Probleme der Ägyptologie Herausgegeben von
Wolfgang Schenkel und Antonio Loprieno
26. BAND
The Modal System of Earlier Egyptian Complement Clauses A Study in Pragmatics in a Dead Language
By
Sami Uljas
LEIDEN • BOSTON 2007
This book is printed on acid-free paper. Library of Congress Cataloging-in-Publication data is available on https://catalog.loc.gov.
ISSN 0169-9601 ISBN 978 90 04 15831 3 © Copyright 2007 by Koninklijke Brill NV, Leiden, The Netherlands. Koninklijke Brill NV incorporates the imprints Brill, Hotei Publishing, IDC Publishers, Martinus Nijhoff Publishers and VSP. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA. Fees are subject to change. printed in the netherlands
table of contents
v
TABLE OF CONTENTS Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ix
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.1.1 Orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.1.2 Assertion and Non-assertion . . . . . . . . . . . . . . 0.1.3 The System in Earlier Egyptian and the Structure of the Present Work . . . . . . . . . . . . . 0.2 Methodology, Terminology, Morphology and Sources
1 1 1 13 23 26
PART ONE MODALITY IN AFFIRMATIVE COMPLEMENT CLAUSES AFTER GOVERNING VERBS 1.
Introduction to Part One . . . . . . . . . . . . . . . . . . . . . . . . . .
39
2.
Affirmative Object Complementation after Notionally Assertive Verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 2.1 Locution: The Verb Dd ‘Say’ and Indirect Speech . . 50 2.1.1 Core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 2.1.2 An Example Text . . . . . . . . . . . . . . . . . . . . . . . 66 2.2 Verbs of Cognition and Perception . . . . . . . . . . . . . . 73 2.2.1 Seeing is Believing and Other Truisms: Realis, Irrealis, Knowledge and Perception . . . . . . . . . 73 2.2.2 The ‘Evaluative’ Use of Irrealis . . . . . . . . . . . . 84 2.3 Interim Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 2.4 The Paradigm after ntt/wnt: Form and Function Revisited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.
Affirmative Object Complementation after Notionally Non-assertive Verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 3.1 wD and dbH: Between ‘Ontological’ and ‘Attitudinal’ Modality . . . . . . . . . . . . . . . . . . . . . . . . . 126 3.2 Verbs of Preventing and rdi . . . . . . . . . . . . . . . . . . . . 140
vi
table of contents 3.3 3.4 3.5
Pragmatics of Desire: the Verb mri . . . . . . . . . . . . . . 153 Some Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 Interim Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 PART TWO
MODALITY IN OTHER TYPES OF EARLIER EGYPTIAN COMPLEMENT CLAUSES 4.
Modality in Affirmative Subject Complement Clauses . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Asserted versus Non-asserted Subject Complements . 4.3 The Verb xpr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
173 173 181 194
5.
Modality in Negative Complement Clauses after Governing Predicates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Assertion: Complement Clauses with iwt and ntt n 5.3 Non-assertion: tm and nfr-n . . . . . . . . . . . . . . . . . . . .
203 203 206 210
6.
7.
8.
Modality in Complement Clauses after Prepositions . . . . . 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 The Typology of Irrealis in Complement Clauses of Preposition: the ‘Obvious’ Instances . . . . . . . . . . . 6.3 ‘Indicative Non-indicatives?’ The Broader System of Preposition Complement Modality in Earlier Egyptian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Clauses with m-xt . . . . . . . . . . . . . . . . . . . . . . . . . . . .
221 221
238 260
Earlier Egyptian Supplementary Patterns of Complementation after Verbs and Prepositions . . . . . . . . . 7.1 The sDm.n=f . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 The Particle is . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3 The Element r-Dd . . . . . . . . . . . . . . . . . . . . . . . . . . . .
264 264 278 284
225
Predicate Complement Clauses . . . . . . . . . . . . . . . . . . . . . . 287 8.1 The Syntax of Type B Nominal Sentences with Clausal Predicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
table of contents 8.2
8.3 9.
vii
Assertion and Non-assertion in Bipartite Nominal Sentences with Clausal Predicates and the Limits of ‘Nominal’ Predication . . . . . . . . . . . . . . . . . . . . . . . . . 296 Interim Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
At the Crossroads of Tempus and Modus: the AspectualModal Correspondence and the Conceptual Foundations of Irrealis Modality in Earlier Egyptian Complementation 9.1 The Interrelation between Temporal and Modal Functions of the Active Suffix-Conjugation Forms in Complementation and Beyond. . . . . . . . . . . . . . . . 9.2 Too Far or Too Close? The Modal-Aspectual Continuum of the Earlier Egyptian Irrealis-Marked sDm=f Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Interim Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . .
321 329
10. Conclusion: Retrospect and Prospect . . . . . . . . . . . . . . . . . . 10.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 The Diachronic Status of the System . . . . . . . . . . . . . 10.3 Extending the Hypothesis . . . . . . . . . . . . . . . . . . . . . . 10.4 Final Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
335 335 341 347 360
308
309
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363 Index of Cited Passages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379 General Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419 Index of Languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427 Key to Abbreviations and Symbols Used . . . . . . . . . . . . . . . . . . 429
table of contents
ix
ACKNOWLEDGEMENTS This book represents a revised and extended version of my 2005 University of Liverpool doctoral thesis. The original work underwent a series of transformations before reaching its current form. At one point it was turned into a shorter study on complements of verbs that in 2004 gained me a research fellowship at Trinity College, University of Cambridge. Certain parts of it also integrate results from work carried out in the framework of the project ‘Basel Diachronic Grammar’, which began in 2005. I would like to express my gratitude to all those who have assisted me and in various ways taken part in the preparation of this book: Dr Mark Collier for his supervision of and comments on the work on which this volume is based as well as to my examiners Dr Christopher Eyre and Dr Jean Winand for their critical remarks. Prof. Antonio Loprieno and Prof. Wolfgang Schenkel not only deemed my work worthy of being published in the series Probleme der Ägyptologie, but also offered invaluable criticism and suggested many improvements to the text. Prof. James P. Allen similarly drew my attention to certain issues that I had overlooked as well as kindly provided me with information and material not available in a published form. Mr Michiel Swormink of Brill Publishing offered assistance of a more technical kind, besides displaying considerable patience and flexibility as regards matters such as dates of submission of the manuscript etc. A special word of gratitude I would like to express to my friend and colleague Dr Matthias Müller with whom I discussed a great many topics treated below and who, as the first critical reader of the manuscript, saved me from various embarrassing slips and inaccuracies. Of those that still remain, I bear the sole responsibility. Also my own and my wife’s family have acted as an ever-reliable supporting team at every stage of my research. Finally, my most heartfelt thanks I wish to tender to my wife Tuuli, to whom I am happy to dedicate this work. Cambridge, December 2006 Sami Uljas
introduction
1
INTRODUCTION 0.1 Preliminaries 0.1.1. Orientation “With the means at our disposal it is not possible to distinguish different moods in Egyptian, if such existed”. Thus Sir Alan Gardiner in his Egyptian Grammar, still justly the most admired and widely used work on the Ancient Egyptian language in general and Middle Egyptian in particular.1 After nearly eighty years since Gardiner first made his brief remarks, the role of modality—a cover term for a wide array of subjective speaker ‘attitudes and opinions’—and its different manifestations in the grammar of Earlier (Old and Middle) Egyptian remain relatively little explored.2 Indeed, gleaned through the prism of modern Egyptology, the early language of the pharaohs, the medium of some of the earliest and most cultivated literary and philosophical works in world history, appears as somewhat of a linguistic anomaly. Of the three verbal ‘meaning domains’ of tense, aspect and modality, (henceforth tam) the first two seem to have had a notable role in the grammatical organisation of the language, whereas the third is at best marginally represented. Earlier Egyptian appears to have been manifestly a ‘talanguage’ largely bereft of the sophistication of expression attained by means of grammatical mood and modality in almost all other linguistic systems, both ancient and modern. The present work seeks to elevate the role of modality in Egyptological linguistics by proposing a novel, semantic-pragmatically based analysis of the grammar of Earlier Egyptian complementation, with a focus on fully clausal complements.3 Traditionally, complement clauses have been viewed as subordinate construals whose basic character is analogous to 1
Gardiner 1957 (henceforth GEG) § 294. It is not the purpose of the present work to provide a historical overview of modality in Egyptological linguistics; the most relevant works pertaining thereto will be noted and discussed as occasion arises. ‘Earlier Egyptian’ is defined here as in Loprieno 1995, 5–6, but excluding the so-called Égyptien de tradition. 3 With the infinitive, Earlier Egyptian governing verbs occasionally develop meanings different from their lexical semantics, some of which are clearly modal (see Uljas 2003, 393–95). However, the infinitive itself has no TAM-profile and is not marked for 2
2
introduction
substantives. Rather akin to nouns, in English a clause may function e.g. as an object complement of a transitive verb as in (i) below, or as a subject complement of e.g. a passive predicate as in (ii):4 (i) (ii)
Jill knew [that Jack had left] [That Jack had left] was known (by Jill).
Besides entering the structure of the main clauses, the subordinate propositions in (i)–(ii) also flesh out the semantics of the superordinate verb. Characteristically, the former constitutes a compulsory, albeit not a very prototypical, semantic argument of a main (matrix) verb.5 Yet, this is only insofar as the governing verb is a truly lexical predicate. In Egyptian as elsewhere, auxiliary verbs also occur with syntactic complements, but the argument-projecting facilities of the former are no longer functional.6 Further, the complement clause need not be associated with a verb at all, but may also appear e.g. as a predicate of a nominal sentence, as in (iii) below, or as a complement of a preposition (iv):7 (iii) (iv)
Jill’s opinion was [that Jack had left] It was a question of [whether Jack had left].
In Earlier Egyptian, clausal object- and subject complements of verbs display two main types of construal.8 In the affirmative, the more
person/number/gender. Consequently, infinitival complement clauses fall beyond the scope of the present study. 4 However, particularly in object complementation this provides merely the most basic blueprint for the structure, which is much less static than might seem. See chapter 1 for a brief discussion. 5 Notably, a clausal complement can never serve e.g. as an affected patient, which is the most typical semantic role of nominal objects of active and subjects of passive verbs. 6 As a signal of this, the complements themselves are almost invariably infinitival (see Uljas 2003, passim for Egyptian). Auxiliation is a diachronic process and auxiliaries are derived from earlier fully lexical verbs. One case of such a development in Earlier Egyptian is discussed in depth in 4.3 below. 7 Cf. Ransom 1986, 29 n.2. This shows e.g. that complements are not necessarily ‘entailed’ by the semantics of some other state of affairs (contra Cristofaro 2003, 38, 95). 8 In addition to the construction types illustrated below, clausal complements may also be used as direct or indirect genitives in Earlier Egyptian (see GEG §§ 191–92; Gunn 1949). Discussion of these must be left to a further occasion.
introduction
3
prevalent of these is directly embedded bare, or un-introduced, suffix-conjugation forms, typically a sDm=f of some sort.9 In example (1) below, the clause qnn=f functions syntactically as the object of the governing matrix predicate mAA, ‘see’, as does Hs Tn nTr=tn in relation to the verb mri ‘want’ in (2) and sipw sw wrty-HqAw in relation to wD ‘order’ in (3). In example (4), mss=s serves as a clausal subject of the adjective verb qsn ‘be difficult’: (1)
Amenemheb testifies to the valour of his royal master on battlefield:
iw wHm.n=i {n} mAA qnn=f iw=i m Smswt=f Again I saw how brave he was, when I was in his following. (Urk IV 892, 6–7)
(2)
From an Appeal to the Living:
[i anxw tpw] tA iw.t(y)=s[n] r [b]iA pn mr=Tn Hs Tn nTr=tn Dd=Tn [xA] m t Hnqt O the living upon earth who will come to this mining-region; as you will want your god to favour you, may you say: “A (Sinai 167, 1–3) thousand bread and beer…” (3)
The deceased is stated to be under Osiris’ command:
iw wD.n wsir sipw sw wrty-HqAw Osiris has ordered the two Great-of-Magic to examine him. (CT I 131a-b/L2Li)
(4)
The story of the birth of three kings describes their mother’s travails:
wa m nn hrw xpr wn.in rwd-Ddt Hr Snt=s qsn mss=s One of these days, Redjedet was suffering because her labour (pWestcar 9, 21–22) was difficult. As can be seen, the un-introduced sDm=f of mutable roots occurs in various different forms and sometimes shows characteristic endings -w and -y. In example (1) the ultimae infirmae root qni shows gemination (qnn), whereas in (2) the root Hsi from the same morphological class does not (Hs). In example (3) the caus. 2rad root sip shows an ending -w.
