarian, consult Gerdžikov (1984/2003). Gerdžikov (1984/2003: 252–
254) argues against the widely accepted view that Bulgarian evidentials calqued the cor-
responding Turkish forms. See Bybee and Dahl (1989: 73ff.) and Izvorski (1997) for more
general pathways of development from perfect to evidential. Narrative uses are reported
even for the perfect in English (Bowie et al. 2013, with references).
20 Cf. Gallis (1960: 184ff.) and Georgiev (1976: 300–302) for more details and some additional
points and refinements of this scenario. Similar scenarios for other languages are out-
lined in Drinka (2017: 91).
258 chapter 7
b. има-м я в-ложе-н-а
ima-m ja v-lože-n-a
have-pres.1sg acc.f.3sg pref-place/invest-pptcp-f.sg
‘I have it placed/invested’/‘I have placed/invested it’ [from Mirčev
(1976: 315)]
If the seeds of this phenomenon had already been sown prior to the Turkish in-
vasion of the Balkan Peninsula, it would be hard to motivate the appearance of
the have-perfect by saying that it arose because the old be-perfect had started
to express evidentiality as a result of contact with Turkish. Yet the have-perfect
could have existed in a low-key way, never really taking off, but it was only
The Perfect in the History and Structure of Bulgarian 259
when Turkish influence affected the reading of the be-perfect that it had a rea-
son to be taken up more substantially. It does not need to have been invented
after the Turkish-influenced change to the be-perfect (Steven Kaye, p.c.). The
interactions between these categories, the diachronic developments attend-
ing them and the chronology will be further examined below and in the next
chapter.
Modal uses of the original be-perfect seem to have existed very early on,
long before the modality-based category (or additional meaning) of renarra-
tion/evidentiality developed (cf. Chapter 8 and Dobrev 1973). According to
Huntley (1993: 153), if we take out of the equation doubtful examples and po-
tential scribal errors, we have attested about eighty instances of the l-participle
without an auxiliary in OB/OCS – the form now used as evidential. Huntley
(1993: 153) notes that, unlike the present and past perfect proper, when the OB/
OCS past participle stands alone without an auxiliary, it is not marked for rela-
tivity/time correlation; it is instead opposed to the aorist and imperfect tenses
in expressing modality, i.e. the narrator’s attitude to the event:
Whereas the aorist provides an objective account of the event, the past par-
ticiple събрали/sъbrali expresses the speaker’s attitude, which could be a
feeling of regret (as in (41)), or rejoicing, adding a rhetorical tone to the mes-
sage (Huntley 1993: 153, cf. Dobrev 1973). The significance of this for my the-
sis is that perfect-like forms already had more than one function in OB/OCS,
including non-temporal/aspectual overtones, even before the appearance of
the modal category/meanings of renarration associated with the l-participle
(cf. Gerdžikov 1984/2003: 112–113 fn. 7, Elliott 2001: 20–21). It is therefore prob-
lematic to see the emergence or spread of the new have-perfect as a backlash
against any presumed increase in the functional load of the old construction.
According to Gerdžikov (1984/2003: 259–260), renarrated/non-witnessed
forms themselves first appear in the Bulgarian charters of Wallachia and the
Prayers of Cserged, so these forms must have originated in the late 12th and
260 chapter 7
early 13th centuries, i.e. before contact with Turkish (though it is highly de-
batable whether the Prayers of Cserged are a truthful reflection of 12th- and
13th-century Bulgarian, G. Ganeva, p.c.). Antedating non-witnessed forms to a
period before the traditional date (i.e. the 17th century) casts even more doubt
on the widely accepted chronology and causal relationships pointed out above.
21 I consciously eschew any form of political or sociolinguistic debate regarding the status of
the two varieties.
22 Cf. Balan (1957: 27–28); for a more complete verbal paradigm of Macedonian, including
be- and have-perfects, see Friedman (1993: 280–281).
The Perfect in the History and Structure of Bulgarian 261
b. сум моли-л
sum moli-l
be.pres.1sg ask-pptcp[m.sg]
‘I have asked’ [from Friedman (1993: 280)]
b. бев моли-л
bev moli-l
be.past.1sg ask-pptcp[m.sg]
‘I had asked’ [from Friedman (1993: 280)]
etc.
Macedonian is much more advanced than its eastern neighbour in the gram-
maticalisation of the have-perfect.23 Example (44) illustrates the first stage
along the grammaticalisation path, where the participle still agrees with the
object. Note though that the finite verb and the participle are already next to
each other. However, the later stage in which agreement is absent by now pre-
vails in Macedonian (Georgiev 1957: 44; cf. Balan 1957: 28). This is illustrated
in (45).
23 See Georgiev (1957: 44), Gallis (1960), Elliott (2001) and Kuteva and Heine (2004: 45); for
more on the perfect in Macedonian, consult Friedman (1976, 1993: 270–272), as well as
Bužarovska and Mitkovska (2010).
262 chapter 7
(46). This would be impossible in Standard Bulgarian (cf. Section 7.4.7 below,
as well as Chapter 2, Section 2.2.3).
As Georgiev (1957: 44) observes, there are only rare examples of lack of agree-
ment in Bulgarian (as of the 1950s), and there are no occurrences whatsoever
of the have-construction with intransitive verbs (at least in the standard liter-
ary language). As pointed out above, the have-perfect is most fully grammati-
calised in the south-western dialects of Macedonian, from where it spread
towards the north-east (see Elliott 2001: 43ff. and elsewhere, Bužarovska and
Mitkovska 2010, with references), and perhaps only then into Bulgarian, where
it is still much less entrenched. This geographical distribution, coupled with
the rarity of have-perfects outside of Western Europe (Drinka 2017, with refer-
ences), lends credibility to a contact scenario regarding the emergence of this
pattern on South Slavonic territory. It must have entered from the south-west
and diffused eastwards.
(47) O V S APLACE
... смоковьниц-ѫ имѣаше единъ въ виноградѣ своемь
… smokovьnic-ǫ iměaše edinъ vъ vinogradě svoemь
fig.tree(f)-acc.sg had one in vineyard own
CO
въсажде-н-ѫ
vъsažde-n-ǫ
plant-pptcp-acc.f.sg
‘A certain man had a fig tree planted in his vineyard’ [Luke 13:6, English
translation from the King James Bible]24
24 The adjunct of place might actually be a dependent of the participle, not a separate con-
stituent on the level of the clause. It is also worth noting the non-adjacency of the finite
The Perfect in the History and Structure of Bulgarian 263
Both Bulgarian and Macedonian later lost their case declensions, but ad-
jectival number and gender survive. Agreement in this type of periphrastic
construction sometimes disappears and the participle can default to neuter
singular even when the object is, for instance, plural or feminine singular,
as in the Modern Bulgarian example in (36), or the Macedonian one in (45).
Unlike the material from English, the morphological marking in Bulgarian
and Macedonian is usually clear and there are hardly any cases which are am-
biguously interpretable as showing either presence or absence of agreement
(though see (51)). Lack of matching adjectival inflection was mentioned in
the previous sections as one of the factors which point to greater grammati-
calisation of the have-perfect. It has also been observed in other, arguably well
established, structures; they are said to be fully grammaticalised but neverthe-
less might on rare occasions admit mismatched/default agreement, although
normally standard agreement would be typical of them.
Agreement can default in the established Bulgarian be-perfect, as in (48)–
(49). The intransitive clause in (48) is especially curious, since the participle
has neuter singular -o (probably under the influence of subjectless neuter sin-
gular impersonal constructions; cf. Kovatcheva 1982: 5), while the expected
standard ending would have been the zero of the masculine singular in order
to agree with the subject ‘rain’. This is the exact opposite of what we saw in
English – here, a zero exponent showing agreement has been replaced by non-
zero marking clearly indicating a mismatch. In (49) and (50), a neuter singular
participle fails to copy the features of its feminine singular or feminine plural
subject.
verb and the participle in both the OB/OCS and the Early Modern English translations.
Importantly, the Greek original from which the OB/OCS was translated, following it close-
ly, has the same type of construction with have (see Drinka 2017: 97, 296–297). I return to
this in Section 7.4.7.
264 chapter 7
As demonstrated in (50) above and (51) below, the same breakdown might
happen in the context of stand-alone l-participles used for the purpose of re-
narration (ultimately derived from the perfect construction), as well as in the
well-established passive voice – (52).
In (52), the attributive adjective mirrors the features of the noun, but the pred-
icative participle does not. As in (48), it chooses neuter singular -o in preference
to masculine singular zero (cf. analogous disparities between the attributive
and predicative domain in OE from Chapter 4).
All of the above considerations cast doubt on the influence of agreement on
the grammaticalisation of a new have-perfect in Bulgarian, or its unassailable
usefulness as a diagnostic. Maybe the abandonment of matching adjectival
25 It is conceivable that било/bilo might agree with the neuter singular demonstrative това/
tova, which is also a potential subject candidate in this copular clause. Certainly, how-
ever, feminine singular била/bila is standardly acceptable. A more general case could be
made that neuter singular participles with neuter singular agreement controllers are not
clear and can in theory represent either presence of agreement or default neuter singular
non-agreement where the (irrelevant) controller just happens to match the features by
accident (see Elliott 2001: 27); the latter position strikes me as unjustifiable in the light of
the rest of the material.
The Perfect in the History and Structure of Bulgarian 265
b. той е падна-л/
toj e padna-l/
he.nom.m.3sg be.pres.3sg fall-act.pptcp[m.sg]
дошъ-л
došă-l
come-act.pptcp[m.sg]
‘He has fallen/come’ [standard Bulgarian]
The meaning is active in either case and this contamination of the two types of
participle in the South Slavic be-perfect might have occurred since there were
no separate active and passive past participial forms in Latin/Romance. It is
certain that ambiguity and zero marking cannot have played a part in this re-
placement due to the transparent morphology. If anything, the mix-up disrupts
hitherto functionally and formally clear categories, blurring the distinction be-
tween perfect and passive. This change flies in the face of ‘optimal function-
ality’ since it has created be-perfects formally identical to be-passives (both
constructions are also available with transitive verbs, as in the Macedonian
яден сум/jaden sum, which can now mean ‘have eaten’ or potentially ‘be eaten’,
though a prefixed stem, изеден/izeden ‘eaten’, might be preferred for the pas-
sive, D. Jangelovski, p.c.; similarly, сум омажен/sum omažen ‘have/be married’
can be active or passive – see Elliott 2001: 50ff., 64). Once again, functionality
cannot be a cross-linguistically explanatory notion since distinctions have been
dismantled here rather than upheld by relying on the hitherto formally distinct
participles. This formal coalescence further undermines any suggestions that
English be-perfects were sacrificed in order to achieve more optimal form-to-
function mapping of perfects and passives and points to grammatical change
that is much more mechanistic (i.e. indifferent to functional considerations),
driven by natural cognitive abilities to reinterpret linguistic structure and/or
by contact. For comparison, the reader is referred to Chapter 3, esp. Sections
3.1.2, 3.2, 3.4.4, where remarks were made about the nature of participles in OE,
including their reinterpretation in terms of voice from adjectival and passive
(‘I had him bound’) to verbal and active/perfect (‘I had bound him’). A pertinent
The Perfect in the History and Structure of Bulgarian 267
26 The discussion and the examples are based on Mladenov (1979/1929: 292). See Nicolova
(2008: 437–438) for some more special cases and exceptions. Modern Bulgarian verbal
morphology is in certain respects reminiscent of that of OE, especially in the marking of
transitivity.
27 Aromanian is a member of the Eastern Romance group spoken as a minority language
in Greece, Macedonia, and Thrace. See https://www.ethnologue.com/language/rup (ac-
cessed on 29 April 2016).
28 See Gallis (1960), Kuteva (1998), Kuteva and Heine (2004: esp. 41, 46, 51–56, 66–67),
Danylenko (2005) and Drinka (2017) for the role of contact in the development of an
analogous construction in Northern Russian and more generally across Europe (includ-
ing critical appraisal). Croft (2000: 146) mentions a have-perfect in Polish (used as a past
tense), which according to him arose due to interference from German. There have also
been suggestions that the Germanic have-perfect itself evolved under the influence of
Romance (or ultimately Greek) – see Drinka (2003, 2017), Kuteva and Heine (2004: 41),
Danylenko (2005), Łęcki (2010: 226 n. 12), Macleod (2012: 18–19, also providing convinc-
ing evidence against this hypothesis), Johannsen (2016: Section 3). While doubting the
more radical hypothesis that the Germanic perfect is wholly due to Latin, Macleod (2012:
268 chapter 7
Trying to shift the focus further east, Bojadžiev (1968) deplores the fact that
so much attention has been paid to the have-perfect in western, especially
Macedonian, varieties of South-Eastern Slavic and neglected elsewhere. He
stresses that it also exists in the (eastern) Thracian area of Bulgarian and is
indeed all-pervasive in this dialect continuum. Bojadžiev (1968: 462–463) re-
ports similar confusion of the two participles (past active and past passive)
in the Bulgarian dialects of Thrace and joins Mladenov (1979/1929: 292) in
seeking a foreign influence on the development of the have-perfect in Thrace,
this time influence coming from Greek (consult Gallis 1960, Asenova 1987,
Xaralampiev 2001: 144–145, Stoevsky 2006: 142, as well as Drinka 2017: 284, fea-
turing intransitive have-perfects with past passive participles from Thrace –
имам дойдено/imam dojdeno ‘I have come’; cf. Elliott 2001: 79ff., 94, for some
Romanian-influenced dialects of Bulgarian with a more grammaticalised pos-
sessive perfect, hence multiple points of entry for this construction are quite
possible).
Dialectal data with non-agreeing have-perfects from the Thessaloniki region
recorded in the 1970s is provided in (54) and (55). The participle ‘hung’ in (54)
might be given an adverbial interpretation, providing information about how
(in what state) the bag was ‘had’ – i.e. it was ‘had’ in a ‘hanging state’, so some
people would treat this as a converb stage in the grammaticalisation of the
have-perfect (see above). (55) is arguably a little more verbal.
170–172) concedes that there might be indications of at least some influence, albeit lim-
ited, of mediaeval Latin on European vernacular usage, including OE. The emergence of
perfects with habere ‘have’ in Latin itself (basically identical in every step of the process)
is discussed in Hopper and Traugott (2003: 62–65) and Drinka (2017: Chapter 6).
The Perfect in the History and Structure of Bulgarian 269
29 In principle, certain instances of /u/ in some dialects could be a reflex of the OB/OCS
back nasal, including remnants of accusative feminine singular inflections, but this is
unlikely here.
270 chapter 7
of different perfects in the rest of Europe (as noted in Chapter 6), coupled with
the fact that this construction is rare outside of Western Europe. The differ-
ent participial profile of Eastern languages with two past participles (active
and passive) is significant too, as is mixing them up despite their morphologi-
cal distinctness – they are thus brought into line with those of the Romance
and Germanic languages. All of this is suggestive of Western European and/
or Greek influence on the emergence of this new perfect. The textual record,
including translation practice, furnishes further proof for external models.
7.5 Conclusion
For centuries, Bulgarian has relied on a construction with be to form the per-
fect of all verbs, transitive and intransitive alike. It has more recently started to
grammaticalise have for the same role, as has the closely related Macedonian,
where the process has gone further. Formally, the new South Slavic have-perfect
construction looks almost identical to the complex transitive proto-perfects in
Old English. However, Bulgarian and Macedonian normally distinguish past
active from past passive participles and do not exhibit cases of ambivalent
zero exponence comparable to Old English. The fact that the same reanaly-
sis is nevertheless underway in these Slavic varieties and the agreement has
started to default (especially in Macedonian) indicates that grammaticalisa-
tion/reanalysis does not depend on formal prerequisites such as ambiguous or
missing morphosyntactic marking. The dialectal confusion between the for-
mally distinct active and passive past participles confirms this impression and
probably points to Western European influence, where only one all-purpose
past participle is available.
Earlier scholars have attributed the rise of the new have-perfect to the fact
that Bulgarian/Macedonian came to use the old be-perfect forms to renarrate
non-witnessed events, probably under the influence of Turkish. Be supposedly
became too overloaded with expressing perfect and evidential meanings so the
time was ripe for change. This functional overload account faces serious prob-
lems, just as it did for English. Firstly, have itself already served multiple func-
tions at the time it was recruited for the role of perfect auxiliary (see Section
7.3). Secondly, there are attestations of proto-have-perfects which predate con-
tact with Turkish, so the chronology of the functionalist story does not quite
add up. Furthermore, the Old Bulgarian be-perfect had modal uses even before
it acquired the modal connotations of evidentiality, so contact with Turkish
did not make it much more overloaded. Finally, the appearance of evidenti-
ality itself might actually predate the Ottoman conquest, which additionally
The Perfect in the History and Structure of Bulgarian 271
8.1 Introduction
Therefore I concur with MacRobert (1981: 156) that the Tale ‘is not to be treat-
ed uncritically as reliable evidence for fourteenth century popular Bulgarian’. It
is safer to take it to represent more general South Slavonic usage from the 14th
century, with an intermixture of Bulgarian and Serbian/Croatian features, both
literary and perhaps some spoken (see Vasilev 1972, MacRobert 1981: 156–157,
179ff., with references). MacRobert (1981: 183–185) concedes that the text may
still betray some of the developments in spoken syntax, albeit somewhat indi-
rectly, recapitulating as follows:
The Pritcha is isolated in Bulgarian literature. The few other secular nar-
ratives extant in MSS of Bulgarian redaction, the Tales of Alexander and
of Aesop, are thought to derive entirely from Greek sources and are writ-
ten in a Bulgarian Church Slavonic which lacks the modernistic pecu-
liarities of the Pritcha. In the absence of a body of comparable material,
and of detailed information about the development of Bulgarian Church
Slavonic syntax, the Pritcha can be used as a source for the history of the
Bulgarian language, literary or colloquial, only with the greatest caution.
The text examined here will be based on the Vatican MS (Cod. Vat. slav. 2),
made electronically available on the Cyrillomethodiana website (http://hist-
dict.uni-sofia.bg/textcorpus/show/doc_165, accessed on 06 March 2017) and
in Ruseva (2011), with a glossary. The results are presented in the following
section.
table 1 Perfects, conditionals, passives, aorists and imperfects in the Middle Bulgarian
Tale of Troy
1 The transliteration of Middle and Early Modern Bulgarian adopted here follows the same
conventions that are generally familiar from the rendition of Old Bulgarian/Old Church
Slavonic. Yet Slavonicists will be aware that some of the letters (especially those representing
276 chapter 8
The present perfects (a total of 29) typically appear in direct speech and ex-
press resultativeness, as in (4)–(6). The same link between perfects and direct
speech, hence more colloquial language, was noted for English in Chapter 4.
This association apparently holds true of (Middle) Bulgarian as well, where
the perfect is widespread in representation of conversation (cf. the following
section too).
the nasal vowels, the jers and the jat vowel) probably were either no longer pronounced or no
longer had their original phonetic values in the later periods.
The Perfect in Middle and Modern Bulgarian 277
For (5) and (6), like many others, it is clear from the context that the speaker
witnessed the event described by the verb. The present perfect is also typically
used to express past experiences which could affect the present, situated in a
time period leading up to the moment of speech, as in (7):
For at least eight or nine of the examples, however, perhaps even more (see
Appendix 10;2 cf. Vasilev 1972: 348ff., with references), the context might pro-
vide some indication that the speaker did not witness the event he or she is
describing, as in (8).
2 The Appendices of this book have been made available online. They can be accessed through
a QR code which can be found in the References section of this book.
278 chapter 8
Although this is listed as a third person singular perfect in Ruseva (2011: 96),
it might simultaneously be treated as a potential (proto-)evidential reporting
where the apple of discord was found. Arguably the gods to whom the words
in (8) are attributed were not present when the apple was discovered (though
they are probably omniscient). The perfect here is at the very least conclusive/
inferential, i.e. the speaker deduces/infers something they have not seen and
draws some conclusion based on the evidence. This is the bridge to the system-
atic non-witnessed evidential function which is to evolve later (cf. Vasilev 1972:
348ff., Gerdžikov 1984/2003: 29, 229, 231–236, 255–257, Izvorski 1997, Nicolova
2013). Gerdžikov (1984/2003) indeed notes that the perfect had contextual
non-witnessed uses as early as OB/OCS. Likewise, (9) could foreground the
resultativity but it additionally carries evidential and conclusive/inferential
meanings. It is clear from the context that the person who reports these events
did not witness them, at least not directly.
In (10) below, presumably the female speaker did not see Paris take Helen, and
only concludes that he has indeed left her for Helen (indirectly, based on the
available evidence). The aorist and the perfect appear side by side, both ex-
pressing second-hand information, rather than being harnessed for first-hand
The Perfect in Middle and Modern Bulgarian 279
(aorist) vs. second-hand reporting (perfect). Dobrev (1973: 11–12) too rejects
earlier claims that the aorist here simply denotes a witnessed event, whereas
the perfect denotes a non-witnessed one. This is not borne out by the story-
line. Instead, Dobrev believes that the choice of the perfect is motivated by
expressive reasons due to its greater emotional colouring and emphasis on the
relevance of the event (for the expressivity of periphrastic circumlocutions, cf.
Chapter 2).
As becomes clear from the contexts of the sentences or from general knowledge
(e.g. about when prophets lived), the examples in (11) do confirm that the aor-
ist can be used freely in the text to report non-witnessed events. Those include
reports based on typical hearsay evidence, (11a), counterfactual events which
would certainly require evidential forms in Modern Bulgarian, (11b), as well
as other mediated information, including the retelling of dreams, (11c)–(11d).
These uses of the aorist would all sound rather unusual in Modern Bulgarian.
Thus, both perfects and aorists are unproblematically selectable for non-
witnessed events in this Middle Bulgarian text. This is certainly not typical of
the aorist today. Only the perfect retains this function now, while things were
much more indiscriminate earlier. It once again becomes clear from (12) that
the mediaeval perfect had at least a contextual use in which it could be recruit-
ed for non-witnessed events, apparently in free variation with the aorist (cf.
Gerdžikov 1984/2003: 254 for an analogous state of affairs in OB/OCS, which
lacks a morphological category of evidentiality). It might be anachronistic to
The Perfect in Middle and Modern Bulgarian 281
spot evidential uses at a time when both the aorist and the perfect can have
them, as the evidential is not yet a fully-fledged category but just a fact about
the situation (Steven Kaye, p.c.). In English too, it would be possible to point
to examples where the speaker has seen an event for themselves and others
where they have only heard about an event, but that barely interacts at all with
which tenses are used (Steven Kaye, p.c.). These are legitimate objections but,
despite the anachronism, I record uses like the one in (12) in view of the later
developments on the way to Modern Bulgarian.
So the potential evidential use of the perfect can already be detected even be-
fore contact with Turkish. It is true that the auxiliary is never omitted in this
mediaeval passage the way it regularly is in Modern Bulgarian; but it should
also be noted that even in Modern Bulgarian a third-person perfect can act as
an evidential with or without omission of the auxiliary, so these alternatives
are still in variation.