9
Discussion of complements with bare sDm.n=f will be postponed to 7.1 below.
4
introduction
These morphological peculiarities have, as shall be discussed shortly, aroused much speculation among Egyptologists. The second principal type of affirmative complements after verbs is clauses introduced by the element ntt/wnt:10 (5)
The king ends his letter announcing his accession to his Nubian viceroy:
hAb pw r rdit rx=k st ntt pr-nsw aD wDA This is a correspondence to let you know it, and that the royal (Urk IV 81, 2–3) house is sound and prosperous. In Earlier Egyptian clauses may also appear in various other complement environments, most notably as objects of prepositions used as conjuncts. In the affirmative, suffix-conjugation forms, particularly sDm=f, may again appear directly embedded, as in examples (6)–(8) below, or the complement may be introduced by ntt (rarely, wnt) as in example (9): (6)
Sinuhe expresses his opinion to his host on the motives of an adversary:
rqt-ib [pw] Hr mAA=f wi Hr irt wp[wt]=k It is but envy because he sees me carrying out your business. (Sin R 141–42)
(7)
Hapdjefa gives instructions on the use of a taper by an official performing his mortuary-cult:
Aw=f s(y) n Hm-kA=i r-sA iri=f irt=f im=s m Hwt-nTr He is to pass it on to my kA-priest after he has done what he is (Siut I 297–98) to do with it in the temple. (8)
From a medical instruction on treating a bleeding swelling:
dd=k Hr=s Ast nt ds dd=k st r hAw snf Whenever you put a splinter of firestone on it, apply it in (pEbers 88, 19) such a way that blood can flow down.
10
Less common methods of introducing complements will be discussed in 7.2 and 7.3 below.
introduction (9)
5
Amenemhat I, assassinated whilst taking repose, advices his son from beyond the grave:
sDr=k sAw n=k ib=k Ds=k Hr ntt nn wn{n}11 mr n s hrw n qsnt When asleep, keep yourself alert, because no man has a servant (Amenemhat IIIa–b) on a difficult moment. Fully clausal complements occur also as predicates in bipartite nominal sentences with the element pw. However, in this class of complement structures there exist no ntt/wnt-introduced variant to the type of sentence exemplified in (10) below with the bare geminating sDm=f. The construal in example (11) is modally equivalent to the former, but it is affiliated to a wholly different set of construals than complement clauses and has its own semantic-pragmatic characteristics, which explain its peculiar form: (10)
A medical gloss explaining a term:
ir nnw mwyt=f hAA mwyt pw m Hnn=f n sA.n=s n=f As for ‘his urine is lazy’; this means urine issues from his (pEdwin Smith 10, 21–22) penis without control. (11)
Queen Hatshepsut tells the background for her decision to erect two large obelisks for Amun:
ink pw snDm.n=i m aH sxA.n=i qmA wi I relaxed in the palace, and I took thought of the one who (Urk IV 364, 16–17) created me. In negative complements, the same overall formal division appears. Again, after verbs and prepositions the complement may be unintroduced, in which case the negative verb tm or the negative construction nfr-n appears. In introduced complements, the element iwt (or, alternatively, ntt + n-negation) is used: (12)
Hapdjefa describes his impeccable conduct in life:
ink dr bXbX m qA-sA sgr qA-xrw r tm=f mdw I was one who removed pride from the arrogant and silenced the loud-mouthed one so that he would not speak. (Siut I 229) 11
So pMillingen 1, 5; other variants have correctly wn.
6 (13)
introduction The king notes to his envoy concerning a dwarf that the latter is bringing:
Dd.n=k xr Hm(=i) iwt zp in.t(i) mit(y)=f in ky nb You have said to my majesty that never has his like been brought (Urk I 129, 2–3) by anyone else. In negative nominal sentence predicate complements only un-introduced clauses with tm are attested: (14)
A medical gloss explaining a term:
ir r=f mr... tm=f wn r=f pw mdw=f As for ‘his (the patient’s) mouth is tied’… this means that he does not/cannot open his mouth so that he might speak. (pEdwin Smith 4, 2–3)
The current opinio communis among students of Egyptian is that the guiding parameters of this grammatical and morphological organisation are primarily syntactic in character. The roots of this conception lie to a notable degree in wissenschaftsgeschichtliche trends peculiar to Egyptological linguistics. Early discussions of Earlier Egyptian complementation were already characterised by a rather scant concern with the particulars of meaning of clausal complements, most scholars contenting themselves with descriptions of their syntax and typology.12 However, the structural orientation in the analysis of the grammar of these constructions was, it seems, perpetuated by the so-called Standard Theory (henceforth ST) of Egyptian based on the work of H.J. Polotsky.13 Although since the early 1990s many of the basic tenets of the ST have been challenged, the effects of the ‘Polotskyan revolution’ are still widely felt in Egyptological linguistics and, arguably, nowhere more so than in the manner which the grammar of complement clauses is viewed. Prior to the advent of the ST, the use of the different complement patterns had been described in terms of dialectics between form and function. 12 See e.g. Sethe 1899, §§ 142, 150–01, 366; Erman 1928, §§ 356, 523–25a, 531– 32d; GEG §§ 154–57, 183–92, 223; Lefebvre 1955 (henceforth LGEC) §§ 612–16, 687–711; DeBuck 1952 §§ 119, 126; Edel 1955–64 (henceforth EAG) §§ 1014–27; Westendorf 1962 (henceforth WGMT) § 436. 13 For bibliographies and summaries of the rise and key hypotheses of the ST, see Depuydt 1983; 1995; Junge 1989, 11–18; Schenkel 1990, 145–58.
introduction
7
For example, according to Gardiner, in un-introduced complements the various verb-forms employed as subjects and objects of verbs and after prepositions had these syntactic roles only ‘virtually’ without being specialised for such uses.14 In the ST, however, form and function were largely synonymous; the geminating sDm=f and its non-geminating counterpart, analysed as a separate (or various separate) ‘prospective’ form(s), were interpreted as nominal forms or nominal ‘transpositions’. Their ability to function directly as complements, as well as their negation therein by tm, seen as the negator of all ‘nominal forms of the verb’, was cited as particularly transparent testimony of this inherent syntactic nature.15 One of the most persuasive arguments seemingly favouring this interpretation is the conspicuous absence of the bare geminating sDm=f/tm from the paradigm following the elements ntt/wnt as well as iwt.16 Accordingly, these elements were seen to allow forms and constructions somehow syntactically inappropriate for ‘nominal’ use to function thus.17 The ST analysis of the assumed inherent syntactic properties of the forms and elements employed in complementation as the primus motor of their grammar not only shared some resemblance with structuralist theories aiming at mathematising language and greatly in vogue between 1950s and 1970s. It also seemed, for the first time, to provide a model capable of accounting for all complement patterns by reference to the shared criterion of ‘nominality’. For example, before the ST there had been no explanation to the use of ntt, which e.g. Gardiner describes merely
14 GEG § 182; this is a clear functionalist formulation of grammatical use; see 0.2 below. 15 Polotsky 1944, §§ 25, 28, 30–31; 1964, 276–77; 1969, 470; 1976, 2.7.2, 2.7.5, 2.7.6; 1987, 19; De Cenival 1972, 41–42; Frandsen 1975, 18–20, 56–57, 69; Schenkel 1975, 41; Silverman 1985, 281; Depuydt 1983, 29–30; 1993, 19 among others. tm is characterised as ‘nominal’ already in Sethe 1899 § 994. Further proof of this was also seen in the obvious morpho-syntactic similarities between the ‘nominal’ and relative forms, the former of which were, from early on, seen as actual ‘abstract relative forms’ (Polotsky 1944, § 25 and passim; most recently Allen 2000, chapter 25 and Schenkel 2006). This topic will not be discussed in the present work, but certain remarks pertaining thereto will be found in the conclusion (section 10.3). 16 Cf. Polotsky 1944 §§ 25, 28. 17 Gilula 1970, 213; 1971, 16; Junge 1979, 84; Satzinger 1986, 307; 1989, 216; Allen 1986b, 25–27, 33; Silverman 1985, 272; 1986a, 38; cf. also Doret 1986, 34 n.264. Already Gunn (1924, 176) and Erman (1928, § 531; 1933 § 723) refer to ntt/wnt as ‘nominalisers’. Hypotheses on the precise syntactic character of the forms/constructions after ntt/wnt/iwt are discussed in 2.4 below; for the ‘prospective’, see 0.2 and 2.4 below.
8
introduction
as an element “occasionally used for ‘that’” after selected verbs and prepositions, without proposing an explanation as to what might condition its use.18 However, the ST ‘nominal hypothesis’ entailed an assumption of the grammar of complement clauses (and Egyptian in general) as almost wholly driven by syntactic rules. In many discussions within this school of thought one discerns a tendency to regard the ‘transpositions’ system an sich as a sufficient explanation for grammatical phenomena. Polotsky himself reserved some of his most ‘revolutionary’ fervour against the earlier views of Gardiner et al when discussing just the complement-uses of his ‘nominal forms’, at times even arguing that they represented purely syntactic entities wholly beyond the system of tam.19 Junge’s dismissal of the complement form-variation as mere hesitation of ‘speakers’ intuition’ or as nebulous irregularity in the ‘surface structure’, expressive of tam ‘Zusatzinformation’ represents the most explicit formulation of this theoretical vantage point.20 Yet, most ST treatises did usually not share such extreme views. Indeed, the Polotskyan School never sought a complete breakaway from earlier views that had stressed the role of tempus, particularly verbal aspect, in the organisation of the different complement patterns.21 Also Polotsky’s own later work is characterised by increasing stress laid on the importance of semantics (again mostly aspect) in the verbal system of Earlier Egyptian, particularly in case of the geminating sDm=f.22 Nevertheless, the ST analysis of complementation as based on incorporation of nominal verb-forms and nominalised clauses into their predetermined syntactic slots has proved to be lasting. At pace with the gradual rise of the ST as the most widely accepted paradigm for analysing Egyptian, also in the domain of complementation the number of approaches deviating from its precepts was reduced to naught.23 Among the advocates of the ST or more broadly in the 18 GEG §§ 187, 237; regarding prepositions + ntt, Gardiner notes that these complexes primarily express cause, (§ 223) but not in all instances. See 6.3 below. 19 Particularly the geminating sDm=f; Polotsky 1964, 281. 20 Junge 1978a, 98, 104, 109. 21 See Schenkel 1990, 151 for this point. 22 See Polotsky 1976, 2.3.1 and chapter 9.2 below. 23 Even the few approaches seeking an alternative to the ST make concessions to its tenets in case of complement clauses. For example, Thacker (1954, 331) claims that by Middle Kingdom the geminating sDm=f had come to express ‘energic’ sense—except in complementation, for which he accepts Polotsky’s analysis of the forms used as relative
introduction
9
intellectual climate permeated by it, there appears to have been little enthusiasm for specifically studying the semantic or pragmatic characteristics of the ‘nominal forms’ in complementation. Practically all post-1950s and pre-1990s discussions of the sDm=f-forms and tm in these construals assume their putative ‘nominality’ therein. Also in the current predominantly ‘post-Polotskyan’ landscape of Egyptological linguistics, views on the grammar of complement clauses continue to reflect the deeply entrenched conception of its syntactic underpinning. The assumption of the ‘nominal’ or ‘nominalised’ character of the bare complement sDm=f forms and tm/nfr-n-clauses, as well as the role of ntt/wnt as ‘nominalisers’ or nominal converters of various ‘nonnominal’ patterns not ‘specialised’ for such use still serves as the point of departure for well-nigh all discussions, differences in theoretical meta-language notwithstanding.24 The hypotheses on tempus/aspect and on the other semanticpragmatic characteristics of Earlier Egyptian complementation will be discussed later in the present work. Yet, due to its fundamental status in all research since Polotsky’s seminal work, the assumption of the role of syntax as the driving force behind the grammatical organisation of complement constructions must be addressed presently. In spite of the admirable consensus prevailing among researchers, at closer inspection the current ‘nominal hypothesis’ of Earlier Egyptian complementation turns out to be bedevilled by various sorts of problems. The most immediately apparent difficulties pertain to the mutual distribution of the bare ‘nominal forms’ and ntt/wnt. The ‘nominal hypothesis’ suggests that complements with and without the latter are in principle mutually substitutable. The tacit assumption seems to be that, as syntactic ‘converters’, ntt/wnt themselves are essentially meaningless. However, their actual distribution does not support this view. For example, ntt and wnt are widely used to introduce object complements forms of some sort (209). Even Gardiner, in his otherwise highly critical review of Polotsky 1944, felt obliged to state that it contained “no more penetrating pages” than those devoted to tm (1947, 99 n.4). Borghouts (1985, 36) argues that the use of the geminating form is motivated solely by temporal factors, but notes also that in complementation the ‘nominal’ character of the verb is undeniable. Yet, since all suffix-conjugation forms (allegedly) originate in nouns, here this inherent character merely “shimmers through” (for an outright rejection of this argument, see Satzinger 1993a, 205). 24 See e.g. Collier 1990a, 83–84; 1991a, passim; 1999, 57; Grandet & Mathieu 1997, 406, 444; Malaise & Winand 1999 §§ 574, 601–03, 611, 612–18, 895–923; Allen 2000, 137. This claim has been made by the present author as well (Uljas 2000, 130).