3 The form glossed as ‘would’ here is a special conditional form of the verb ‘to be’. I return to it
later in this section.
282 chapter 8
Some speakers might feel that the first case (with omission of ‘be’) approxi-
mates something along the lines of ‘My grandfather told me so’ (renarrative),
whereas the second is more like ‘I guess so’ (conclusive/inferential), though it is
not always easy to keep this demarcation up in investigating actual usage. This
is evident in (14), taken from a corpus of contemporary spoken Bulgarian – the
first and the second attestations in bold have an auxiliary, while the third does
not, but all three of them report the same event:
The Perfect in Middle and Modern Bulgarian 283
Going back to Middle Bulgarian, example (15) from my corpus again appears
in direct speech. It demonstrates a further, less clearly defined, modal use of
the present perfect (cf. Chapter 7, Section 7.4.4, for the emotional colouring
of the OB/OCS l-participle). This is a conditional context with future time
reference and modality is further suggested by the (optative?) subordinator/
complementiser да/da. This attestation fits the description of emotional uses
of l-forms from the previous chapter, as well as Dobrev’s claim above that the
perfect might impart greater expressiveness to the utterance.
Apart from the perfect, there is another structure which involves a combina-
tion of ‘be’ and a past active participle ending in -l. ‘Be’ appears in a special form
(either an old optative or an old aorist), here glossed as ‘would’, and the whole
combination expresses conditional meaning, again in the realm of modality
(for the conditional in Old Bulgarian/Old Church Slavonic, see Duridanov et
al. 1991: 312–314). There were 21 such instances in my sample (see (12) above, as
well as Table 1), confirming the multiplicity of functions of ‘be’ + past participle
constructions (though the different form of the auxiliary must surely matter).
(17) shows masculine agreement with the name of Troy. Although the actual
name ends in -a in the nominative and is therefore expected to be feminine
(cf. (11a), where Troy takes the accusative feminine singular -ѫ/-ǫ), the generic
The Perfect in Middle and Modern Bulgarian 285
word for city is masculine, so this might be the reason for the masculine par-
ticiples in (17), essentially an instance akin to semantic or (better) associative
agreement.4
The second clause in (18) exhibits strictly grammatical agreement with the
neuter quantifier ‘many’ (in the comparative degree), not the partitive genitive
dependent governed by it (as with the perfect in (15) above):
4 We might be dealing with a mismatch between the declensional class and the grammatical
gender of Troy.
286 chapter 8
Here, I have chosen the Trojanski damascene from the 17th century, which is
made available electronically at http://histdict.uni-sofia.bg/textcorpus/show/
doc_11 (accessed on 21 March 2017). It is designated in Latin as Damascenus
Troianensis, and appears under MS number № ІІ, 11 or № 88 at the Bulgarian
Academy of Sciences. Ivanova’s (1967) is a print edition with a glossary and
notes.
in fact be smaller due to issues of symbol compatibility when the text from
the website was pasted into MS Word for word processing. The first part, a
sermon on the Ten Commandments, is mostly not about narrating past events,
so this might explain the lower incidence of the perfect there. The following
story about the life and miracles of Saint Simeon does narrate past events and
it also has direct speech, where many of the present perfects are to be found.
This once again reaffirms the relationship between (present) perfects and
direct speech already noted in the preceding section and in earlier chapters.
There are only four pluperfects in the entire sample as opposed to 44 pres-
ent perfects, which are common in dialogue because dialogue naturally views
(recent) past events through the prism of the moment of speaking, so current
relevance is more prominent and brought to the fore by perfect forms. There
is much less need for these perfect highlighting devices in narratives of purely
past events characteristic of chronicle entries which are interested in (more
distant) history (see Chapter 4). In the light of this, a pertinent question to ask
would be: are there more perfects in Damascenus Troianensis than in the Tale of
Troy because the category is more grammaticalised in 17th-century Bulgarian
or because there happens to be more dialogue in this sample?
Another issue has to do with the potential influence of Turkish on the
Bulgarian verbal system, since this is a post-contact text. Although lexical
borrowings do not guarantee structural ones, contact with Turkish is abun-
dantly evident from Turkish loanwords such as арти́сува/artisuva ‘be left
over’, файдъ́/fajdъ ‘benefit, advantage, use’ (ultimately Arabic, via Turkish),
душма́нин/dušmanin ‘enemy, tormentor’ (and derivatives), among others
The Perfect in Middle and Modern Bulgarian 289
(cf. (31) below). There is also the typical Balkan Sprachbund use of the da-
tive for the genitive: и тогiа стро́шишь глава́ дiа́вол-у/i togia strošišь glava
djavol-u ‘and then you’ll break the devil’s head’ [11] (see (31f) below); женѫ́
бли́жн-ому и съсѣ́ д-а твое́го/ženǫ bližn-omu i sъsĕd-a tvoego ‘[thou shalt not
covet] the wife of thy friend or of thy neighbour’ [12], with a dative (in -omu)
and genitive (in -a) side by side. In the following (almost minimal) pair, the
same dative -u ending serves to express the indirect object and the possessor:
ре́коха iгумен-у/rekoxa igumen-u ‘they said to the abbot’ [20] [dative indi-
rect object]; посте́лiaта С[ве]т-му Сѷмеѡ́н-у/posteljata Svet-mu Simeon-u
‘Saint Simeon’s bedding’ [18] [dative possessor].
Despite the obvious contact with Turkish and structural developments
typical of the Balkan Sprachbund,5 this stretch of text does not yield eviden-
tials of the type one might have expected to find, i.e. with missing auxiliaries
and supposedly due to Turkish influence. The perfect sometimes denotes
non-witnessed events, as confirmed by the context, and the overall situation
appears more or less identical to that in the pre-conquest Tale of Troy. The dis-
tribution of the different patterns is similar, as is the way they are used.
And yet, some crucial shifts deserve special mention. Although the synthet-
ic preterital forms (aorist and imperfect) remain dominant, the percentages of
passive and perfect periphrases with be have increased more than twofold: be-
passives accounted for 0.62% of all recorded verbal constructions in the Tale
of Troy, whereas they now have a 3.24% share. Perfects have gone from 4.72%
in the Middle Bulgarian Troy sample to 10.37% in this early modern text. It will
emerge below that the perfect periphrasis has started to be consolidated not
only in numerical terms but also in terms of typical usage, though there are
still many similarities to the mediaeval period and it has not yet attained its
present-day status. It would come as no surprise that this Early Modern text is
somewhat of a bridge half way between the Middle Bulgarian passage exam-
ined above and the modern day.
As noted in the previous section and earlier in the book, a characteristic
usage is the so-called experiential perfect, focusing on the expression of one’s
own past experiences which may affect the present, as in (20a). Similarly, it
is clear in (20b) that the speakers report their own actions and experiences,
which additionally have current relevance.
5 For the Balkan Sprachbund, see Trask and Millar (2015: 303–304, with references).
290 chapter 8
For many perfects, though, it was not clear whether they were meant more as
highlighters of resultativity or whether they were intended primarily to de-
scribe non-witnessed events:
The writer could hardly have witnessed the creation of the world or what went
on between God and Joseph, so the material in (23) certainly describes non-
witnessed events, though arguably it might still be intended to foreground the
resultativeness of what is reported:
For (24), it is not clear that the events are not witnessed, since the spirit that
speaks these words is probably omniscient (cf. a similar example from the Tale
of Troy). In any case, the 3sg auxiliary would not be omissible today from the
final clause in (24), which might suggest that this is a bona fide perfect, not a
proto-evidential.
292 chapter 8
Probably the most typical and convincing example of reporting, where evi-
dential forms with an ellipted auxiliary would be appropriate in Present-Day
Bulgarian, is provided in (25). The narrator most likely presents the event of
the main character telling the monks as witnessed by himself too, so he uses
the aorist of ‘say’. However, the narrator did not witness the character’s dream
for himself, hence the l-participle of ‘see’. The two preterital forms are now
functionally differentiated in this material, though the perfect auxiliary is re-
tained (cf. the discussion of (10) and (11) above, especially the analogous (11c),
again reporting a dream but with two aorists).
At least a quarter, and probably more, of the 48 perfects here indeed de-
scribe non-witnessed events, as suggested by the context (see Appendix 11).6
However, as noted above, not all of them could translate into renarrated forms
without an auxiliary today. For instance, the context suggests that the Saint’s
actual death in (26a) was not witnessed by anyone, but an evidential construc-
tion with missing ‘be’ would hardly be resorted to for that subordinate clause
6 The Appendices of this book have been made available online. They can be accessed through
a QR code which can be found in the References section of this book.
The Perfect in Middle and Modern Bulgarian 293
Despite the incipient differentiation between aorist and perfect in (25) and
(26), this post-conquest text also exhibits aorists in non-witnessed/evidential
function, just like the Middle Bulgarian Tale of Troy, written before contact
with Turkish. Such ‘evidential’ aorists were counted and recorded on the first
three pages only, just to illustrate this point.7 The word count for the first three
pages is ca. 727 words, and there are also imperfects throughout the text which
describe non-witnessed events, though they have not been recorded systemati-
cally. In the brief stretch of only 727 words, I found many more ‘non-witnessed’
aorists, apparently much more numerous than the perfects. Pages 1 to 3 con-
tain only one perfect, as opposed to at least eighteen aorists describing non-
witnessed events (though the reader should keep in mind the introductory
remarks about genre). This should additionally be placed in the wider context
of perfect vs. aorist usage in the entirety of this sample compared to the previ-
ous one. While the imperfect has remained fairly constant (at roughly 17% in
both passages), compared to the Tale of Troy, the perfect has carved out more
territory from the aorist here (an increase from 4.72% to 10.37% for the perfect,
accompanied by a drop from 74.91% to 67.82% for the aorist). Nevertheless, the
aorist remains the dominant preterital form (cf. Tables 1 and 2). Crucially, it can
still relate second-hand information.
The two constructions might in fact serve subtly different functions, ex-
pressing slightly different meanings, so a future investigation could shed
more light on the differentiation between them (cf. MacRobert 2013). Let us
consider (27):
The reason for using aorists in (27) (cf. the perfect in (23) above) could be that
the writer wanted to make what he describes in (27) sound more ‘real’, as if he
had been there and seen it himself. This might make his message more vivid
and therefore more convincing (though some scholars ascribe greater vivid-
ness to the periphrases, e.g. Dobrev 1973; see preceding section). All things
considered, however, one can conclude that there is no systematic correlation
between form and meaning at this stage. Non-witnessed meaning can be ex-
pressed both by the perfect and by the aorist (and the imperfect too), so contra
Gerdžikov (1984/2003: 256, 260), it is perhaps too early to talk about a fully-
fledged grammatical category of evidentiality, even though evidential and
‘double-evidential’ (admirative/dubitative/incredulous) forms do exist else-
where in damascene literature, including in this very collection (as described
by D’omina; G. Ganeva, p.c.).8 In some respects, this 17th-century document
is little different from the 14th-century Tale of Troy or from the earliest OB/
OCS records. Thus, Jesus’s life can be summarised in the aorist, as in (28a) (cf.
Dobrev 1973: 13–14, 17–18, and MacRobert 2013: esp. 391, 397–399, for classical
OB/OCS analogues where aorists alternate with perfects (with or without an
auxiliary) in what appears to be more or less free variation). The aorist in (28b)
is also non-witnessed and might even indicate a certain amount of resulta-
tiveness. Perfects and aorists coexist in (28c) and arguably both constructions
indicate resultativeness. Unlike the parallel examples in (26), (28d) has an aor-
ist in a counterfactual context, where a perfect would be obligatory today (cf.
previous section too). All of this suggests that the perfect-aorist dichotomy has
not yet attained its present status, though the specialisation of the two might
be underway.
8 See Gerdžikov (1984/2003: 104 fn. 2, 256, 260–261, with references), D’omina et al. (2012: xii),
inter alia. Below is one such random example from the same collection but outside the scope
of my sample. Vindicating my thesis, the stand-alone evidential participle in -l coexists with
an aorist (случи/sluči) which reports the same event. Both forms would be expected to get -l
in Present-Day Bulgarian.
(i) И тоги́ва случи́ се ѿ индiѧ нѣ́ кой члкь купе́ць
I togiva sluči se ot indię nĕkoj člkь kupecь
& then happen[aor.3sg] refl from India some person merchant
до́шь-л-ь вь iерслимь
došь-l-ь vь jerslimь
come-pptcp-m.sg in Jerusalem
‘And then it happened that a merchant from India came to Jerusalem’ [Source: https://e
-medievalia.uni-sofia.bg/moodle/mod/page/view.php?id=2839 (accessed on 11 Sept 2017)]
296 chapter 8
In addition to its other functions, the perfect in this sample also demonstrates
a certain amount of emotional colouring, indicating immediacy and greater
vigorousness, even impatience and annoyance, compared to the plainer/less
marked simple present which might have occurred in (29) instead:
As an aside, this might be a good point to note that so-called reflexive passives,
i.e. semantically passive constructions with an active verb and a reflexive clitic,
are treated as morphologically active (see (28a)). The number of passives fea-
tured in Table 2 and in Appendix 119 only includes combinations of be and a
9 The Appendices of this book have been made available online. They can be accessed through
a QR code which can be found in the References section of this book.
298 chapter 8
past passive participle. Thus, it might be speculated that the perfects here (and
in the previous Troy sample) are much more numerous than the participial
passives due to the existence of ‘reflexive’ passives as an alternative option for
marking voice in Bulgarian. It is hard to tell if this is one of the reasons why
be-perfects survive, though so do the (underdog?) be-passives. This line of rea-
soning is reminiscent of the ruminations involving the two passive auxiliaries
of early Germanic (Chapter 3, Section 3.4.3, and Chapter 6).
Finally, I detected no traces of perfects with have in this sample, at least
no obvious ones. However, the verb for ‘have’ did appear delexicalised in
constructions with object nouns, such as those in (31), similar to earlier and
modern English combinations like have need/shame/a fight etc. (31a)–(31g)
demonstrate the wide range of nouns that can combine with ‘have’ in these
delexicalised constructions: душманлъ́кь/dušmanlъkь ‘enmity’, файдъ́/fajdъ
‘benefit, advantage, use’, кара́зь/karazь ‘fight, argument’, любо́вь/ljubovь ‘love’,
проще́нiе/proštenie ‘forgiveness’, as well as добринiа/dobrinja ‘good(ness)’.
(31f), (31h) and (31i) demonstrate that other generic verbs than ‘have’ are
available in such delexicalised constructions, including ‘do’ and ‘hold’. This
indicates that it is such generic verbs that, due to their non-specific mean-
ings, are likely to be delexicalised even further and to experience grammati-
calisation. The appearance in (31i) of the synthetic verbal form душма́нишь/
dušmanišь ‘(lit.) [thou] tormentest’, related to a noun in the delexicalised V +
N combinations, confirms that these patterns are equivalent to verbs in that
they convey dynamic meanings. The same applies to the verb and the noun
for ‘love’ in (31b) and (31e). Once a stative verb denoting possession, delexical
have has started to acquire a more dynamic interpretation arguably condu-
cive to onward grammaticalisation. The same equivalence between the verb
надзе́мувай/nadzemuvaj ‘take too much, overcharge’ and the V + N combina-
tion зе́ма артъ́кь/zema artъkь ‘take a surplus/take too much, overcharge’ is
in evidence in (31j), where we additionally see that these verb-like, dynamic V +
N constructions can undergo subsequent nominalisation into deverbal (com-
pound?) nouns: артъ́кь зема́нiе/artъkь zemanie ‘surplus-taking, taking too
much, overcharging’.
‘have’ which was prone to get even less concrete and more dynamic over time.
By the time of this representative of 17th-century Bulgarian, I have not detect-
ed any significant shifts in the distribution or functional load of the be-perfect
as compared to the 14th century, including evidential uses (despite the incipi-
ent functional differentiation of the various preterital forms). Stand-alone re-
narrative l-participles exist outside of my sample, but they are still in variation
with aorists. In the next section, I look for realignments within the system in
the 18th century.
consulted here is that of Peev et al. (2015), which is based primarily on what is
believed to be the original draft, but the editors supply readings and emenda-
tions from the later copies too. In the preparation of this chapter, parallel use
was made of the electronic publication available on the Cyrillomethodiana
website (http://histdict.uni-sofia.bg/textcorpus/show/doc_137, accessed on 04
April 2017). Each example is identified with folio and line numbers.
table 3 Perfects, evidentials, conditionals, passives, aorists and imperfects in the preface
to Paisius’s Slavonic-Bulgarian History (1762)
a Of the 15 past participles counted as instances of the perfect, only 8 are combined with an
overt occurrence of the auxiliary be. As long as there was an overt auxiliary that could com-
bine with multiple conjoined participles, all were taken to implicitly go with that auxiliary.
This was done in order to achieve a principled way of categorising the forms, in line with the
introductory remarks at the opening of this chapter (see Section 8.2.2 and the discussion
below).
304 chapter 8
10 It should be noted that the line numbers given in square brackets after the examples refer
to the numbers in the original editions, which do not correspond to the lines as they ap-
pear here.
The Perfect in Middle and Modern Bulgarian 305
Importantly, all past active participles in (32) above are treated as separate
tokens of the perfect (as opposed to the evidential), because there is an aux-
iliary which arguably applies to all of them. Although the events are not wit-
nessed, the author may have intended to additionally foreground and reinforce
the resultativeness arising from the situation described.
The decision to treat such participles as perfects with an auxiliary recover-
able from elsewhere is debatable. Strictly speaking, of the 15 past participles
counted as instances of the perfect, only 8 are combined with an overt oc-
currence of the auxiliary be, as indicated underneath Table 3, so the number
of actual perfects might have to be revised downwards. This applies in equal
measure to example (33).
Other manuscripts omit the auxiliary in (33) (see Peev et al. 2015: 76, as well
as Appendix 12,11 for further similar editorial and scribal interventions). It
is therefore questionable whether all analogous participles should be taken
to covertly combine with a single instance of auxiliary be. These scribal in-
terventions also bear testimony to the variability which existed at the time
11 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
306 chapter 8
It is worth noting two more relevant points regarding (35). Assuming that
наiде/najde ‘finds/found’ is an aorist (rather than present tense, with which
it is syncretic), there is a contrast between the event expressed by the aorist
(experienced first-hand) and the second-hand information reported with the
perfect. Secondly, singular наiде/najde (also part of a so-called reflexive pas-
sive construction) fails to agree with its plural subject, probably because of the
impersonal flavour associated with such ‘passives’. As pointed out in earlier
sections, shedding agreement is a wide-ranging phenomenon which applies to
the whole spectrum of verbal constructions (see further below).
Virtually absent from the earlier samples, evidentials are now abundant.
Compared to other verb forms that could serve the same role, there were many
more evidentials with no auxiliary employed to report second-hand informa-
tion here – at least 27, even more if the unclear cases above are included for
which it is not certain whether to relate a single auxiliary to multiple parti-
ciples. In either case, this type of construction now seems fully established
and is indeed the predominant way in which Paisius narrates historical events
(see Section 8.4.1). As with the perfect proper, there is often a clear-cut contrast
between the non-witnessed evidentials (l-forms highlighted in (36a)) and the
witnessed aorist (напи̑сахъ/napisaxъ ‘I wrote’ at the end of (36a) and another
two aorists in (36b)).
12 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
308 chapter 8
13
The unabridged quotation on which this translation is based can be found in
Appendix 12.
The Perfect in Middle and Modern Bulgarian 309
The -лъ/-lъ of прославiлъ/proslavilъ in (38) was inserted later above the line
in the draft. Without it, the remainder is an ordinary aorist, so this might con-
firm the interchangeability and coexistence of evidentials and aorists with the
same functions in the minds of speakers (unless it is merely a scribal oversight).
14 It is not absolutely certain that оувѣщава/uvĕštava is an aorist, and some versions sup-
ply an unambiguous present-tense form here (see Peev et al. 2015: 66). The hyphenation
in that unclear word in (37) is from the original. The interchangeability of aorists and
evidentials in damascenes and in the National Revival period is noted by Gerdžikov
(1984/2003: 182 fn. 4), possibly attributable to Church Slavonic influence; like OB/OCS,
Church Slavonic does not possess evidentiality as a grammatical category.
15 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
310 chapter 8
One reason for using ‘second-hand’ aorists suggested in Section 8.3.2 was to
lend vividness to the retelling of Biblical events. As in the earlier periods, in
(42a)–(42b) the aorist and the imperfect describe non-witnessed events in a
Biblical context, so one wonders if the Bible might have acted as a conserva-
tive force to retain more archaic ‘non-witnessed aorist uses’ in the face of rising
evidentials, or as a context encouraging ‘vivid, as-if-witnessed’ aorists in order
to lend credibility to Christian doctrine. In (42c), however, the vividness effect
is not connected to the life of Jesus or other figures in Christian belief, so this
is a more general strategy which is not prompted solely by religious doctrine.17
Then again, the ‘incoming’ perfects/evidentials are used for similar material
in (42d), repeated from (33). In this respect, the situation, with its unabated
characteristic variation, is not too far removed from the picture in the earlier
periods. After all, language change is a gradual and incremental process.
16 The Appendices of this book have been made available online. They can be accessed
through a QR code which can be found in the References section of this book.
17 This can occasionally be encountered today, especially in the speeches of politicians
when they describe historic events. Gerdžikov (1984/2003: 34–36) supplies modern ex-
amples of the coexistence of evidentials and aorists for non-witnessed events (cf. Drinka
2017: 16, with references).
312 chapter 8
Apart from periphrastic perfects, there are three conditional structures and
six be-passives in this sample. Two of the past passive participles, in (43), ap-
pear on their own without an auxiliary (in a quasi-impersonal post-modifier
clause); they fail to agree with the feminine singular antecedent they appar-
ently modify.
18 The grapheme ѧ, originally employed to represent the Old Bulgarian/Old Church
Slavonic front nasal vowel, is probably used here (and elsewhere in the material, includ-
ing Damascenus Troianensis) to represent /ja/ under the influence of Russian Church
Slavonic. The pronunciation of this letter in Bulgaria during the National Revival is dis-
cussed in passing by Peev et al. (2015: 48).
314 chapter 8
As with the previous samples, no have-perfects proper were found in this 1762
excerpt, but instances of delexical have abound, perhaps paving the way for
the rise of the Modern Bulgarian have-perfect. Examples from Paisius’s History
outside the scope of my sample quoted in the literature include the following:
а болгари имеяли распра между собою .../a bolgari imejali raspra meždu sobo-
ju ‘and the Bulgarians had an argument with one another’, да има разговоръ/
da ima razgovorъ ‘(lit.) to have a conversation’ (Georgieva et al. 1989: 63, 66). In
addition to these instances of have with objects denoting actions, my material
yields (46a) and (46b), though here и̇меютъ/imejutъ and немашъ/nemašъ
could merely indicate possession of a tangible or non-tangible thing (cf.