10
introduction
of verbs of locution, cognition and perception, but they never occur e.g. after verbs of volition, ordering or preventing:25 †
iw mr.n=f ntt ink Dd=i st He wanted that it would be me who will say it †
iw dbH.n=f ntt Dd=s n=f st He asked that she would say it to him †
iw wD.n=f ntt=f r Dd st He ordered that he say it
†
iw xsf.n=f ntt Dd=f st He prevented him from saying it
Similarly, there are only certain prepositions that combine with ntt/ wnt-clauses, whereas many combinations such as e.g. †m ntt, †n-mrwt ntt or †n-aAt-n ntt are never found.26 This clearly has nothing to do with the syntactic nature of the construals following ntt/wnt, and it remains unclear why such restrictions should apply if the role of these elements is simply to ‘nominalise’ the following construal and if the clauses thus introduced represent mere “analytic counterparts” or substitutes of the ‘nominal forms’.27 There are also certain gaps in the ‘nominalising’ function of the ‘converters’. For example, ntt/wnt cannot ‘nominalise’ such clearly ‘non-nominal’ expressions as e.g. adverbial phrases:28 † †
iw mA.n=f ntt m pr/aA He saw that in the house/here
The same holds also for fully clausal adjuncts introduced by prepositions: 25
Cf. Uljas 2000, 126; 2003, 388. However, at this juncture it should be noted that in many analyses the elements ntt/wnt after prepositions are not interpreted as introducing complements at all, but are seen to form non-isomorphic ‘adjunct connectors’ of the type ‘preposition-ntt’ with the preceding prepositions. For discussion of this topic, see 6.3 below. 27 Polotsky 1976, 2.3.1; Satzinger 1986, 306; cf. also Meltzer 1991, 227. 28 Cf. Collier 1990a, 78. 26
introduction † †
11
iw mA.n=f ntt xft irr=f st He saw that when he does it
Worse still, as will be discussed later on, there are main verbs that accept some or all of the so-called ‘non-nominal’ forms and constructions as their objects and subjects without a ‘converter’, and a similar anomaly is found in nominal sentences, where un-introduced ‘non-nominals’ nevertheless appear as predicate complements.29 Yet, the syntagmatic properties of the alleged ‘converters’ have aroused remarkably little interest. As before the ST halcyon days, comments to this effect have largely amounted to mentions that they occur only after ‘certain verbs’ and prepositions.30 Most researchers have focussed on their paradigmatic character and the nature of the forms and constructions after ntt/ wnt rather than on the clauses as a whole. These are, naturally, of fundamental importance.31 However, the restrictions in distribution of the ‘converters’ noted show that it is at least equally important to understand where and when ntt/wnt can or cannot occur and why. Moreover, there is a more general issue involved here. If ntt and wnt were truly required to introduce forms and constructions syntactically unsuited for being embedded as complements, one may justly ask why did they need to be so embedded at all if the language allegedly possessed also a set of ‘nominal forms’ as the ‘analytic counterparts’ of many of them? Why were the syntactically ‘unsuitable’ patterns not simply replaced by the ‘nominal forms’ consistently? Why was the clause ntt pr-nsw aD wDA in example (5) above not replaced by aD wDA pr-nsw, i.e. ‘nominal sDm=f’s’ of some sort? What motivated the use of ntt + the ‘unsuitable’ stative here but not e.g. in example (1), where the complement is similarly an intransitive state, but now appears as qnn=f, a bare geminating sDm=f? In a similar vein, if prepositions are assumed to govern ntt-complements, it may be asked what exactly prompted the use of ntt + nn wn in example (9) above, when a bare preposition Hr + the ‘nominal’ negation tm wn mr n s hrw n qsnt could, in principle, have been ‘substituted’ in its stead? Issues and questions such as these make it abundantly clear that it simply 29
See 2.4; 4.3 and chapter 8 below for all these construction types. Gilula 1970, 213; 1971, 16. Cf. also e.g. Allen 1986a, 11; 1986b, 25; Doret 1986, 28 n.146; 34 n.263; 48; Grandet & Mathieu 1997, 444, 446. 31 See 2.4 below for a full discussion. 30
12
introduction
does not suffice to insist that ntt/wnt are required for purely syntactic reasons when they obviously are not so used universally in what is, after all, constantly the same syntactic environment. Consequently, in order to understand what determines their use, clauses with ntt/wnt should be treated first as units. After the parameters conditioning their occurrence have been clarified, one may turn into the question whether or not the forms and patterns they contain have some shared feature(s).32 Just as analysing ntt/wnt as mere ‘nominalisers’ or ‘converters’ seems thus to be an oversimplification, the alleged ‘nominal’ character of the bare suffix-conjugation forms used in complementation is similarly disputable. In terms of finiteness, valency, and the manner in which the bare complement forms project arguments, they clearly behave as verbs, not nouns. Alternatively, the label ‘nominal form’ should perhaps not be understood simply as denoting a ‘noun equivalent’, but rather to mean something like ‘nominalised verb’. However, when assessed against properties whose clustering is usually seen as prototypical indicators of such a status of verbs, the bare suffix-conjugation forms used in complementation fare rather poorly.33 In these environments, they can indeed be seen as behaving as the syntactic ‘head noun’, but they do not e.g. acquire nominal morphology,34 do not mark their subjects or objects as genitives, do not lose their ability to express tam, nor accept nominal determiners such as articles, demonstratives, adjectives, numerals or relative clauses. Yet e.g. with the infinitive, a truly ‘nominal’ verb-form, most of these criteria are met.35 Thus, since they are clearly not substantives, but neither fit the profile of verbal nominalisations, it is wholly unclear exactly what sort of ‘nominals’ or ‘syncategorisations’ the bare forms are supposed to be. The ‘nominal’
32
This latter issue will be the topic of section 2.4. Sasse 2001; Givón 2001, vol.2, 25, 35; for Egyptian in particular, see Claudi & Mendel 1991. 34 Cf. Polotsky 1944 § 26.1–2; Gardiner 1947, 98; Reintges 1997, 115 n.10. The gemination displayed by the ultimae infirmae roots is surely not to be seen as such (cf. Eyre 1986, 134) particularly as this would not apply to the ‘nominal’ ‘prospective’ and sDm. n=f form(s). 35 Thus pA prt ‘the going’, prt iri.n=f ‘the going which he did”, prt 3 ‘three goings’ etc. The only exception here is the feature of nominalising morphology, which is universally absent from infinitives. 33
introduction
13
character of the geminating sDm=f, the sDm.n=f 36, the alleged ‘prospective’ and the negation tm is also seriously in doubt generally, and not just in complementation. All the counter-arguments above against the ‘nominal’ nature of these forms in the latter environment apply similarly to their other ‘nominal’ uses. Further, the assumption of the ‘nominal’ character of the ‘nominal forms’ in second tenses, the most widely attested use of the geminating sDm=f, rests on a dubious syntactic analogy (not similarity, as in complementation) with adverbial sentences plus a highly suspect theory on ‘predication’.37 It is also disputable whether any ‘nominal sDm.n=f’ exists at all, 38 and insofar as there are one or more ‘prospective’ sDm=f forms, these also occur in patently non-nominal environments such as in final ‘so that’-clauses, which even the most ingenious ST postulates have failed to analyse as nominal in character.39 The last property pertains also to tm, whose capability to negate adjunct clauses is in fact not restricted to final clauses,40 and which is also the negation of sequential sentences with the elements ix, xr and kA.41 When it is added that there are good reasons to doubt the entire division of the sDm=f forms in Earlier Egyptian,42 the assumption of ‘nominality’ as the founding principle in the grammar of complement clauses in this language seems unlikely to be correct and should be seriously reconsidered. 0.1.2 Assertion and Non-assertion It is argued in the present work that instead of syntactic characteristics of the forms and constructions employed, the grammatical organisation 36 The issue whether the sDm.n=f in the ‘nominal’ uses differs morphologically from that employed in ‘non-nominal’ environments will be discussed in 7.1 below. 37 See Collier 1990a, 80–82; 1992 and 10.3 below. 38 See 7.1 below for full discussion. 39 Attempts to divide the ‘prospective’ sDm=f into ‘adverbial’ and ‘nominal’ forms (Allen 1982, 25; Depuydt 1993a) as well as hypotheses of the possibility of using ‘nominal’ forms ‘adverbially’ (Allen 1984 § 290; Doret 1986, 43) have been rife. The initial uses have been argued to represent complements of some zero ‘ø main predicate’ (Schenkel 1975, 41; 1978, 113–14; Junge 1978a, 122; Depuydt 1983, 46; Allen 1984 § 255; 1991, 7–8; Kammerzell 1988, 41; Grandet & Mathieu 1997, 253. Jansen-Winkeln 1995 extends this to even final uses; for counter-arguments, see Uljas 2000, 126– 27) or ‘single noun clauses’ (Polotsky 1964, 271). 40 See 5.3 below for a brief discussion. 41 There has been no shortage of ST attempts to interpret constructions with ix/xr/ kA as ‘nominal’, see Satzinger 1968, §§ 73, 78 for one. 42 See 0.2 below.
14
introduction
of Earlier Egyptian complement clauses is based on modality and differentiation between asserted and non-asserted complements. Since these concepts belie great complexity, a discussion and illustration of their meaning and nature is indispensable. The term assertion refers to a particular kind of speech act, the performing of which is regulated by the co(n)text of the communication and its intent. Asserting requires the following preparatory conditions to be fulfilled:43
The S(peaker) has a reason (evidence etc.) for believing in the veracity of the p(roposition); S is committed to p; p is not obvious to both S and the H(earer).
In addition, and following from the last criterion above, the speaker’s aim must be that of informing H of p and of his commitment thereto, typically in view of convincing H of p. This requirement for assertion is known as the speaker’s illocutionary intention:
S assumes that saying p is relevant to, and in the interest of H; S wishes H to believe p and that S is committed to p.
Thus, p is asserted if S indicates some degree of positive commitment towards the information it conveys and the latter is not something that S knows or assumes to be already known by H or lacking in relevance—i.e. asserted propositions have a proportionally high information value. Conversely, and abstracting away from the more precise taxonomy of speech acts, propositions in which one or more of these criteria are not fulfilled represent non-assertions. This provides a notional definition for assertion and non-assertion as particular kinds of illocutionary forces. However, the grammatical coding of the notional concepts is an altogether different issue, and here assertion and nonassertion become a matter of modal organisation. In general terms, modal systems and grammatical coding for modality are activated by the status of propositions as assertions or non-assertions. In particular, the use of grammatical indicative or realis corresponds to expression 43 See Searle 1969, 67; Bach & Harnish 1979, 42; Edmondson 1981, 145; Levinson 1983, 277; Allan 1986, 193; Wierzbicka 1987, 321; cf. also Allan 1998, 925–26; Langacker 1991, 496.