Timberlake 1993: 874–875 for Russian, where the cognate of Bulgarian ‘have’
appears to have specialised in expressing the possession of abstract qualities
such as honour, influence or authority, unlike the cognate of ‘be’).
19 One might argue that, as in Modern Bulgarian, the adjective славни/slavni ‘glorious,
famous’ is distinguished from the participle славени/slaveni ‘worshipped, honoured,
glorified’.
The Perfect in Middle and Modern Bulgarian 315
The meanings in (47) are arguably more dynamic. In (47a)–(47b), the abstract
objects of love, peace and hatred presuppose feelings/relations which one has/
experiences, rather than physical objects (concrete or otherwise) which one
can possess (similar examples were recorded in Damascenus Troianensis). In
(47c), Paisius exhorts readers to keep his book so that it does not get lost – the
2pl imperative form indicates a more dynamic sense of have, with the address-
ees capable of acting and taking care to preserve Paisius’s work. Static posses-
sive have would not typically appear in the imperative.
c. и̇ и̇ме-и-те ю да се не
i ime-j-te ju da se ne
and have-imperative-2pl 3f.acc.sg comp refl neg
погуби̇ [f. 3v, l. 9–10]
pogubi
perish
‘and have [i.e. keep] it [the book] so that it doesn’t get lost’
(48) O S V
сегашни̇ хiтрци̇ що гi ти̇ и̑ма-шъ
segašni xitrci što gi ti ima-šъ
present cunning.men that them thou have-pres.2sg
CO
за почестъ [f. 4v, l. 14–15]
za počestъ
for honour
‘our own cunning men, whom you hold in high regard’
Existential have in the default third-person singular has also been recorded
in (45) above; it arguably bears some resemblance to a nascent proto-have-
perfect (though identified with the benefit of a great deal of hindsight and
power of the imagination and far from convincing as evidence). At least it
could be viewed as a stepping-stone to the new perfect, especially with the pas-
sive participle in the relative clause and the existential use of have in the main
clause, as already noted in the previous chapter, though the link remains rather
tenuous. The final example does not showcase have at all (not in the relevant
first clause, that is). In (49) is a construction with another existential verb of
similar ontological category alongside a passive past participle as modifier
The Perfect in Middle and Modern Bulgarian 317
which (again with some hindsight) might remind the observer of certain prop-
erties of the later have-perfect, which was just beginning to be attested in the
varieties of Macedonia at about that time (see Chapter 7, Section 7.4.5). The
participle ‘written’ could be taken to modify the PPs ‘by little’ and ‘in short’,
but it could alternatively be a substantivised adjective acting as subject (itself
modified by those PPs).
Like the 14th- and 17th-century samples, the preface to Paisius’s 1762 master-
piece allows the use of be-perfects and aorists to express non-witnessed events.
What is new here is the frequent omission of the perfect auxiliary in order
to express non-witnessed events with a stand-alone participle in -l. In other
words, evidentials proper already turn up in great numbers in this text. Unlike
the aorist and the ‘real’ perfect, they are restricted to second-hand reports and
can hardly be met with in first-hand accounts. However, there is a good deal of
variation when it comes to second-hand reporting and all three constructions
can naturally be employed in such contexts, often in close proximity to each
other. Different copyists exhibit different preferences too, with a marked ten-
dency for later copyists to eliminate the auxiliaries (which then often resurface
in the modern translations). Crucially, the advent of evidentials has not led
to an immediate deployment of have-perfects in order to alleviate the burden
of be. As in the earlier centuries, there are nevertheless examples of delexical
have which might be seen as intermediate stages in the grammaticalisation
of this verb. And yet, it would be risky to posit a causal relationship between
the rise of evidentials and the new have-perfect because of the rather long
period of reasonable stability in a system which functioned happily without
318 chapter 8
The sentences in (56) and (57) are also rather dubious as have-perfects, again
because the participles behave syntactically like attributive modifiers within
the object NP (preceded by determiners). Once again, it is plain to see in (56)
that the subject of has (Slavija) is not the same as the implied agent of the
participle (cf. Chapter 7, Section 7.4.1). Even though they were not included
in Table 4, clauses like (56)–(57) might be seen as (remote) potential precur-
sors in the cline leading to the formation of a real have-perfect, especially the
impersonal (57).
The Perfect in Middle and Modern Bulgarian 321
The sentence in (58) was discarded since the past participle is itself modified
by an adverb. Moving to the plural, (59) and (60) were likewise excluded be-
cause it is again obvious that the pro-dropped first-person plural subject of
имаме/imame is not the agent of the activity expressed by the passive past
participle. In other words, it was not us that supplied the woman in question
with weapons, and it was not us that wounded the person. All the examples so
far have been with passive past participles suffixed with -n or -t. A similar case
but with an active l-participle is illustrated in (61) (cf. Chapter 7).
Based on this corpus evidence, I conclude that the have-perfect is still not en-
trenched enough in Bulgarian, and the language community does not seem to
‘feel a need’ to use it, despite the neat picture painted in the functionalist ac-
count whereby have-perfects are expected to ‘come to the rescue’ and diminish
the functional load of be in this domain. Interestingly enough, the material ex-
amined demonstrates more modal uses for have, some of them listed in (62)–
(67). (63) is noteworthy because there is no NP object of have or of the lexical
verb. The word order in (66) favours a grammaticalisation account (unlike in
the perfect above; cf. (64) and (65), with a more conservative word order). In
(67), the clubs are not yet in possession of the money, so semantic bleaching
has set in. There were of course plenty of delexical uses as well, not unlike
those from earlier centuries, but it is not necessary to dwell on them again.
This diversity of ‘polygrammaticalisation’ paths for various auxiliary roles does
indicate that generic verbs like have are the most suitable candidates for re-
cruitment as function words.
In (70), again from Tobacco, have retains its meaning of possession but the ob-
ject has a past passive participle as a pre-modifier. It is clear from the adverb
добре/dobre that разработено/razraboteno is an adjectival attributive modi-
fier of дюкянче/djukjanče, but this construction could still be perceived as a
rather remote precursor/‘distant relative’ of the have-perfect (cf. the analogous
sentence in (58) above). This is the closest to a possessive perfect that one gets
in these fiction samples.
central Bulgaria (though they both moved around a good deal). This outcome
might in part be due to the supposedly initially ‘colloquial nature of the ima
construction, which [causes] it to be avoided in the literary language’ (Elliott
2001: 46). Despite these inconclusive results from an admittedly small-scale
study, it can still be surmised that the geographical spread of these new tense/
aspect forms must have followed a trajectory from west to east, in line with
the fuller grammaticalisation in the modern standard and dialectal varieties
in the Republic of Macedonia as compared to those of Bulgaria, where have
remains at an embryonic stage as a perfect auxiliary. Much more extensive
work is needed in order to shed light on the actual transmission routes. This is
outside the scope of the present book and will have to await future research.
8.6 Conclusion
As I did for English in the previous chapters, here too I have tried to analyse
small chunks of the history of Bulgarian which give insights into the bigger
picture, including inter- and intra-systemic relations, as well as the impact of
external factors such as contact. The selection included representative texts
from the 14th, 17th and 18th centuries. The 14th-century Middle Bulgarian text
yielded some breakdowns in agreement with periphrastic verbal construc-
tions, which however are in a clear minority (and remain so throughout). No
have-perfects were found, nor were any structures that could be perceived as
their immediate precursors, so be is the only auxiliary in this role here. Some
of the instances of the be-perfect are contextually used for non-witnessed
events (and have been since the earliest records), though the auxiliary is never
omitted and the aorist can report non-witnessed events as well. There are also
indications that even in Present-Day Bulgarian, omission of the auxiliary in
third-person evidentials is not consistent, so some things have changed very
little since mediaeval times. Even today, speakers might occasionally use aor-
ists for non-witnessed events. The same range of options is still available but
their distribution has shifted dramatically by now.
By the time of the representative of 17th-century Bulgarian post-Ottoman-
contact, I detect no significant shifts in the distribution or functional load of
the be-perfect compared to the 14th century, including the same contextually
evidential uses. Although evidentials proper exist in damascene literature
more widely (and probably predate it), they do not show up in my sample,
where the perfect and aorist are still in free variation, so an obligatory category
of evidentiality has, strictly speaking, not yet come into being. No have-perfects
were employed either, though there were instances of delexical transitive have
The Perfect in Middle and Modern Bulgarian 327
Conclusions
contact might work in a way that overrides frequency. Similarly, in Old English
the ubiquity of legitimate zero marking was proved to play no part in the loss
of overt agreement (Chapter 4). Agreeing plural passives held their ground in
spite of being in the minority, and so did perfect be in English, for a very long
time. Rare(r) items are not necessarily doomed. Though this might be hard to
admit, the messiness and unpredictability of language change often thwarts
even the most ingenious attempts at imposing neatness and orderly patterns
on its vagaries. Delineating violable tendencies is sometimes the best one
can do.
I hope that this study gives a taste of how complex the developments under
consideration are, especially when approached from different angles. Mapping
out parallel developments in two languages has helped to highlight that and
to cross-pollinate ideas from the two scholarly traditions. Another possible
cross-pollination might involve an exciting link between some of the regulari-
ties that have emerged for both languages under investigation, as pointed out
to me by M. Kovatcheva (p.c.). A connection was noted in Chapter 3, Section
3.5, between schemas based on I have something and something is at/to me,
both featuring generic and extremely common lexemes such as be and have
(cf. Drinka 2017: 89–90, with references). In addition, it should be stressed that
past participles can have an adverbial function, and it is precisely the adver-
bial suffixes that appear to act as defaults in certain contexts. In Slavonic, -o
serves both as neuter singular marker on adjectives and as a marker of adverbs,
whereas -e, apart from frequently indicating nominative/accusative plural on
OE adjectives, also marked adverbs in early Germanic, including Old English
(see Chapter 4). These observations would account for the interchangeabil-
ity of the auxiliary verbs in the perfect and the possibility of one ousting the
other, as well as for the changes in the morphological exponence. Due to the
above-mentioned equivalence of the schemas, it no longer seems strange that
a passive participle can acquire an agent (in the perfect), or that an active par-
ticiple (e.g. arrived) can come to combine with have as opposed to be (or vice
versa: see below for transitives with perfect be). Grammaticalisation is indeed
accompanied by loss of agreement, but rather than being a trigger, the loss of
agreement (or the collapse of other morphological distinctions) is the ultimate
outward sign that the construction has been reanalysed. This would enable
the transition from (1a) to (1b) to (1c), all of them paradigmatically related.
The transition must have proceeded via an ‘adverbial’ stage which bestowed
more verbal features on the originally adjectival participle. Adverbs are not
expected to agree, which would in turn account for the shedding of inflections
or switching to petrified neuter singular forms. Essentially, no zero morphol-
ogy is needed as a bridging context. To the extent that agreement in Modern
332 chapter 9
Bulgarian tends to survive much better than it did in late Old English and early
Middle English, this probably has to do with the better preserved overall mor-
phology of Slavic compared to that of Germanic, as suggested in Chapter 1.
Additionally, this has to do with the lesser degree of grammaticalisation of
have-perfects, as confirmed by comparing Bulgarian to Macedonian (where
agreement is dropped).
(1) a.
Бележк-и-те са написа-н-и
Beležk-i-te sa napisa-n-i
note(f)-pl-def be.pres.3pl write-pptcp-pl
‘The notes are written’/‘The notes have been written’
Benson, Larry D. (ed.) et al. 1987a. The Riverside Chaucer. 3rd ed. Boston: Houghton
Mifflin Company.