introduction
15
of asserted propositions; in case of non-assertion, languages resort to non-indicative or irrealis patterns. The formal method of coding these two categories displays notable variation. Realis may be overtly indicated as such, but is usually unmarked,44 whereas expression of the often more marked irrealis ranges from particles and modal auxiliaries to verbal inflection, primus inter pares the subjunctive mood of numerous (particularly Romance) ‘Standard Average European’ languages.45 For example, in Spanish the use of the subjunctive is directly linked to non-assertion; a proposition may appear in the subjunctive if46 The speaker has doubts of its veracity (v below) It describes an unrealised situation (vi) The information conveyed is presupposed (vii) (v)
Dudo que sea (sub) buena idea “I doubt that’s a good idea”
(vi)
Necesito que me devuelvas (sub) ese libro “I need you to return that book to me”
(vii)
Me alegra que sepas (sub) la verdad “I’m glad that you know the truth”
Particularly the first and the last of these properties fail the preparatory condition for assertion: doubt equals lack of commitment and presupposition represents unchallenged information which the speaker assumes to be already obvious to, and accepted by the audience.47 44
Cf. Givón 1982, 155; 2001 vol.1, 330–31. Many languages are said to have a specific ‘irrealis’ rather than subjunctive mood, but this is largely a (Whorfian) research tradition: many modern linguists seem to have opted to employ the label ‘irrealis’ in their descriptions of various ‘exotic’ languages, perhaps to stress their uniqueness (cf. Palmer 2000, 185). Yet, both subjunctives and the ‘irrealis’ are grammaticalisations of the same notional category of non-assertion. The traditional view of the subjunctive, apparent in the etymology of the word itself, (Italian subiung , ‘subjoin’) is that it is the mood of subordination. This old assumption has been rightly discarded in linguistics; its inadequacy is clear from the data below. 46 Data from Lunn 1995, 430; see also the fundamental study by Hooper & Terrell 1974. 47 It is to be noted that presupposition is not tantamount to logical ‘necessarily true’. Presupposition as a linguistic concept refers solely to assumption of shared knowledge between speakers and hearers; it is perfectly possible to assert propositions whose logical 45
16
introduction
Presupposed information is part of the discourse background and hence lacks relevance: communicating what is already shared cannot be the speaker’s illocutionary intention and remains, by definition, unasserted. Notably, the above examples involve complement clauses, and there are no syntactic reasons barring the use of the indicative. Instead, the subjunctive is chosen for each proposition because “potentially assertable information must have two qualities: it must be both reliable… and informative as to news value. Information that is lacking in either of these qualities… is unlikely to be asserted”.48 However, the words ‘potentially’ and ‘unlikely’ in the last quote already raise various issues concerning this seemingly simple set of principles that are decisive for the understanding of Earlier Egyptian complementation. Assertion and non-assertion are prototype-based, rather than ‘classical’ Aristotelian categories whose all members are equally representative of the category as a whole: both can be viewed as encompassing certain core semantic notions surrounded by a peripheral domain of other more or less ‘good examples’ of (non-)assertion.49 As a result, languages vary considerably in what they classify as modally realis and irrealis within and between themselves, and to what extent. Particularly in irrealis, devices for expressing gradations of non-assertion are commonplace. Thus the grammatical treatment of e.g. epistemic expressions, i.e. speaker attitude towards the authenticity-status of a proposition, as either realis or irrealis universally reflects the degree of tentativeness expressed. Cf. the following Spanish and English expressions:50 (viii)
Tal vez me-estaba (ind)/me-estuviera (sub) esperando “Perhaps s/he was/might have been waiting for me”
truth-value is co(n)textually uncancellable, provided that the speaker does not assume this to be common ground with the hearer; cf. Hooper & Terrell 1974, 485; Lambrecht 1994, 61–63; Palmer 2000, 3–4; Cristofaro 2003, 30. This issue is also relevant with socalled ‘factive’ and ‘implicative’ verbs (see chapter 1 and sections 2.2.2 and 3.2 below). 48 Lunn 1989b, 691, emphasis in the original. 49 See Lunn 1989b and Bybee 1998 in particular. Thus e.g. when referring to irrealis-use, many authors prefer to speak of ‘reduced assertion’ (Bell 1980) or use expressions such as ‘suspension of affirmation’ (cf. Lunn 1989b, 687). 50 Spanish data from Givón 2001 vol.1, 314–15; cf. 1994, 301.
introduction (ix)
17
He’s a spy; I know it He must be a spy: look at him He may be a spy, who knows”
Much of deontic modality, i.e. expression of will, is grammatically irrealis across languages. But deontic is also subject to gradations of politeness, the assessed potential for control and manipulation over the ‘targets’ of the attitude, and thus in the likelihood of the state of affairs referred to.51 Consequently, the strongly demanding imperative often varies with less direct modal auxiliaries or subjunctives/jussives as in the English and Biblical Hebrew examples (x)–(xi) below. In Caddo, (N. Iroquoian) affirmative imperatives are realis-marked whereas negative imperatives and ‘obligative’ expressions (‘should’) are irrealis, reflecting the degree of expectation of compliance and realisation (xii):52 (x)
Go!
(xi) (xii)
You must/should/may/might go
( שמרimp)
( תשמרjuss)
“Guard!”
“May you guard”
(real) (irr) (irr)
“Look at it” “Don’t look at it” “He should sing”
This also shows an important faculty of ‘non-realisation’ vis-à-vis irrealis. There is no requirement that situations treated as irrealis be ‘unreal’. It is not ontology or, again, ‘truth values’ that determine the modal status of propositions. These concepts are logico-philosophical ‘absolutes’, whereas irrealis relates to speaker assessments of low information value, as in case of most certainly ‘real’ presupposed and their “judgments concerning the degree to which their ideas accord with what they believe to be objective reality”.53 Hence languages often divide the expression of e.g. futurity into ‘more’ and ‘less’ ‘objective’ types. The following examples from Central Pomo (N. California) illustrate this well:54 51
Cf. Givón 1975a; 1994, 298–300; Fleischman 1989, 8–12; Mithun 1995, 377. Hebrew and Caddo data from Greenberg 1965, 51 and Chafe 1995, 356-58 respectively. For further comments on ‘obligatives’, see chapter 6 n.23 below. 53 Chafe 1995, 364; emphasis by SU; cf. also Givón 2001 vol.1, 302 and Bendix 1998, 253–54. 54 Data from Mithun 1995, 370, 378–79. 52
18
introduction
(xiii)
•• (real) •= “We’ll go around eating” (certain) •=• (irr) • == “I’ll go to town and bring it back” (less certain)
Another more familiar example of the same phenomenon is the variation in Earlier Egyptian between the bare initial ‘prospective’ sDm=f and the ‘pseudo-verbal’ iw=f r sDm, recognised by most commentators as that between ‘subjective’ and ‘objective’ future respectively.55 The latter tends to occur when there are sufficient grounds for the speaker to be committed to the eventual realisation of the situation, which makes it more assertable. It is no accident, then, that e.g. first person intentions are mostly expressed by iw=i r sDm, as people tend to be more committed to and in control of their own actions than of those of others, or that the same pattern is the favourite also when the situation is somehow expectable through ‘experience’, ‘norm’, or e.g. divine prognostication.56 The use of such variants is not based on any more ‘objective’ criteria than approximation of likelihood and possibility of the future state of affairs; the final say on what to consider realis or irrealis belongs to the speaker.57 This is also apparent in the way in which speakers may withhold from asserting if e.g. they do not accept the proposition, as in the Albanian example (xiv) below, or because they do not wish to imply that what they say represents their own commitment, as e.g. in German ‘journalistic’ style (xv):58
55
See e.g. Hannig 1982, 47; Vernus 1990, 24–27 and numerous others. Thus for example, Peas B1, 42 mk wi r nHm aA=k sxty “I am going to take your donkey, peasant” does not portray the speaker as “compelled to seize the donkey as a logical punishment of the transgression” (Vernus 1990, 14) but expresses a firm conviction to carry out the said act. However, these words do not protest the speaker’s willingness to act, unlike e.g. the long string of first person bare sDm=f’s in Sh.S. 139–46: i.e. “I will (=am willing to) spread your fame at home” etc. Similarly, the standard expression iry=i r Hst=k “I will do as you ask” (EAG § 472) expresses readiness and willingness to obey, but no personal motivation as such; the variant iw=i r irt r Hst=k is notably rare. 57 A fine empirical study to this effect in Spanish is Lavandera 1983. 58 The Albanian ‘admirative’ usually expresses surprise, but as here, it can also indicate mockery and non-acceptance, both of which represent reduced assertion. See Haiman 1995, 333–34 for this and for similar examples from other languages. For the German data, see Palmer 2000, 42; in French, the conditional is similarly used, see Fleischman 1995, 533. 56
introduction
19
(xiv)
E na i dashka (adm) bullgaret.. ai e.. e pse keshtu u pritka nje Bullgar? “And he ‘likes’ Bulgarians. Him? Ha! After all, is that how you treat a Bulgarian?”
(xv)
Er sei (sub) mit S in Streit geraten und habe (sub) sich von diesem bedroht gefühlt “He had allegedly become involved in a quarrel with S and had felt threatened by him”
Also the requirement of ‘informativeness’ and relevance from asserted propositions is a very flexible concept.59 For example, in Spanish pairs such as the following are not uncommon:60 (xvi)
Aunque es (ind)/sea (sub) mi hija, la encuentro muy guapa “Although she is my daughter, I find her very pretty”
Here the indicative signals that the information about ‘she’ being ‘my daughter’ is communicated both as actual and relevant, whereas the subjunctive codes it as concessive and presupposed. Variation like this reflects speaker evaluation of the proposition information value and ‘background-ness’ and does not follow from any mechanical rule. Understanding it here requires particular attention to the co(n)text, as speakers’ motives for treating propositions as more or less informative can only be recovered by examining the discourse as a whole rather than isolated clauses or sentences. Further, there is no hard and fast limit to what speakers may and may not evaluate as ‘informative’. This depends on the quality of their performance in a given linguistic and even ‘extra-linguistic’ context. For example, in Spanish the subjunctive can signal that the audience is assumed as already ‘reasonably’ acquainted with the information, and the extent to which this can be taken is quite remarkable. In journalistic Spanish a prior mention of some proposition in the headline may suffice to prompt the use of the subjunctive when it is resumed in the main text:61 59 A superb discussion of the concept of ‘relevance’ and communication in general is Sperber & Wilson 1995. 60 Data from Lunn 1989b, 697–98; cf. Klein 1975, 364 n.8. 61 Data from Lunn 1995, 433.
20 (xvii)
introduction Headline: La bandera que besó es la que, en su día, también besó (ind) el Rey don J, y bordó (ind) su tatarabuela la Reina doña M. “The flag that he kissed is the one that one day king J also kissed and his great-great-grandmother queen M embroidered” Text: al final, besó la bandera roja y gualda que hace treinta años besara (sub) su padre el Rey y que un día bordara (sub) su tatarabuela la Reina doña M. “At the end, he kissed the red and gold flag that his father the king had also kissed thirty years ago, and that his greatgreat-grandmother queen M had once embroidered”
The next example derives from a gossip-magazine. Its author assumes the information in the relative clause to be ‘old news’ to faithful readership and indicates this with non-assertion:62 (xviii)
La pareja, que se hiciera (sub) famosa por interpretar el papel de marido y mujer en El pájaro espino, es (ind) en la vida real un matrimonio feliz “The couple, who became famous for their role as husband and wife in ‘The Thorn Birds’, is happily married in real life”
Here the principal information with the highest ‘news value’ is in the main clause, where the indicative appears. However, e.g. Russian and Polish dispense with the notion of ‘relevance’ from their grammaticalisation of modality altogether and code even presupposed information as indicative.63 This emphasises that the grammaticalisation of meanings as realis and irrealis is also a crosslinguistic variable. There are certain senses such as e.g. counterfactual,
62 Data from Lunn 1989a, 254; 1989b, 693; 1995, 433; cf. also Lavandera 1983, passim. Interestingly enough, since the antecedent of the relative clause is definite, the use of subjunctive should be ungrammatical. Yet, this rule is clearly ignored herein in favour of greater expressiveness—see also n.69 below. 63 Noonan 1985, 99; Wierzbicka 1988, 152.
introduction
21
whose non-assertivity is somehow signalled in almost every language.64 Yet the treatment of e.g. interrogatives, negatives and conditionals, all notional non-assertions, varies greatly. For instance, simple negated propositions in Alamblak (Papuan) are irrealis-marked, but realis in Central Pomo:65 (xix)
fiñji noh-r-fë-r (irr) “He did not die”
(xx)
=í í (real) “Because they didn’t keep up the rancheria...”
This is because negation can be viewed either as expression of total lack of commitment (negated assertion) or of commitment to falsity (negative assertion)—or kept outside the system of modality altogether, as in most European languages. Finally, assertion and non-assertion overlap conceptually with semantic categories other than modality, and this has grammatical consequences. For example, in some languages certain types of verbal aspect prompt irrealis-marking due to perceived similarity with ‘unreality’. Thus e.g. in Caddo, infrequent (xxi) and in Bargam (Papuan) past habitual propositions (xxii) are irrealis:66 (xxi)
(irr) “I seldom see it”
(xxii)
miles-eq leh-id (irr) “when (the pig) used to return…”
These meanings involve characteristics that set them metaphorically ‘close’ to irrealis. Something taking place seldom may be subjectively ‘as good as nothing’, and in Caddo this conceptual ‘family resemblance’ results in grammatical marking of such situations as irrealis. Habitual refers to a mass-like series of events rather than any one particular 64 For representative samples of languages, see D. James 1982, 378–84; Roberts 1990, 392. For counterfactuals in Earlier Egyptian, see 6.2, end and 7.1 below. 65 Data from Roberts 1990, 390 and Mithun 1995, 382 respectively. 66 Data from Chafe 1995, 357 and Roberts 1990, 384.