Benson, Larry D. 1987b. The canon and chronology of Chaucer’s works. In: Benson,
Larry D. (ed.) et al., The Riverside Chaucer. 3rd ed., xxvi–xxix. Boston: Houghton
Mifflin Company.
Benson, Larry D. 1987c. The texts. In: Benson, Larry D. (ed.) et al., The Riverside Chaucer.
3rd ed., xlv–xlvii. Boston: Houghton Mifflin Company.
Berdicevskis, Aleksandrs and Hanne Martine Eckhoff. 2016. Redundant features are
less likely to survive: Empirical evidence from the Slavic languages. In: Roberts, S.
G., C. Cuskley, L. McCrohon, L. Barceló-Coblijn, O. Feher and T. Verhoef (eds.), The
Evolution of Language: Proceedings of the 11th International Conference (EVOLANG11),
69–77.
Bergs, Alexander and Gabriele Diewald (eds.). 2008. Constructions and Language
Change. Berlin: Mouton de Gruyter.
Biber, Douglas, Susan Conrad and Geoffrey Leech. 2002. Longman Student Grammar of
Spoken and Written English. Harlow: Longman.
Bischoff, Shannon T. and Carmen Jany (eds.). 2013. Functional Approaches to Language.
Berlin: De Gruyter Mouton.
Bojadžiev, Todor. 1968: Бояджиев, Тодор. 1968. Аналитични форми с глагола имам в
тракийските говори [Analytic forms with the verb have in the varieties of Thrace].
Известия на Института за български език XVI: 459–465.
Bojadžiev, Todor, Ivan Kucarov and Jordan Penčev. 1999: Бояджиев, Тодор, Иван
Куцаров и Йордан Пенчев. 1999. Съвременен български език: фонетика,
лексикология, словообразуване, морфология, синтаксис [Modern Bulgarian:
Phonetics, Lexicology, Word-Formation, Morphology, Syntax]. София: ИК Петър
Берон.
Bowie, Jill, Sean Wallis and Bas Aarts. 2013. The perfect in spoken British English. In:
Aarts, Bas, Joanne Close, Geoffrey Leech and Sean Wallis (eds.), The Verb Phrase in
English: Investigating Recent Language Change with Corpora, 318–352. Cambridge:
Cambridge University Press.
Brinker, Klaus. 1990. Das Passiv im heutigen Deutsch. In: Petkov, Pawel and Diana
Sliwkowa (eds.), Ausgewählte Theoretische Literatur zur Morphologie der Deutschen
Sprache, 398–435. Sofia: Sofioter Universität “Kliment Ochridski”.
Brinton, Laurel J. 1988. The Development of English Aspectual Systems: Aspectualizers
and Post-verbal Particles. (Cambridge Studies in Linguistics 49.) Cambridge:
Cambridge University Press.
Brinton, Laurel J. and Leslie K. Arnovick. 2011. The English Language. A Linguistic
History. 2nd ed. Oxford: Oxford University Press.
Brown, Keith and Jim Miller. 2016. A Critical Account of English Syntax. Grammar,
Meaning, Text. Edinburgh: Edinburgh University Press.
References 337
Burnley, David. 1992. The History of the English Language: A Source Book. London:
Longman.
Burrow, J. A. and Thorlac Turville-Petre. 2005. A Book of Middle English. 3rd ed. Oxford:
Blackwell Publishing.
Bužarovska, Eleni and Liljana Mitkovska. 2010. The grammaticalization of habere-
perfect in standard Macedonian. Balkanistika 23: 43–66.
Bybee, Joan L. 1988. The diachronic dimension in explanation In: Hawkins, John A.
(ed.), Explaining Language Universals, 350–379. Oxford: Blackwell.
Bybee, Joan L. 2003. Mechanisms of change in grammaticization: The role of frequency.
In: Janda, Richard and Brian Joseph (eds.), The Handbook of Historical Linguistics,
602–623. Oxford: Blackwell.
Bybee, Joan L. 2013. Usage-based theory and exemplar representations of construc-
tions. In: Hoffmann, Thomas and Graeme Trousdale (eds.), The Oxford Handbook of
Construction Grammar, 49–69. Oxford: Oxford University Press.
Bybee, Joan L. and Östen Dahl. 1989. The creation of tense and aspect systems in the
languages of the world. Studies in Language 13: 51–103.
Bybee, Joan L. and Paul J. Hopper (eds.). 2001. Frequency and the Emergence of Linguistic
Structure. (Typological Studies in Language 45.) Amsterdam/Philadelphia: John
Benjamins.
Bybee, Joan L., Revere D. Perkins and William Pagliuca. 1994. The Evolution of Grammar:
Tense, Aspect and Modality in the Languages of the World. Chicago: University of
Chicago Press.
Bybee, Joan L. and Sandra A. Thompson. 1997. Three frequency effects in syntax.
Berkeley Linguistics Society, 378–388.
Carey, Kathleen. 1995. Subjectification and the development of the English perfect.
In: Stein, Dieter and Susan Wright (eds.), Subjectivity and Subjectivisation, 83–102.
Cambridge: Cambridge University Press.
Chankova, Yana. 2007a. Possible Scandinavian influence on the grammar of Middle
English: The uses of the perfect in Early Middle English reconsidered. Contrastive
Linguistics XXXII(1): 44–55.
Chankova, Yana. 2007b: Чанкова, Яна. 2007b. Употреба на перфекта в превода на
Уиклиф и превода на Тиндейл в сравнение с оригиналите [Use of the perfect
in Wycliffe’s and Tyndale’s translations compared to the originals]. В: Багашева,
Александра (съст.), За човека и езика. Сборник научни статии, посветени на
60-годишнината на проф. д.ф.н. Майя Пенчева, 510–524. София: Университетско
издателство “Св. Климент Охридски”.
Chankova, Yana. 2008: Чанкова, Яна. 2008. Произход и развитие на аналитичен
перфект в английския език [Origin and development of an analytic perfect in
English]. Дисертация за присъждане на образователната и научна степен
“доктор”. ЮЗУ “Неофит Рилски”, Благоевград.
338 References
Danylenko, Andrii. 2005. Is there any possessive perfect in North Russian? Word 56(3):
347–379.
Davis, Norman. 1987. Language and versification. In: Benson, Larry D. (ed.) et al., The
Riverside Chaucer. 3rd ed., xxix–xlv. Boston: Houghton Mifflin Company.
Denison, David. 1993. English Historical Syntax: Verbal Constructions. London/New
York: Longman.
Denison, David. 1998. Syntax. In: Romaine, Suzanne (ed.), The Cambridge History of
the English Language. Vol. IV: 1776–1997, 92–329. Cambridge: Cambridge University
Press.
Dobrev, Ivan. 1973: Добрев, Иван. 1973. Произход и значение на старобългарския
перфект [Origin and meaning of the Old Bulgarian perfect]. Известия на
Института за български език ХХII: 3–20.
Dolberg, Florian. 2012. Gender change from Old to Middle English. In: Hegedűs, Irén
and Alexandra Fodor (eds.), English Historical Linguistics 2010: Selected Papers from
the Sixteenth International Conference on English Historical Linguistics (ICEHL
16), Pécs, 23–27 August 2010. (Current Issues in Linguistic Theory 325), 263–288.
Amsterdam: John Benjamins.
Dolberg, Florian. 2018. Looking further afield: An interdisciplinary approach to
language contact and grammatical change. Paper presented at the Twentieth
International Conference on English Historical Linguistics (ICEHL 20), Edinburgh,
27–31 August 2018. Powerpoint available at http://www.conferences.cahss.ed.ac.uk/
icehl20/handouts-and-powerpoints/ (accessed on 19 Sept 2018).
Dolberg, Florian. forthcoming. HITting on migration in the murky Middle Ages –
advocating an interdisciplinary approach. In: Thomson, Simon C. (ed.), Strangers
at the Gate! The (Un)welcome Movement of People and Ideas in the Medieval World.
(Explorations in Medieval Culture.) Leiden: Brill.
D’omina et al. 2012: Дьомина, Евгения и съавт. 2012. Речник на книжовния български
език на народна основа от XVII век (върху текст на Тихонравовия дамаскин)
[A Dictionary of Written Bulgarian Based on the 17th-century Vernacular (Based on
the Text of the Tixonravov Damascene Book)]. София: Издателство “Валентин
Траянов”.
Dresher, B. Elan and William J. Idsardi. 1999. Prerequisites for a theory of diachronic
adaptation. Zeitschrift für Sprachwissenschaft 18(2): 212–215.
Drinka, Bridget. 2003. Areal factors in the development of the European periphrastic
perfect. Word 54: 1–38.
Drinka, Bridget. 2017. Language Contact in Europe: The Periphrastic Perfect through
History. Cambridge: Cambridge University Press.
Dujčev and Lixačev 1988: Дуйчев, И. С. и Д. С. Лихачев (ред.). 1988. Среднеболгарский
перевод Хроники Константина Манассии в славянских литературах [The
340 References
Laing, Margaret. 2010. The reflexes of OE beon as a marker of futurity in early Middle
English. In: Lenker, Ursula, Judith Huber and Robert Mailhammer (eds.), English
Historical Linguistics 2008: Selected Papers from the Fifteenth International Conference
on English Historical Linguistics (ICEHL 15), Munich, 24–30 August 2008. Volume I:
The History of English Verbal and Nominal Constructions, 237–254. Amsterdam/
Philadelphia: John Benjamins.
Lass, Roger. 1980. On Explaining Language Change. (Cambridge Studies in Linguistics
27.) Cambridge: Cambridge University Press.
Lass, Roger. 1994. Old English. A historical linguistic companion. Cambridge: Cambridge
University Press.
Łęcki, Andrzej M. 2010. Grammaticalisation Paths of Have in English. (Studies in English
Medieval Language and Literature 24.) Frankfurt am Main: Peter Lang.
Lee, Jeong-Hoon. 2003. The “have” perfect in Old English: How close was it to the
Modern English Perfect? In: Minkova, Donka and Robert Stockwell (eds.), Studies
in the History of the English Language: A Millennial Perspective, 373–398. Berlin, New
York: Mouton de Gruyter.
Lehmann, Christian. 1985. Grammaticalization: Synchronic variation and diachronic
change. Lingua e stile 20: 303–318.
Levinson, Stephen C. 2000. Presumptive Meanings: The Theory of Generalized
Conversational Implicature. Cambridge, Massachusetts: MIT Press.
Lightfoot, David W. 2002. Review of William Croft, Explaining Language Change: An
Evolutionary Approach, 2000. Journal of Linguistics 38(2): 410–414.
Los, Bettelou. 2015. A Historical Syntax of English. Edinburgh: Edinburgh University
Press.
Lutz, Angelika. 2012. Language contact in the Scandinavian period. In: Nevalainen,
Terttu and Elizabeth Closs Traugott (eds.), The Oxford Handbook of the History of
English, 508–517. Oxford: Oxford University Press.
Macleod, Morgan D. 2012. The perfect in Old English and Old Saxon: The synchronic
and diachronic correspondence of form and meaning. Doctoral thesis. University
of Cambridge.
MacRobert, Catherine Mary. 1981. The decline of the infinitive in Bulgarian. DPhil the-
sis. University of Oxford.
MacRobert, Catherine Mary. 2013. The competing use of perfect and aorist tenses in
Old Church Slavonic. Slavia 82(4): 387–407.
Mathieu, Eric and Robert Truswell (eds.). 2017. Micro-change and Macro-change in
Diachronic Syntax. Oxford: Oxford University Press.