22
introduction
situation occupying a specific locus in time. The degree of individuation of the ‘sub-events’ is low and the proposition may even refer to a mere tendency.67 Thus habitual events may be viewed with precisely that “lack of belief in or lack of commitment to… the reality, realization, or referentiality of an event or sequence of events”68 characteristic to irrealis.69 This apparent lack of ‘coherence’ of realis and irrealis has led to occasional doubts of their validity as grammatical concepts.70 However, it could be argued that it rather shows their fundamentally scalar nature. The basic classes of non-assertion and assertion receive almost as many expressions in modal systems as there are languages and they can be further modulated into a notable degree. Yet, there is remarkable agreement across languages on e.g. what sorts of nonassertions constitute the ‘nucleus’ of irrealis and qualify for modal coding as such. Overall, however, realis and irrealis form a continuum of linguistic meaning without clear-cut lines of demarcation mutually or with other categories of grammar. In addition, and as a corollary of the previous point, irrealis and realis meaning is associated with linguistic expressions (verb-forms, particles etc.) in a non-arbitrary manner. In certain languages, the metaphorical linkage between e.g. habitual aspect and irrealis results in a linguistic mapping of both onto the same or co-occurring grammatical expressions. The task of the linguist is to define and explain this sort of procedures in the grammar of the specific language under study.
67 Cf. Givón 1994, 270–71; 2001 vol.1, 305; Fleischman 1995, 537–39; Palmer 2000, 179, 190–01. This is reflected in various ways. For instance, nouns under the scope of the habitual ‘hybrid modality’ are non-referring: in “he used to buy a newspaper every day”, ‘newspaper’ is non-referential and lacks a specific counterpart in ‘reality’. Note also the (particularly American) English use of ‘would’ instead of ‘used to’ for past habitual. 68 Fleischman 1995, 522, 537. 69 Irrealis and realis marking of verbs is sometimes affected even by case-marking of associate nouns and their definiteness; see e.g. Martin 1998, 199. In Romance, attributive relative clauses of indefinite nouns commonly employ the subjunctive; cf. n.62 above.. 70 See e.g. Bybee 1998. In Bybee et al 1994, 236–39 it is maintained that irrealis and realis are not appropriate concepts for assertion and non-assertion, which they see as the decisive factors in modal organisation. See Givón 1994, 320–28 and Palmer 2001, 188–91 for counterarguments.
introduction
23
0.1.3. The System in Earlier Egyptian and the Structure of the Present Work In Egyptology attitudes towards even the general status of modality in Earlier Egyptian have tended to be somewhat reserved.71 Polotsky’s dismissal of anything like an ‘echter modus’ in Egyptian reverberated through the writings of his followers,72 and the ST preoccupation with syntax sidelined modality in favour of speculations on structure and tense/aspect.73 In complementation, modality has hardly been discussed at all, although some researchers have e.g. proposed geminating sDm=f complements to portray the situation as ‘objective fact’ or ‘indicative’ and those with a bare sDm=f without gemination to entail “an element of possibility or doubt”.74 Most have had little to say about the pragmatic profile of ntt/wnt/iwt-clauses,75 and comments on modality and tm/nfr-n-complements are extremely scarce.76 The term ‘assertion’ has been applied to the illocutionary force labelled thus—particularly in connection with the auxiliary iw, which has been characterised as an ‘assertion-particle’, but assertion has mostly not 71 However, Gardiner’s stance in this matter, quoted in 0.1.1 above, did not amount to denial of modality in Egyptian a priori, but merely recognised the then current lack of suitable methodological tools for exploring this phenomenon. These are apparent e.g. in Sander-Hansen’s (1941) inspired, but perhaps premature attempt to distinguish moods in Egyptian. In Late Egyptian, however, there has recently been a surge of extremely interesting research, such as Winand’s fascinating study on the conjunctive, (2001) which substantially extends the pioneering work by Borghouts (1979) and demonstrates the irrealis character of this construction beyond doubt. 72 Polotsky 1964, 272; one may note e.g. the marginal remark by Depuydt 1993a, 13 n.7 or Callender’s (1986, 11) claim that on ‘internal grounds’ the notion of mood is redundant in Egyptian, justly disputed by Junge (1989, 39 n.62). 73 An example of the results of this is the synopsis of modality in Earlier Egyptian future expressions by Vernus, (1990) which is praiseworthy for its pioneering approach but which is unfortunately marred by a too great focus on tempus and adherence to ST dogmas. This problem is only partially remedied by reviewers (Ritter 1992; Schenkel 1992). In some treatises modality and pragmatics generally are seen to have given way diachronically to syntax as the driving force in grammar, including that of ‘originally’ modal patterns such as the geminating sDm=f or tm in the mature ‘syntactic system’ of Classical Egyptian (Loprieno 1991a, 215, 223; 1995, 82). 74 Doret 1986, 23, 39, 49; see also Gilula 1971, 16; Allen 1984, §§ 365, 703, 706; Malaise & Winand 1999, § 908. However, there has been some divergence in views as to how many different ‘prospective forms’ the latter disguises (see 0.2 below). 75 Notable exceptions here are Satzinger 1968 § 100, Allen 1986a, and Loprieno 1991a, 214. See 2.1.1 and 5.1 below for discussion. 76 But see Loprieno 1986b, 280–81; 1991b, 231–35 and 5.3 below. Silverman (1985, 281) and Vernus (1990, 119) note tm to be the negation of both geminating and non-geminating sDm=f-complements, which may indicate belief that it inherits the properties of both.
24
introduction
been treated as a modal concept.77 ‘Irrealis’ has been evoked but seldom, and always as a synonym for ‘unreal’.78 It is argued here that when understood as modal categories, realis and irrealis are analysable as the determining concepts behind the grammatical organisation of Earlier Egyptian complementation. The system can be summarised as follows: 1. The basic division of complement clauses into introduced and un-introduced types corresponds to differentiation between assertion and non-assertion. When the prerequisites for assertion are satisfied, the subordinate clause is specifically marked as asserted by the modal operators ntt/wnt or the negative iwt. Such complements are modally realis. 2. If the complement is not asserted, ntt/wnt or iwt are not used, and the clause remains un-introduced. In the affirmative, bare suffix-conjugation forms are employed. The patterns mostly used are various types of sDm=f, but also the bare sDm.n=f has its role to play in the expression of non-assertion. In the negative tm and nfr-n appear. Although less marked than ntt/wnt-clauses, complements of this sort are modally irrealis. 3. Unlike in many modern languages with a singular conjugated irrealis, e.g. a subjunctive mood, in Earlier Egyptian unintroduced sDm=f forms divide up the domain of irrealis and non-assertion in a principled and meaningful way. Their mutual variation in complementation serves to differentiate between hierarchically ‘stronger’ and ‘weaker’ non-assertion. However, in negated clauses, the modal (and formal) distinctions between these two degrees of non-assertion are neutralised and they share the common irrealis negations tm and nfr-n. 77 See Assmann 1974, 65; Junge 1989, 105; Hannig 1982, 43; Loprieno 1995, 169; Ritter 1995, 99; Ray 2000, 225 among others. Only Eyre (1991, 113) and Winand (2006, 166) treat iw as modal in its ‘assertive’ role. Zonhoven (1997b, 387) contrasts assertion and presupposition, Assmann (ibid) treats the former as equivalent to ‘predication’ and views adjuncts as ‘non-asserted’. Callender (1983, 93) correctly calls Hm a particle of “enhanced veracity and… relevance” and of ‘assertion’. In Late Egyptian the issues of assertion and illocution have received more attention; the most recent overall study to this effect is Sweeney 2001. 78 Early uses of the term in Egyptology are Erman 1933 § 536 and Till 1933. See also Vernus 1990, 16; Reintges 1997, 56; Hannig 2003, 1 for Earlier Egyptian.
introduction
25
The remainder of this work demonstrates how this system functions in practise and what are the parameters for coding a complement as an assertion or non-assertion in Earlier Egyptian. The point of departure for the discussion are affirmative object complements of main clause predicates, after which both asserted and un-asserted clauses are attested and which best illustrate the key principles of their variation. This is followed by a discussion of transitive verbs that, for reasons explained, systematically disallow realis ntt/wntintroduced object clauses, but which are shown still to form part of the same overall system of assertion and non-assertion covering all complementation in Earlier Egyptian. Subsequent chapters elaborate the image of this organisation by focussing on the little-studied affirmative clausal subjects and negative complement clauses after verbs. The discussion is then extended to preposition complements, for which a modal reanalysis is proposed along the lines established for complementation after governing predicates. Finally, after a discussion of certain less central patterns of complement clauses and predicate complements of nominal sentences, the ‘cross-categorical’ and more abstract conceptual foundations of the grammatical ‘mapping’ of Earlier Egyptian irrealis modality as revealed in complementation are surveyed. The use of the bare complement sDm=f forms is found to share notable parallelism with their temporal properties, particularly aspect. These two apparently disparate domains of use and meaning are shown to be inherently related and many previous theories focussing on the aspectual characteristics of the forms in question are recast into a more global analysis merging their tam- properties into one semantic-pragmatic profile. The analysis presented here is also intended to provide an initiative and methodology for tackling many further problem-areas of Egyptian grammar from a modal perspective. It is suggested at various points, and in the conclusion, that modal irrealis and realis play a role in the grammar of Earlier Egyptian far beyond the relatively restricted domain of complementation. Nevertheless, complement clauses offer an ideal basis for investigating modality at this stage of the linguistic history of Egyptian. These constructions provide sufficient data (largely) free from the burden of various complicating factors pertaining to initial clauses, most notably the character of auxiliaries, the issue of predication and the question of theme-rheme/
26
introduction
topic-comment-oppositions.79 They reveal a system for expressing modal meaning whose basic characteristics are recognisable as language universals, whose expressive potential is remarkable, and whose functional flexibility is deeply impressive. 0.2 Methodology, Terminology, Morphology and Sources Before embarking on the discussion of the system outlined above, certain remarks on methodological and related issues must be made. The approach adopted in the present work has, to an extent, a comparative element. Studying Ancient Egyptian is archaeology of a dead language in which cross-linguistic comparisons provide the only support available for hypotheses on semantic-pragmatics. The imperfectly preserved evidence cannot be understood or analysed in isolation. In most instances, the comparisons will be drawn from languages with no relationship to Ancient Egyptian. Genetic matches demonstrate whether a certain feature exists or does not exist in some language group. However, parallelisms with unrelated languages have the advantage of showing that if a given meaning suspected as grammatically realised in the language under study is actually so treated in languages more far a field, this is likely to reflect some underlying cognitive aspect of human communication in general. Further, the present study entails a conscious rejection of Bloomfieldian views according to which researchers of language have no direct access to ‘meaning’ but only form. It is argued here that systematic application of comparisons from other languages and close attention to the co(n)text in which the expressions studied occur allow substantiated (and not necessarily purely hermeneutic) judgements to be made of even such subtleties as the expression of attitudinal information in extinct languages. As a concrete illustration of this, one may consider the Biblical Hebrew examples in (xi) above. The discussion relies extensively on linguistic data. The reason for this is not only the need to avoid idiosyncrasy in analysis; employing linguistic principles and methodology in the study of Egyptian also renders the language more accessible to non-Egyptologists. 79
All these issues will be touched upon several times in the course of the present work and in some more depth in the conclusion (section 10. 3).
introduction
27
Ancient Egyptian is but seldom quoted in linguistic literature,80 and a significant contributing factor to this state of affairs is arguably the combined effect of the syntactic models with which the language has been described and the sense of overwhelming ‘otherness’ entailed. Postulates such as the various ‘nominal transpositions’ and ‘conversions’ create a false impression of it as somehow fundamentally alien, incomprehensible and best left to Egyptologists. But, if and when it is shown that Earlier Egyptian possessed clear and definable means of expressing modality in complementation that share resemblance with those found in other languages, this should attract the interest of students of language more widely. The same holds also for the special features of Earlier Egyptian modality—in particular the internal division of irrealis, the use of irrealis negations, and the generally ‘upside-down’ character of the system where irrealis is less marked than realis. The relationship with linguistics can perhaps finally be made more reciprocal. Just as methods and data from the latter can be appropriated for the benefit of Egyptology, a solid description in similar terms of Egyptian, the language with the longest documented history, can contribute significantly to the general scientific discussion on language and communication. The analysis presented is not couched in the vocabulary of any particular theory, but shares affinities with Cognitive Grammar, Functional Grammar and the so-called (Radical) Construction Grammar. 81 From the first-mentioned approach is adopted the view of the ideational basis of modality as ‘force dynamics’, i.e. as metaphorical abstraction of physical forces and barriers into the domains of compulsion and obligation in deontic, as well as reasoning in epistemic modality.82 A modal ‘force’ may be someone’s authority 80 And, alas, when this happens, the evidence is often misinterpreted. Two examples will suffice: Sasse, in his discussion of the thetic/categorical divide, refers to the Earlier Egyptian element iw as an ‘existential marker’ used in ‘thetic’ statements (1987, 576 n.19). However, this analysis is clearly misguided. Cristofaro claims that Ancient Egyptian uses only direct speech to convey reported statements “and possibly also reported believes and commands” (2003, 46). As will be seen, this too is incorrect. 81 A succinct summary of the basic tenets of the first two closely related approaches is Horie & Comrie 2000; for an outline of Construction Grammar and its developed form Radical Construction Grammar, see Croft 2001, 14–29. 82 Talmy 1988 is the basic theoretical text, Sweetser 1990, chapter 3 an illuminating practical discussion. The most recent overall treatment is Talmy 2000, 409–549, particularly 440–54.
28
introduction
and status in deontic and some indubitable set of premises and evidence in epistemic modality, which function metaphorically as ‘real’ physical forces affecting material beings. This approach explains well e.g. the intuition that ‘epistemic necessity’ (must) quite literally presupposes ‘compelling evidence’ which ‘forces’ one to conclude something. The ‘forces’ involved are subjected to equally abstract ‘barriers’: e.g. in “he may not go” an authority in deontic, or some ‘mental block’ in the epistemic reading prevents the action or conclusion. From functionalist approaches to language derive the view of structures as non-arbitrary, of meaning as context-dependent and non-atomic, of categories as non-discrete and of grammar in general as non-static and emergent. From there is also adopted the focus on explaining the communicative function of grammatical structure as well as the demarcation between form and function, which is similarly one of the central tenets of Construction Grammar. From the latter derives the view of constructions as the central unit of syntactic description and their decisive role in determining and assigning semantic-pragmatic function to their constituent parts.83 These issues pertain particularly to the role active sDm=f forms in Earlier Egyptian complementation. Indeed, in addition to analysing the grammar of these constructions, the present work seeks to contribute generally to the ongoing discussion of the ‘sDm=f theory’ and propagate a particular sort of ‘minimalist’ approach to formal issues.84 The expression of non-assertion in affirmative complement clauses is primarily based on the use of un-introduced forms of the sDm=f formation. As seen, in such instances the sDm=f of verbs with ult(imae) inf(irmae) roots appear both with and without gemination and some also with additional consonants. Thus e.g. the 3ae inf. verbs iri ‘do, make’ and hAi ‘descend’ are found written respectively as ir, irr, iry, irw, and hA, hAA, hAy and hAw. Doubling verbs such as the 2ae gem. mAA ‘see’ occur both with and without doubling of the last consonant. Anomalous roots display further peculiarities—e.g. iwi ‘go’ occurs both as iw and iwt—whereas immutable ones show no variations.
83
Croft 2001, particularly 4–5, 45–46. For the issue of form and the sDm.n=f, see 7.1 below. The past passive sDm=f will be treated as a independent form due to its combinability (at least in Classical Egyptian) with noun subjects only, which indicates that it no longer forms part of the suffixconjugation. 84
introduction
29
The most widely shared assumption is that, of the various sDm=f forms postulated, employed herein are the geminating ‘nominal sDm=f’ and one or several ‘prospective forms’. The latter do not geminate and appear variously with and without the endings -w and -y in the ult. inf. and caus(ative) classes, show both doubling and no doubling in doubling classes plus diagnostic spellings such as iwt in anomalous roots.85 Opinions diverge on how many ‘prospective’ forms there are as well as whether they exist in Earlier Egyptian throughout or whether some are restricted to the earliest stratum of the language. A widely shared assumption is that originally there were (at least) two ‘prospective’ sDm=f’s, which either largely merged by Middle Egyptian or one suppressed the other;86 total separation after this is suggested but seldom.87 There is further debate on the more general question concerning the extent to which particularly the assumed prospective form(s) represent full paradigms. According to the predominant view, formal distinctions such as the endings -w and -y observable in singular roots can be generalised, and ‘forms’ defined as paradigms of writings of all roots sharing the same syntactic position with the ‘diagnostic’ writings.88 However, this old hypothesis, which underpins most current divisions of sDm=f ‘forms’ in general, is suspicious. It seems fallacious to argue that, say, a writing such as hA=f of the ult. inf. root hAi represents some ‘prospective sDm=f’ only because it occurs e.g. in final clauses with future-modal meaning. This is particularly so if it is simultaneously maintained that hA=f is ‘also’ the writing of some other—in this case ‘circumstantial’—sDm=f form of the same verb hAi in gnomic main clauses after iw or in adjuncts with relative present tense. What can actually be observed in writing suggests that hA=f is the same form in each case, but its semantic-pragmatic and syntactic function is not. Of course, theoretically it is possible that as in Semitic generally, the inflection of the Early Egyptian verb was not based only on consonantal, 85 Depuydt 1993a provides a comprehensive summary of all the views on ‘prospective forms’ in Egyptological language studies. See also Schenkel 2000b on the endings -w and -y in Coffin Texts. 86 Allen 1984 § 266; Schenkel 1981; Loprieno 1991a, 210–17; 1995, 81; Grandet & Mathieu 1997, 206; Malaise & Winand 1999 §§ 445–46; Schenkel 2005, 199, among others. 87 See n.39 above. 88 This principle goes back to Erman 1889a, 11 n.2 and was embraced by ST (Polotsky 1969, 466–67; 1990, 769 and especially 1965, 9).
30
introduction
but also vocalic variation. Native speakers confronted with an unfamiliar text might have utilised syntactic and syntagmatic context to separate forms differing only in vowels—i.e. having recognised the function of some string of written consonants, proceeded to inflect it into a fully vocalised form. However, it is not known whether or to what extent this strategy was actually employed, and its application is in any case forever lost to modern Egyptologists, who have access to Earlier Egyptian only as it survives in the vowel-less writing.89 When all is said and done, the unpleasant reality is that ultimately grammars of this language describe merely the behaviour of the bare consonantal skeletons of words. The extent to which this corresponds to the living, spoken mode of communication is mostly unknown, although here one should be reminded of Junge’s admirably common-sense argument that if a feature is of any grammatical significance, it will not be invisible in writing.90 In what survives of Earlier Egyptian, then, form must equal written morphology, and, as a consequence, researchers are bound primarily to write grammar of function rather than of form. The two cannot be equated, as this will result in postulating ‘forms’ with
89 There are, of course, not a few reconstructed vocalisations of the alleged forms. The sole sufficiently extensive piece of evidence here derives from the Coptic t-causatives, (tako, tanxo etc.) which since Erman 1884 have been used as proof for a ‘subjunctive’ sDm=f in hieroglyphic Egyptian (see Elanskaya 1981, 80–81 for a research history on this topic). In Coptic, the Formenbildung of the t-causatives is regular, which may be taken as indicative of an Earlier Egyptian unitary paradigm of sDm=f in the combination rdi + sDm=f, from which the Coptic lexemes derive (see 3.2 below). In this paradigm the same roots occur written with and without the ending -y but not with the ending -w. From this it may be deduced that forms with -y and -w do not belong to the same formal paradigm and that the y-form may sometimes be written without the ending. In other complement environments, -y and -w-forms occur together along with forms with gemination or no ending, i.e. the paradigms are clearly mixed. Now, one may assume that the last-mentioned are still ‘defectively’ written ‘subjunctives’, as they might be after rdi, but there is no definite proof of this, nor any certainty that the -w cannot also be subject to such omission. If it can, beyond rdi it is impossible to say which form is in question. In any case, there is a deeper problem with all this reasoning. Even if it were the case that after rdi a single paradigm of sDm=f was used in Earlier Egyptian, from this it does not follow that in other complement environments the choice must be only between the ‘subjunctive’ and the -w-form whenever there is no gemination. That is, the supposed fact that after rdi only one kind of sDm=f is used does not indicate that elsewhere in complementation the number of non-geminating sDm=f’s of weak roots cannot be higher than two. This assumption is, once again, based on theory-internal expectations of inherent syntactic properties of the forms involved rather than on phonological or diachronic evidence (cf. e.g. Schenkel 2000b, 71). 90 Junge 1989, 16.
introduction
31
no morphological identity whatsoever. For example, the paradigm of the so-called ‘circumstantial sDm=f’ contains not a single writing that does not also serve some other function than adjunct use. When ‘forms’ become defined as nothing but function, one runs a serious risk of a categorical error. Also the proliferation of ‘identically written forms’, which is an inevitable outcome of such an approach, amounts to a flagrant aberration of Occam’s Razor—one of the most sensible principles in philosophy of science. An alternative approach to the question of form and function has been advocated by Allen, who views the occurrence of differently written sDm=f’s in a given syntactic position as variation of functional counterparts.91 This interpretation accepts that e.g. such writings as hAA=f of the ult. inf. root hAi and spr=f of the immutable root spr are not the same form, but may share the same function. In complement clauses, Allen distinguishes three inflected bare sDm=f forms: a ‘nominal aorist’ active sDm=f signalled by gemination in root-classes showing this property, a ‘prospective’ sDm=f with the ending -w, and a ‘subjunctive’, morphologically distinct only in case of the verbs iwi and ini, which are written iwt/int.92 Other sDm=f’s without these morphological features are considered as functional counterparts of these three types of forms. The division also touches upon modality: the ‘aorist’ describes a “simple occurrence of an action” whereas the ‘subjunctive’ “carries a sense of contingency or necessity that the prospective seems to lack”.93 The present work seeks to build on Allen’s original insight, but treats form as strictly isomorphic of semantic-pragmatic and syntactic function, which relate to use instead.94 The geminating and ‘prospective’ forms of the sDm=f are not understood to exist as paradigms of the Earlier Egyptian verb as a whole.95 What will, for reasons explained in time, be termed proximal irrealis represents a modal function of the sDm=f that is a particular sort of non-assertion and contrasts with another irrealis function of the sDm=f, termed distal. Verbs with immutable 91
Allen 1984 § 212; Allen’s object of research are the Pyramid Texts. Allen 1984 §§ 258–64. 93 Allen 1984 §§ 258, 267; cf. also § 364 and Allen 1982, 22 as well as Malaise and Winand 1999 §§ 593–94. Rather differently in Loprieno 1991a, 215–17; 1995, 81. 94 Comparable views have been expressed by Eyre (1994, 118–19). 95 The same holds also with the ‘circumstantial’ sDm=f, which ought to disappear altogether. 92
32
introduction
roots are interpreted as allocating both these functions to the one and only form of sDm=f in which they occur. Verbs of ult. inf. class possess a specific form for the proximal irrealis, namely the geminating hAA=f/ nTrr=f, but assign the distal function mostly to the non-geminating sDm=f. However, many roots in this as well as in the causative class also possess forms for the latter function: hAw=f, sAxw=f, iry/mry/Hsy=f etc., which do not occur in other functions96 and are not seen as ‘full writings’ of something else any more than hA=f, sAx=f, iri=f etc. are understood to represent ‘Defektivschreibung’. These distal irrealis forms are but rarely used and apparently under diachronic pressure of elimination and morphological collapse, with their function(s) increasingly assigned to their counterparts the non-geminating sDm=f.97 The doubled form of doubling roots such as mAA is associated with proximal function(s), whereas in the classical language the form mA=f of the verb mAA at least seems to have been established as a specific distal irrealis form after the demise of the independent ‘indicative’, which survives only in the bound negative n sDm=f.98 The apparent discrepancy in the form-function divide between ult. inf. (geminating = specific proximal, non-geminating = non-specific distal) and doubling roots (doubling = non-specific proximal, no doubling = specific distal) is in no way surprising. Doubling and gemination are distinct phenomena in that the former is a matter of lexical root morphology whereas the latter is a method of creating semantic-pragmatic oppositions by mutating the root. As a result, the distribution of modal functions in and between the doubling and ult. inf. classes is different. The anomalous roots show very disparate properties, which shows that they do not form any unitary morphological group. The roots rdi ‘give/cause’ and ini ‘bring’ possess separate proximal forms dd=f and inn=f as does the 96 In the Pyramid- and the Coffin Texts, sDm=f written thus occur with future passive sense, but these most likely represent the same form with no fixed voice as yet; cf. chapter 5 n.32 below for a similar situation with the negatival complement. 97 Rather similarly Schenkel 1975, 62; 1981, 517; Loprieno 1986a, 32 n.20, n.21, 38 n.39, end. 98 Although the paradigm of the ‘indicative’ in the different roots is largely identical to that of the -w/-y-less non-geminating sDm=f forms used in complementation, in the historical language the last remaining independent uses of the indicative can be interpreted as involving a separate ‘form’ that is only formed of transitive verbs, accepts only noun subjects and occurs only in absolute initial position; see Doret 1986, 24–27 and 6.3 below. The status of the form mAn=f (of the root mAA) found occasionally in the same environments as mA=f is obscure.
introduction
33
auxiliary wnn, (wnn=f )99 but in complementation rdi and wnn allocate distal functions to the forms rdi/di and wn.100 The root iwi has no separate proximal or distal forms but rather assigns these functions respectively to the forms iw=f 101and iwt=f, the latter of which appears to be not a sDm=f, but a sDmt=f form in a special use.102 The verb ini does the same with int=f, but it also uses the form in=f for distal. The verb ii appears to have a separate proximal form iy=f,103 but allocates distal functions to i(i)=f and iit=f, the latter of which is, just like iwt=f and int=f, a specialised use of the sDmt=f.104 The geminating sDm=f of ult. inf. roots and forms with the endings -y and -w are thus formally distinct moods; other (bare) forms of active sDm=f have, in complementation, irrealis function only. The former two may be grouped as modally marked non-indicative forms; the profile of the latter forms will be defined later.