McFadden, Thomas and Artemis Alexiadou. 2006. Pieces of the be perfect in German
and Older English. In: Baumer, Donald, David Montero and Michael Scanlon
(eds.), Proceedings of the 25th West Coast Conference on Formal Linguistics, 270–278.
Somerville, MA: Cascadilla Proceedings Project.
348 References
McFadden, Thomas and Artemis Alexiadou. 2010. Perfects, resultatives, and auxiliaries
in earlier English. Linguistic Inquiry 41(3): 389–425.
McMahon, April M. 1994. Understanding Language Change. Cambridge: Cambridge
University Press.
McWhorter, John H. 2002. What happened to English? Diachronica 19: 217–272.
Meillet, Antoine. 1912. L’évolution des formes grammaticales. Scientia (Rivista di
Scienza) 12(26): 6. (Reprinted 1951, Linguistique historique et linguistique generale,
130–148. Paris: C. Klincksieck.)
Menn, Lise, Cecily Jill Duffield and Bhuvana Narasimhan. 2013. Towards an experimen-
tal functional linguistics: Production. In: Bischoff, Shannon T. and Carmen Jany
(eds.), Functional Approaches to Language, 207–245. Berlin: De Gruyter Mouton.
Michaelis, Laura A. 1994. The ambiguity of the English present perfect. Journal of
Linguistics 30: 111–157.
Miller, Jim. 2000. The perfect in spoken and written English. Transactions of the
Philological Society 98(2): 323–352.
Miller, Jim. 2003. Syntax and discourse in Modern Scots. In: Corbett, John, J. Derrick
McClure and Jane Stuart-Smith (eds.), The Edinburgh Companion to Scots, 72–109.
Edinburgh: Edinburgh University Press.
Miller, Jim. 2004a. Perfect and resultative constructions in spoken and non-standard
English. In: Fischer, O., M. Norde and H. Perridon (eds.), Up and Down the Cline: The
Nature of Grammaticalization, 229–246. Amsterdam: John Benjamins.
Miller, Jim. 2004b. Problems for typology: Perfects and resultatives in spoken and
non-standard English and Russian. In: Kortmann, Bernd (ed.), Dialectology Meets
Typology: Dialect Grammar from a Cross-Linguistic Perspective, 305–334. Berlin:
Mouton de Gruyter.
Minčeva, Angelina. 1987: Минчева, Ангелина. 1987. Старобългарският език в
светлината на балканистиката [Old Bulgarian in the Light of Balkan Studies].
София: Наука и изкуство.
Mincoff, Marco. 1958. An English Grammar. Second revised edition. Sofia: State
Publishing House “Science and Art”.
Mirčev, Kiril. 1976[1973]: Мирчев, Кирил. 1976. За съчетанията на глагол имам +
минало страдателно причастие в български език [On combinations of have
+ past passive participle in Bulgarian]. В: Пашов, Петър и Руселина Ницолова
(съставители), Помагало по българска морфология. Глагол, 312–315. София:
Издателство Наука и изкуство.
Mirčev, Kiril. 2000: Мирчев, Кирил. 2000. Старобългарски език. Кратък граматичен
очерк [Old Bulgarian. A Short Grammatical Outline]. Велико Търново: Фабер.
Mitchell, Bruce. 1985. Old English Syntax. 2 vols. Oxford: Oxford University Press.
References 349
Petré, Peter. 2013. Passive auxiliaries in English and German: Decline versus grammati-
calisation of bounded language use. In: Diewald, Gabriele, Leena Kahlas-Tarkka
and Ilse Wischer (eds.), Comparative Studies in Early Germanic Languages: With a
Focus on Verbal Categories, 71–100. Amsterdam: John Benjamins.
Petré, Peter. 2014. Constructions and Environments: Copular, Passive, and Related
Constructions in Old and Middle English. Oxford: Oxford University Press.
Petré, Peter and Hubert Cuyckens. 2008. Bedusted, yet not beheaded: The role of be-’s
constructional properties in its conservation. In: Bergs, Alexander and Gabriele
Diewald (eds.), Constructions and Language Change, 133–169. Berlin: Mouton de
Gruyter.
Piantadosi, Steven T., Harry Tily and Edward Gibson. 2012. The communicative func-
tion of ambiguity in language. Cognition 122(3): 280–291.
Pinker, Steven and Paul Bloom. 1990. Natural language and natural selection. Behavioral
and Brain Sciences 13: 707–784.
Pinto de Lima, José. 1995. Pragmatic maxims in explanations of language change?
In: Jucker, Andreas H. (ed.), Historical Pragmatics: Pragmatic Developments in the
History of English, 197–215. Amsterdam/Philadelphia: John Benjamins.
Plummer, Charles and John Earle (eds.). 1892. Two of the Saxon Chronicles Parallel
with Supplementary Extracts from the Others. A Revised Text. Vol. I. Oxford: Oxford
University Press.
Popova, Genoveva. 1931: Попова, Геновева. 1931. Употреба на минало страдателно
причастие с глагола “имам” в българския език [The use of a past passive par-
ticiple with the verb ‘have’ in Bulgarian]. Известия на семинара по славянска
филология VII: 490–501.
Prokosch, Eduard. 2009[1939]. A Comparative Germanic Grammar. Richmond: Tiger
Xenophon.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik. 1985. A
Comprehensive Grammar of the English Language. Harlow: Longman.
Quirk, Randolph and C. L. Wrenn. 1957. An Old English Grammar. 2nd edn. London and
New York: Methuen.
Radeva, Nija. 2013: Радева, Ния. 2013. От езика на “История славянобългарска” към
езика на нейните преписи. Част 1 [From the Language of the ‘Slavonic-Bulgarian
History’ to the Language of its Copies. Part 1]. Велико Търново: УИ “Св. св. Кирил
и Методий”.
Rainer, Eva Maria. 1989. Das Perfect im Spätmittel- und Frühneuenglischen: eine
Frequenz- und Funktionsanalyse anhand von Brieftexten. Innsbruck: Institut für
Anglistik, University of Innsbruck.
Rikov, Georgi T. 1993/1994: Риков, Георги Т. 1993/1994. Следи от индоевропейските
перфектни формации в славянските езици [Traces of Indo-European perfect
formations in the Slavonic languages]. Български език XLIII/XLIV(1): 1–8.
References 351
Sim, Stuart. 1997. Introduction. In: Griffith, Tom (ed.), The Concise Pepys, v–xi. London:
Wordsworth Classics of World Literature.
Smirnickaja et al. 1977: Смирницкая, О. А., Мирра Моисеевна Гухман, Л. С.
Ермолаева. 1977. Историко-типологическая морфология германских языков:
Категория глагола [Historical-Typological Morphology of the Germanic Languages:
Verbal Categories]. Москва: Институт языкознания (Академия наук СССР).
Smith, K. Aaron. 2001. The role of frequency in the specialization of the English an-
terior. In: Bybee, Joan L. and Paul J. Hopper (eds.), Frequency and the Emergence of
Linguistic Structure, 361–382. Amsterdam/Philadelphia: John Benjamins.
Stambolieva, Maria. 2008. Building Up Aspect: A Study of Aspect and Related Categories
in Bulgarian, with Parallels in English and French. Frankfurt am Main: Peter Lang.
Stankov, V. 1980: Станков, В. 1980. За характера на опозицията преизказност в
българския език [On the nature of the evidentiality opposition in Bulgarian]. В
чест на академик Владимир Георгиев: Езиковедски проучвания, 485–491. София:
БАН.
Stoevsky, Andrei. 1999. Remarks on the status of the English perfect. In: Mantcheva,
Alexandra (ed.), Mélanges de linguistique, sémiotique et narratologie: dédiés à la
mémoire de Krassimir Mantchev à l’occasion de son 60e anniversaire, 375–380. Sofia:
Colibri.
Stoevsky, Andrei. 2006. The collocation “имам ‘have’ + the past passive participle” in
Bulgarian. В: Куцаров, Иван, Васил Райнов, Мери Лакова, Владко Мурдаров
и Светла Коева (ред.), Езиковедски изследвания в чест на 75-годишнината
от рождението на ст.н.с. І ст. д.ф.н. Йордан Пенчев Пенчев, 142–152. София:
Артграф.
Straaijer, Robin. 2010. Prescription or practice? Be/have variation with past participles
of mutative intransitive verbs in the letters of Joseph Priestley. In: Lenker, Ursula,
Judith Huber and Robert Mailhammer (eds.), English Historical Linguistics 2008:
Selected Papers from the Fifteenth International Conference on English Historical
Linguistics (ICEHL 15), Munich, 24–30 August 2008. Volume I: The History of English
Verbal and Nominal Constructions, 63–78. Amsterdam: John Benjamins.
Strang, Barbara M. H. 1970. A History of English. London: Methuen.
Szmrecsanyi, Benedikt. 2016. About text frequencies in historical linguistics:
Disentangling environmental and grammatical change. Corpus Linguistics and
Linguistic Theory 12(1): 153–171.
Tagliamonte, Sali A. 2000. The grammaticalization of the present perfect in English:
Tracks of change and continuity in a linguistic enclave. In: Fischer, Olga, Anette
Rosenbach and Dieter Stein (eds.), Pathways of Change: Grammaticalization in
English, 329–354. Amsterdam: John Benjamins.
Tagliamonte, Sali A. 2004. Have to, gotta, must. Grammaticalisation, variation and
specialization in English deontic modality. In: Lindquist, Hans and Christian Mair
References 353
Linguistics (ICEHL 15), Munich, 24–30 August 2008. Volume I: The History of English
Verbal and Nominal Constructions, 217–235. Amsterdam: John Benjamins.
Xaralampiev, Ivan. 2001: Харалампиев, Иван. 2001. Историческа граматика на
българския език [A Historical Grammar of the Bulgarian Language]. Велико
Търново: Фабер.
Zifonun, Gisela, Ludger Hoffmann and Bruno Strecker et al. 1997. Grammatik der
deutschen Sprache. Band 3. Berlin: Walter de Gruyter.
Zimmermann, Rüdiger. 1968. Untersuchungen zum frühmittelenglischen Tempus
system. Doctoral dissertation. Kiel.