99 There is no reason why wnn should be left out of consideration here or given some specific treatment just because it functions as an auxiliary. In this its role, wnn is specialised for expressing TAM—and little else—which in fact makes it an indispensable part of a discussion on modality. wnn can hardly be classified as an ordinary 2ae gem. root seeing that the form used after iw is wn, not wnn, which compares ill with e.g. iw(=f) mAA=f. However, wnn is clearly not a ‘geminating’ form either, insofar as this term is understood technically as referring to the reduplication of a medial consonant. Nevertheless, wnn=f is functionally equivalent to the ‘geminating’ form of ult. inf. roots, but in contrast to e.g. the long form of doubling roots, it has this one function in all instances and its writing does not conceal ‘frozen’ ‘prospectives’ (Allen 1982, 25) etc. See 9.1 for discussion. Similarly, wn is simply the ‘non-geminating’ form of this verb, not several identically written ‘converters’ nor an idiosyncratic spelling for wn.n=f (see 9 n.35). 100 The root rdi also has the rare forms rdiw/rdy/dy but these are not attested in complementation. The forms di and rdi cannot be divided functionally. Examples of both are found in most syntactic or semantic-pragmatic functions argued to be a domain of only one or the other, including future cleft sentences, (PT 990c: di=f; 1093b/ M: di=sn) after the verb rdi, (PT 583c/P2: rdi N) in the past negation n, (MMA 35.7.55, 11 n di=i), in final- and adjunct clauses as well as second tenses (cf. EAG § 459). 101 Polotsky (1965, 5) argues that iw=f should be seen as a ‘circumstantial’ form of this root whereas according to Gilula (1985) it should be analysed as the ‘emphatic’ (‘geminating’) form of this root. Seeing that iw=f occurs after ntt (see 2 n.159) but also demonstrably in second tenses, it seems that iw=f covers both these functions. There exists also the rare form iww=f; (Gilula ibid. 137 n.3) the form iwy=f is rare. 102 See 9.3 below for discussion of this point. 103 Polotsky (1965, 5) views iy=f as ‘circumstantial’, Gilula (1985) as ‘emphatic’. Except for Sin R 15, (ti sw iy=f “Now he was returning”) the uses of iy=f rather support Gilula’s view (see 2 n.88, 2 n.106, 4 n.30 and example (196) below). 104 However, a singular instance of a complement form iw of the root ii will be found in example (168) below.
34
introduction
The Earlier Egyptian formal-functional system of expressing modality with un-introduced sDm=f (plus the three sDmt=f forms iwt=f/int=f/iit=f) complement clauses may thus be tabulated as follows (x = non-existent): proximal Root Class Immutable Mutable Doubling Ult. Inf. Anomalous
rdi iwi wnn ini ii
distal
Specific Form x
Counterpart sDm=f
Specific Form x
Counterpart sDm=f
x
mAA=f x x
mA=f hAw/hAy=f x x x x
x
hAA=f dd=f x wnn=f inn=f iy=f
iw=f x x x
iw=f
hA=f rdi/di=f iwt=f wn=f in/int=f i(i)/iit=f
In sum, some roots possess special forms of sDm=f for the proximal and distal irrealis functions, some do not, and instead employ counterpart forms that may also function differently, depending on where they are used. By necessity, the discussion in the present work will mostly focus on mutable roots, which show clear formal differentiation between the distal and proximal irrealis functions. It should be stressed that this organisation is based on the evidence from complementation only. Whether or not the analysis of the forms grouped as marked or unmarked for some modal value is applicable generally to all their uses is another issue. However, as will be seen in the course of the present study and particularly in chapter 9 below, there are good reasons to believe that this is indeed the case. Yet, it goes without saying that verifying this requires an extensive study of the other uses of all the forms listed above. Finally, the present work has a strongly philological and example-based focus. Particularly when dealing with dead languages, elusive modal nuances can only be extracted from a large amount of data rather
introduction
35
than selected examples. The sources utilised span the period between and inclusive of the late V dynasty and the reign of Amenhotep III. Pre-XVIII dynasty examples derive from texts representing all textual genres and registers, whereas later material is largely restricted to royal and private monumental inscriptions.105 Such a wide diachronic scope is required by the relatively infrequent occurrence of complement clauses with revealing morphology and is perhaps less than ideal. As a precautionary measure, therefore, and notwithstanding the unavoidable degree of arbitrariness associated, the Pyramid Texts have been excluded from consideration.106 Yet, the realis-irrealis divide appears to be discernible in these texts as well and Pyramid Texts passages will also be noted at several occasions for comparison against the data studied.107 On the whole, the modal organisation discussed in the present work remains intact throughout the material studied: signs of its collapse appear late and are at first limited in number.108 In view of the many quite thoroughgoing diachronic developments elsewhere in the language between late Old- and early New Kingdom, the modal system of complementation represents one of the most persistent characteristics of the Ancient Egyptian language.
105
Exceptions are the early New Kingdom medical papyri pEdwin Smith, pEbers and pHearst, as well as the Book of the Dead. Examples from the latter derive only from XVIII dynasty manuscripts, and specific preference is given to the versions in the fine papyri of Nu (pBM EA 10477) and Nebseni, (pBM EA 9900) particularly when the latter differ from the version(s) in Naville’s edition. New Kingdom royal Unterweltsbücher are not utilised. 106 Some Pyramid Texts spells are clearly contemporary Old Egyptian, some use language that seems exceedingly archaic. Although the Coffin Texts incorporate excerpts from the Pyramid Texts, they are nevertheless included in the present corpus, seeing that independently their idiom seems closer to ‘standard’ Middle Egyptian than the language of the Pyramid Texts (cf. Polotsky 1969, 466). 107 For a brief discussion of the system outlined and the PT, see 10.2. 108 See 7.3 and 10.2.
introduction to part one
37
PART ONE
MODALITY IN AFFIRMATIVE COMPLEMENT CLAUSES AFTER GOVERNING VERBS
introduction to part one
39
CHAPTER ONE
INTRODUCTION TO PART ONE Of the two types of complementation after governing verbs, object complement clauses are most widely attested in all languages. Also in Earlier Egyptian, where survival and evenness of the evidence are always an issue, a wide variety of matrix verbs with object complements other than the bare infinitive or the sDm=f of immutable roots abound in the textual corpus. Regardless of the particular language in question, in object complements one notices a certain dichotomy between often relatively clear syntax and highly complex semantic-pragmatics. It has been noted that the degree to which the situations described in the main and subordinate clauses are integrated is reflected iconically in their grammar. The complement displays signs of increasing morphosyntactic dependence from the main clause the more the situations they describe are semantically assimilated.1 Yet, the more generally ‘asymmetric’ hypotactic relationship of the clauses—i.e. one between a dependent subordinate- and a dominant main clause—remains the same.2 However, there also exists a related strategy of clausal combination known as direct speech, (oratio recta) which belongs to the same scale of inter-clausal cohesion between transitive verbs and following clauses as complementation, but which is best considered apart from the latter.3 In direct speech, or ‘quoting’, the relationship between the combined clauses is ‘symmetric’ or paratactic: they have equal status and maintain their grammatical independence.4 For instance, in examples (15) and (16) below the direct quotes after the verbs say and answer are self-contained sentences:
1 See e.g. Givón 1980; 1994, 278–79; 2001 vol. 2, chapter 12; Hopper & Traugott 1993, 171; Cristofaro 2003, 117–22. For Egyptian, see Uljas 2003. 2 Halliday 1985, 218; this functional formulation does not presuppose that the complement clause be morphologically marked for subordination or even treated as syntactically embedded, even if this is usually the case (but see Cristofaro 2003, 96–98 for synchronic and diachronic exceptions). 3 Cf. Li 1986, 36–37; Givón 2001 vol. 1, 155–56; Cristofaro 2003, 47, 99 n.4. 4 Halliday 1985, 250–52.