Zimmermann, Rüdiger. 1973. Structural change in the English auxiliary system: On the
replacement of BE by HAVE. Folia Linguistica 6: 107–117
Appendices
Kuteva, Tania 28, 242, 245n10, 255, 261n23, 103n15, 104n16, 105n17, 106, 107n19,
267n28 108–109, 119n25, 120–121, 122n26, 124,
Kytö, Merja 21n10, 45n11, 48n13, 49, 50n15, 126, 127n29, 137n32, 139n35, 141–142, 149
51, 54n19, 55n20, 61n28 Mitkovska, Liljana 256n17, 260, 261n23, 262
Mladenov, Stefan 264–265, 267n26, 268
Labov, William 18 Mugglestone, Lynda 50n15
Laing, Margaret 45n11, 46 Mustanoja, Tauno F. 34n1, 41n6, 42, 43n8,
Lass, Roger 12n4, 13, 16, 17n7, 54, 83 44n9, 45n11, 48n13, 49n14, 51, 52n16, 53,
Łęcki, Andrzej M. 12n5, 20, 22–26, 33, 34n1, 54n19, 57, 59n25, 87, 149, 157
35n3, 36, 39, 40n5, 41n6, 42–44, 46n11,
48, 51, 52n16, 53, 54n19, 57n22, 59n23, Načeva, Mira 244n7
64, 65n31, 66n32, 67n34, 68, 69n38, Nakamura, Fujio 49, 52n16, 149n1, 157,
70n39, 71–72, 73n41, 75–76, 79, 165n11, 181n18, 207, 209–210, 217
246n10, 267n28 Newmeyer, Frederick J. 9, 12n5, 17–18
Lee, Jeong-Hoon 36, 49 Nicolova, Ruselina 240, 241n2, 242n3,
Lehmann, Christian 221 245n10, 246n11, 251, 254n16, 256n18,
Levinson, Stephen C. 22, 27n13 267n26, 278, 320
Lightfoot, David W. 13 Norde, Muriel 9
Lixačev, Dmitrij S. 273
Los, Bettelou 42, 46, 50, 63, 93, 141, 151, 225 Peev, Dimităr 302–303, 305–307, 309n14,
Lutz, Angelika 162 310, 313n18
Penčev, Jordan 246, 251n15, 252
Macleod, Morgan D. 20, 22, 25, 33, 34n1, Pennisi, Antonino 9
35n3, 37, 45n11, 81, 86–90, 92–94, Petré, Peter 59, 212, 224n1
102–103, 110, 130, 149, 224, 267n28 Piantadosi, Steven T. 16, 220
MacRobert, Catherine Mary 242n4, 244, Pinker, Steven 9, 17
273–274, 287, 294–295 Pinto de Lima, José 12
Mathieu, Eric 11 Plummer, Charles 90–91, 95, 108, 111,
McFadden, Thomas 33, 47, 50n15, 51, 52n16, 121–122
57n22, 80, 224n1 Popova, Genoveva 247n12, 265
McMahon, April M. 15 Prokosch, Eduard 46n11, 135
McWhorter, John H. 59, 80 Pullum, Geoffrey K. 34, 35n2, 67–68, 73n41
Meillet, Antoine 20, 22
Menn, Lise 87 Quirk, Randolph 33–34, 38, 56, 57n21,
Michaelis, Laura A. 33, 157n7 65–67, 70, 73, 157, 185
Millar, Robert McColl 25, 35n3, 40, 62,
289n5 Radeva, Nija 302
Miller, Jim 33, 35n2, 36, 74–76, 157n7, 185 Rainer, Eva Maria 203
Minčeva, Angelina 244n7, 269 Rikov, Georgi T. 242n4
Mincoff, Marco 35n3, 45n11, 50n15, 55n20, Ringe, Don 18n8, 30
57n21, 59, 61n28 Rissanen, Matti 46, 48n13, 49, 53, 54n19,
Mirčev, Kiril 242n4, 244n7, 245n8, 257–258, 55n20, 56, 57n22, 59, 63–64, 72, 172,
302 175n14
Mitchell, Bruce 33–34, 35n3, 36–39, 40n5, Ruseva, Dimitrina 273–274, 278
41n6, 42–43, 44n9, 45n11, 46–47, 48n13, Rydén, Mats 34n1, 46, 48n13, 49n14, 50n15,
49, 50n15, 53, 55n20, 57n22, 59n24, 51, 52n16, 53n17, 54n19, 55n20, 56,
60n26, 60n27, 62, 66n32, 67n33, 68, 57n22, 61n28, 62, 75n42, 80, 89, 97, 152,
70n39, 73, 81n47, 93n3, 95, 96n7, 97, 157–158, 162–163, 172, 174, 182, 194, 209,
98n9, 98n10, 98n12, 99n13, 100, 101n14, 210n25, 211–212, 219, 223, 224n1, 225
Author index (modern authors) 359
imperfect (tense) 19, 240–241, 259, 272, 196, 198–200, 207, 210–213, 224–225,
274–275, 286, 288–289, 294–295, 229–231, 242–243, 261–262
303–304, 311 (see also aorist and
preterite) NICE (properties) 65, 67, 73n41
imperfective (aspect) 33, 240–241, 256n17 non-witnessed (see evidential)
impersonal (construction) 55n19, 62n30, null exponence (see zero marking/zero
111n20, 124, 143n, 154n4, 199, 224, exponence)
246–247, 256, 263, 265, 307, 310, 313,
320–321, 334 Optimality Theory 12n5
indeterminacy/indeterminate (see
ambiguity) paradigm 35n2, 35n3, 39–40, 80, 96, 99,
inferential (see conclusive) 107n, 127–128, 130, 253, 260, 318, 331 (see
intransitive (verb/clause) 28, 36, 41, 45–46, also declension)
48–51, 54–58, 60, 69, 80, 82, 92, 102, 110, parole (performance) 220 (see also langue)
119–123, 128, 131, 139–142, 145, 148, participle (see converb, past participle and
150–154, 158–164, 168–169, 171–172, present participle)
179–180, 184, 187–188, 196–198, 200, passive 19, 36, 44n10, 50, 54–66, 73, 76–77,
202, 206–208, 211–213, 215–216, 218–219, 79, 82–86, 92–98, 101–102, 104–111,
221, 224–225, 228–230, 238, 242–243, 119–131, 134–145, 148–161, 165–168,
256, 261–263, 267–268, 270, 330, 332 172–173, 176–182, 184–185, 189, 191–196,
(see also mutative and transitive) 198–201, 205–207, 211–219, 221, 223,
invisible hand 12–13, 31, 220 224n1, 226–229, 231–233, 235–239, 247,
252, 254n, 256, 264, 266, 272, 274,
language contact 5–6, 28, 59, 68, 70n39, 285–286, 329–331 (see also past passive
80–83, 85, 93, 123, 144–146, 163–164, 167, participle)
172, 203, 220, 222–223, 237–239, 244n, be 59, 105–106, 108–109, 121, 124, 127–129,
257–260, 262, 265–274, 281, 287–289, 139–140, 145, 151, 161, 168, 176, 184, 201,
294, 302–303, 316, 326–331, 334 (see 213, 217, 219, 221, 223, 226–229, 232–233,
also Balkan Sprachbund and 235, 238, 241–242, 251n, 266, 274,
Carolingian Sprachbund) 288–289, 294n, 297–298, 303–304, 313,
langue (competence) 220 (see also parole) 327, 333 (and general entries above)
layering 23–24, 65, 77, 253 become (weorðan/werden) 58–60, 73, 86,
lexicalisation 301 98, 106–109, 121–122, 125–128, 137–138,
LFG 21n9 145, 223, 226–229, 232
light verb (see delexical verb) have 61, 63–65, 67, 72, 187, 201, 213–214,
217, 227–228, 256n18, 320
mechanical/mechanistic 13–15, 19, 77, 146, get 72–75, 217, 221, 227n, 333
222, 266, 333 (see also random) perfect passive 41, 103, 114n, 124, 127n29,
merger/(near-)merger 18, 269 140, 155, 171, 176–177, 185–186, 202,
modal(-ity) 24, 28, 43, 51, 67–72, 73n, 75, 142, 204–205, 219, 236, 314
148, 150, 152, 161, 163–165, 168, 174–176, pseudo-passive 56, 254n
184, 187–191, 196, 200, 202, 205, 211, reflexive passive 297–298, 307, 310
213–215, 222, 224, 228–229, 231–232, Vorgangspassiv 226–229
244–245, 259–260, 270, 283–284, Zustandspassiv 226–229
322–324, 327 past participle 33–42, 44–48, 55–63, 65–67,
mutative (intransitive verb) 45, 48–49, 51, 71–72, 74, 76–77, 82, 86, 93–101, 103–105,
53–54, 56, 94, 122, 148, 151–152, 154, 107, 112–116, 119–123, 129–130, 132–134,
161–162, 168, 170–173, 177, 180, 184, 187, 136–140, 153–155, 158–160, 162n, 163,
Subject index 365
165, 171, 173, 177–178, 186, 188, 197, 199, perfective (aspect) 32–33, 51, 138n33,
201, 203–205, 211, 215, 218n, 219, 224, 240–241, 256n17
226, 236, 238, 241–242, 248–253, phase 34
255–256, 258, 263, 266, 269–270, 331, phonological change (see sound change)
333 phrasal verb 57, 197n
past active participle 8, 36, 46n12, 60, pluperfect (see past perfect)
238, 241–242, 245–247, 259, 264–268, polygrammaticalisation 25, 43, 72, 75,
270, 275, 282–285, 292, 295n, 302, 303n, 82–83, 146, 174–175, 189, 195, 201, 213,
305–310, 317, 319–322, 330–333 215, 222, 237, 322, 327, 333 (see also
past passive participle 8, 36, 46n12, 60, grammaticalisation)
166, 238, 241–242, 245–247, 251, possession/possessive (construction) 24–
261–262, 264–268, 270, 275, 285–286, 25, 34–38, 42–43, 48, 62–65, 68–71,
297–298, 313, 314n, 316–321, 325, 327, 74–76, 81, 83, 103–104, 114, 145, 171, 204,
331–333 214, 222, 244–245, 248, 251, 254–256,
perfect 24–25, 29, 32–34, 48–63, 65, 71–87, 268, 289, 298, 314–316, 322, 324–325,
89–90, 92–96, 98, 102–105, 109–116, 333–334
119–124, 126–135, 139–146, 148–164, pragmatic inferencing/implicatures 22–23,
165n, 166–174, 178–189, 192–204, 26–27, 153, 222
206–231, 233–239, 245, 257, 259, Prague School 12
264–267, 272, 274–284 (see also perfect prefix (see verbal prefix)
passive and perfect progressive) prescriptive (grammar/grammarian) 20,
be 45–48, 52–54, 59–61, 74–77, 79–83, 86, 49, 56
94–95, 98, 105, 119–123, 126–129, present participle 47, 51, 71–72, 99, 104, 159,
139–140, 143–146, 148, 150–154, 156–159, 176, 206, 214–215, 218n
161–164, 166–168, 172–174, 179–184, preterite (past tense) 24, 29, 33, 44, 77–78,
187–189, 192–196, 198, 201, 203, 80, 87, 93, 103, 111, 116, 120–121, 136,
206–208, 210, 212–213, 215, 217–227, 139–142, 153, 156, 157n6, 165, 172, 185,
229–231, 233–239, 241–247, 251–254, 208, 215, 224, 236–237, 267n28, 272, 275,
257–261, 263, 266–267, 270, 272, 286, 289, 292, 294, 302, 304 (see also
274–275, 286, 288–298, 301–307, 311, aorist and imperfect)
317–319, 322, 326–334 Principles and Parameters 12n5
become (weorðan) 58–59, 119, 121–124, progress 28–29
127–129, 231 progressive (aspect) 32, 41, 47, 54, 55n20,
clipped perfect 61 (see also ellipsis) 56, 60n27, 79, 81–82, 92, 99, 102, 104,
double perfect 57 108, 124, 131, 143n, 168, 174–177, 184,
get 73–75, 329, 333 189–190, 194, 196, 200, 204–206, 211,
have 28, 34–45, 65, 67, 72, 79–83, 86, 94, 213–215
100, 103–104, 112, 116, 119, 122, 127–130, passive progressive 204–205
132–135, 139–140, 143–146, 148–154, 157, perfect progressive 41, 164, 171, 175,
161–164, 167–171, 174, 179–182, 184–188, 186–187, 189
195–198, 201–202, 204, 206–207,
210–215, 219, 221, 224–231, 233–239, 242, random(-ness) 29, 31, 221–223 (see also
245–265, 267–272, 286, 296, 298, 302, arbitrariness/arbitrary and mechanical/
314, 316–323, 325–334 mechanistic)
own 41, 43 reanalysis 15, 22–23, 25–27, 31, 35–36, 39–41,
past perfect 33, 51, 87, 89, 92–93, 116, 120, 60–61, 68, 74, 82–83, 85–86, 130, 145, 171,
141, 149, 153, 158, 163, 177, 202–203, 212, 222, 238, 265–267, 270, 327, 330–331,
219, 253, 259–261, 274–276, 280, 288 333–334
366 Subject index