40
chapter one
(15)
The king has been told that a magician has arrived:
Dd.in Hm=f is in n=i sw Then his majesty said: “Go, bring him to me”. (pWestcar 8, 8–9)
(16)
The king has just implied that Sinuhe need not fear, then:
wSb.n=i st m wSb snDw ptr Ddt n=i nb=i I answered to this with an answer of a frightened man: (Sin B 260–61) “What is my lord saying to me?” ‘Direct speech’ need not involve a verb of locution. Humans use language also for thinking, and just as it is possible to quote words spoken, it is also perfectly feasible to quote thoughts: a sentence such as “I thought: ‘I’ll quickly see if he’s in’” is a near-paraphrase for “I said to myself: ‘…’”5 This is possible also in Egyptian: (17)
The sailor is about to encounter the giant snake for the first time:
aHa.n sDm.n=i xrw qri ib.kw wAw pw n wAD-wr Then I heard a tumultuous noise, but only thought: “It is (Sh.S. 56–59) (just) a wave of the sea”. Here the speaker’s thoughts are given a wording, and as in locution proper, this is signalled by the grammatically independent nominal sentence after ‘think’.6 A quote never represents a semantic argument of the governing predicate, but purports to be an authentic reproduction of the words (sounds) or thoughts quoted.7 Also when quoting others, as e.g. in (15) above, the real speaker (here the narrator) lends his voice to the original speaker, (the king) but the quote displays no signs
5
Halliday 1985, 255–56. Thus wAw pw is not an instance of a nominal sentence direct complement of ib, as analysed e.g. in GEG § 186.3 n.7 and Callender 1975, 74. Quoting gives here a more fragmented, dynamic shape to the situation-description, which is well in accord with the texture of the narrative: I only thought: “…”—i.e. in fleeting and without great concern at first. The text does not lose any of its elegance; indeed, the more one is able to recognise different rhetoric devices employed by the ancient authors, the lesser the danger of underestimating their expressive skills. However, in Egyptian parataxis is often evoked simply to explain away complementation violating the ST principles of the use of ‘nominal forms’. See 2 n.167 below. 7 See Haiman 1985, 222–28 for a fine discussion on this issue. 6
introduction to part one
41
of “removal from the reality of the original utterance”.8 In object complementation, by contrast, the deictic elements of the subordinate clause are modified to conform to the perspective of the real speaker— his identity and spatio-temporal locus—regardless of whether or not these are shared with the original speaker, ‘thinker’ etc. For example, in a sentence such as: (xxiii)
Jack said that Jill was going there
the complement stands for the presumed original “Jill is coming here” uttered by the subject ‘Jack’, who is not the real, but the original speaker.9 This is an exemplar of prototypical reported ‘indirect speech’ (oratio obliqua) where a verbum dicendi is followed by a complement clause that does not simply reproduce the original words. For the study of complement modality, direct speech is of limited interest, seeing that, as in example (15) with the imperative, it expresses also these features in exactly the same manner as independent main clauses.10 In object complementation proper, however, assigning a modal status to the subordinate clause is far more complex a matter. Here the semantic-pragmatic dependency of the complement from its governing predicate, the often extreme sensitivity of this to co(n)textual factors, and the role of the real speaker play a decisive role. The effects of these factors must now be briefly illustrated. Object complement modality is inherently linked to the type of semantic-pragmatic dependency of the complement from the main verb. A complement may be ‘meaning-dependent’ of a governing verb with respect to its11 time propositional attitude information value 8
Fleischman 1995, 532; see also Coulmas 1986, 2; Li 1986, 38. The changes in temporal deixis in particular have often been assigned to a set of mechanical rules of ‘sequence of tenses’ (systematised by Comrie 1985, 104–17). But see Palmer 1986, 166–67; Langacker 1991, 253–60 and 9.2 below for the role of tensechanges in modality. 10 But see Palmer 2000, 197–98 and Martin 1998, 205 for some exceptions. 11 Adapted from Noonan 1985, 92. 9
42
chapter one
Complement non-assertion typically results when the time of the subordinate clause is determined by the main verb, the situation it describes is made subject to negative propositional attitude, or presupposed—all characteristic motivations for the use of irrealis/ subjunctive.12 However, this is not merely a matter of main verb taxonomy. In most languages finite complements whose time-reference is predetermined as unrealised vis-à-vis the governing verb are treated as irrealis, and in no language are they asserted.13 Such verbs are e.g. those of commanding, requesting, intending, desiring, preventing etc. which are notionally non-assertive in that their complements are never presented as assertions.14 This is because for the original speaker the complement situation is unrealised and impossible to predict for certain. However, in reports there is potential for the real speaker’s perspective to be taken into account as well.15 In present and future reports, this does not substantially differ from that of the original speaker. For example, in “They order that he go”, it is quite obvious from the sentence that no real situation is referred to in the subordinate clause and whether or not the ordering is or will be successful cannot be known. On the other hand, when it is said “They ordered that he go”, it may actually be the case that at the time of speaking—i.e. in the real rather than the original speech context—the complement situation has been realised. The sentence may e.g. answer the question “Why did Jack go away?” Most languages, including English, Spanish and French, do not take the real speaker’s perspective into account in the modal marking of the object complement here at all. Instead, they assign the modality of complements of notionally non-assertive verbs such as order from the perspective of the original speaker and speech context. Thus, in these languages irrealis is used in the complements of non-assertive verbs also when the former represent past reports:16 (xxiv)
12
Spanish:
Le mandaron que se callara (SUB) “They ordered that he keep quiet”
See 0.1.2 above. In e.g. Bulgarian, this is argued to be the only factor conditioning complement modality (Noonan 1985, 93–94). 14 Cf. Givón 1994, 272; 2001 vol. 1, 309. 15 See Cristofaro 2003, 111–17 for a detailed discussion. 16 Data from Givón 1994, 281 and Judge & Healey 1983, 138 respectively. 13
introduction to part one (xxv)
French:
43
J’ai ordonné qu’il parte (SUB) “I ordered that he go”
However, as can be seen from these examples, the deictic shifts take place as normal. This shows that the overall perspective to the situation is still that of the real speaker, and some languages pay further attention to this vantage point. For example, in Bemba, a Bantu language, in past reported orders different complement patterns are used to indicate whether the situation ordered took place at some point between the time of ordering and the time of speaking:17 (xxvi)
Jack a-à-koonkomeshya Jill a-à-boombele (FIN) “Jack ordered Jill to work” (and she did) Jack a-à-koonkomeshya Jill a-boombe (SUB) “Jack ordered Jill to work” (but maybe she didn’t)
The different complement construals after the same main verb indicate respectively that the ‘working’ took place or that this was not necessarily so. This information is broadly attitudinal seeing that it reflects the subjective relevance attached by the speaker to whether or not the situation was realised. The ‘realisation not indicated’-alternative in (xxvi) is irrealis just as in the more familiar languages. The realised situation is not, but it is not realis either.18 The differentiation is, accordingly, part of the modal system. It does not pertain to the subject of the governing verb, but is added by the real speaker and represents a clear ‘intrusion’ of the latter to the complement situation frame. With verbs that express propositional attitude towards the complement situation/information, or content of communication, the role of the co(n)text and of the real versus original speaker is highly significant across languages. Many such verbs are notionally assertive in that they describe committed attitude and their complements have independent time-reference.19 Verbs of locution, knowledge and perception are obvious representatives of this class. What is said is of course potentially asserted, 17
Adapted from Givòn 1972, 150–51. In other words, the generalisation that the complements of notionally nonassertive verbs are never asserted holds. But see 3.1 for an alleged exception. 19 Hooper 1975, 95. 18
44
chapter one
and knowledge of something equals strongest attainable certainty and commitment to facilitate assertion. Similarly, vision is most objective kind of support for certainty concerning situations in the world, followed closely by auditory evidence. 20 In contrast, some verbs such as doubt describe negative attitude. Expressions of positive and negative propositional attitude are typically associated with asserted and nonasserted complements respectively, but here again various subjective and contextual factors make their effects felt. Although complements of notionally assertive verbs may be assertions, they do not always assert. For example, verbs of belief express commitment and are generally followed by the indicative in Spanish, but the subjunctive may appear if the matter ‘believed’ is still viewed somewhat tentatively:21 (xxvii)
Sospecho que es (IND)/sea (SUB) mentira “I suspect it is/might be a lie”
Similarly, the verb think is followed by the indicative in Spanish and many other languages if it is affirmative, but by the subjunctive, if negated:22 (xxviii) Creo que habla (IND) inglés “I think she speaks English” No creo que hable (SUB) inglés “I don’t think she speaks English”
20
Cf. Sweetser 1990, 37–40; Uljas 2003, 393. Here a note should be made on the issue of ‘factivity’ (Kiparsky & Kiparsky 1971). This term refers to verbs such as ‘see’ and ‘know’ that render their complements ‘necessarily true’. Factivity has been seen as synonymous with presupposition, but this assumption entails confusion between ‘semantic’ and ‘pragmatic’ presupposition (see introduction n.47 above as well as Hooper & Terrell 1974, 485; Hooper 1975, 116–17 and Levinson 1983, 179–81). ‘Factive’ verbs presuppose the logical truth of their complements, but do not in any sense indicate that the information therein is assumed as shared by the speaker and hearer. Nothing prevents from asserting something known and seen, as long as the speaker does not take it for granted that this is already ‘old news’ to the audience. However, there are nevertheless certain instances in which factivity may preclude the latter; see 2.2.2 below for an example of this in Earlier Egyptian. 21 Butt & Benjamin 2000, 258. 22 See Butt & Benjamin 2000, 241; Hooper & Terrell 1974, 486; Hooper 1975, 121 n.21. Also e.g. French, Italian, Portuguese, Russian and Polish behave similarly.
introduction to part one
45
This is because ‘not-think’ is tantamount to doubt, which cancels the assertion in the complement.23 As in main clauses, complement assertion and non-assertion thus often depend of the semantics of the overall co(n)text and of whole sentences rather than individual verbs. The speaker attitude affects the asserted/non-asserted alternation of the complements of ‘assertive’ verbs also when the speaker and the subject of the main clause are, unlike above, not the same. Such sentences potentially contain two assertions; verbs of speaking are the most obvious examples in this respect: (xxix)
Oedipus said that his mother was beautiful → Assertion 1: Oedipus said X Assertion 2: his mother was beautiful
The assertion in the first clause is assigned to the real speaker, whereas the complement represents something said by the referentially different main clause subject. But the deixis in the latter shows that it is anything but faithful to the ‘original’. The real speaker is free to modify and even add information to the complement not conveyed by the original speaker. In (xxix) it may be that Oedipus actually said e.g. “my wife is beautiful”, and it is the real speaker who provides the information that the wife of the original speaker is also his mother.24 In that case the complement represents even less what the original speaker said; it provides a report of this, but it is certainly not his assertion. The real speaker’s ability, willingness etc. to assert or non-assert the complement, and so comment on its content, is decisive to modal status. This is evident e.g. in German reported speech:25 (xxx)
Er sagt, er müsse (SUB) nach Hause/dass er nach Hause muss (IND)
“He says he must go home” The subjunctive in the complement indicates that the speaker does
23 Conversely, the verb doubt is notionally an explicit signal of inability to assert, and e.g. in Portuguese takes subjunctive complements when affirmative, but indicative if negated (Willis 1974, 294). 24 Coulmas 1986, 3–4; this is known as the de dicto/de re -ambiguity. 25 Palmer 2000, 114.
46
chapter one
not necessarily believe or accept the reported proposition, whereas the indicative shows that he does.26 Variation of this sort has little to do directly with the governing verb. The speaker’s role is also revealed in instances where the ‘assertivity’ of say is manipulated, e.g. again through negation. For example, the following German and English utterances are more typical reports of questionings in judiciary systems where suspects are presumed innocent unless proven otherwise: (xxxi)
Er gibt zu, in der Nähe gewesen zu sein, aber er sagt nicht, er wäre (SUB) der Einbrecher “He admits to having been nearby, but he does not say that he would be the burglar”
The non-indicative subjunctive and if-clauses communicate that the speakers are not aware of the true state of affairs described in the complement and cannot assert it. But e.g. in communist East Germany, an interrogator might have reported a failure to extract a confession from an accused whose guilt had already been decided beforehand, as follows: (xxxii)
Er gibt zu, Feindsender empfangen zu haben, aber er sagt nicht, dass er ein Klassenfeind ist (IND) “He does admit to having watched western TV-programs, but he does not say that he is a class-enemy”
Here it is explicitly told that the subject never asserted the complement, but the real speaker is committed to ‘his’ being a ‘class-enemy’ and frames it as an assertion.27 Verbs of cognition and perception abide
26 Cf. Coulmas 1986, 16; Wierzbicka 1988, 150; Palmer 2000, 198–99. The same principle lies behind the main clause use of subjunctive or conditional in French and German ‘journalistic style’ (see 0.1.2 above) where evading responsibility for the ‘truth’ of reported statements is of fundamental importance for reasons of libel. 27 This shows clearly that it is incorrect to argue that the complement assertion is simply assigned to the subject and that the “S(peaker) is not himself asserting the existence of T” (the situation; Langacker 1991, 255; cf. Cristofaro 2003, 36–37). The indicative pertains solely to the real speaker and its presence e.g. here is inexplicable by reference to the original speaker only. Cristofaro (ibid, 37 n.7) argues that the complement clause may be ‘dominant’ in the discourse context, but is never asserted. However, this leaves the variation between indicative and subjunctive mood in coding of reported speech completely unexplained. Cf. also Butt & Benjamin 2000, 254 for subjunctive/
introduction to part one
47
to the same principles based on the knowledge and acceptance of the real speaker. In reported speech proper a reference to the complement as a ‘subject assertion’ may be appropriate when this participant is positively reported to have carried out a speech act. However, otherwise reference to ‘assertion’ is meaningful only with respect to the real speaker. If the complement e.g. merely informs what someone knows or perceives, it need not report anything beyond what the sentence subject can be imagined as having said. The real speaker provides the wording, and again is responsible of the complement modality. For example, in normal circumstances the following English and German sentences are ungrammatical: (xxxiii) ?I don’t see/know that he is here ?Ich sehe/weiss nicht, dass er hier ist (IND) If the speaker says that he does not know something, he cannot phrase the latter as an assertion, which signals that he does know it.28 Yet the English and Spanish sentences below are perfectly acceptable:29 (xxxiv) He does not see/know that/whether he is here Yo no sabía que él estuviera (IND)/estuviese (SUB) ahí “I didn’t know that/whether (s)he was there” In the first English sentence it is innocuous what the subject of the sentence sees or knows. The that-clause shows that the real speaker is in a position to assert the complement because he sees/knows that the situation it describes obtains.30 Whether indicates that this is not so, or indicative expressing the same contrast in Spanish. E.g. Swedish uses conditional vs. indicative and Finnish interrogative vs. indicative to identical effect. 28 Hooper 1975, 119; Wierzbicka 1988, 146. Cf. also Kiparsky & Kiparsky 1971, 349, who, however, explain this phenomenon by reference to ‘factivity’. 29 Cf. Butt & Benjamin 2000, 254. 30 However, unlike with speaking, the real speaker is (almost) bound to belief if reporting someone else’s visual confirmation or knowledge of something; hence the oddity of ?“Jack knows/sees that Jill is there but I don’t believe it”. Nevertheless, “Jack hears that Jill is there, but I don’t believe it” is more acceptable because auditory evidence is less reliable and easier for the real speaker to overrule. This scalar organisation shows that the issue is not that of ‘factivity’. Seeing and hearing are equally ‘factive’, but visual confirmation is more ‘believable’; cf. the different weight assigned to eye-witness testimonies and “hearsay” in court-proceedings, or the ranking of visual evidence higher than auditory in evidential modal systems (see Willett 1988).
48
chapter one
that the speaker does not wish to confirm this. The speaker may withhold the assertion also if it is not his illocutionary intention to provide his audience with the information of whether ‘he’ actually is/was ‘there’, regardless of whether or not he is presently aware of this. In the past tense Spanish sentences in (xxxiv), indicative is acceptable provided the speaker is now in a position to assert—i.e. capable and willing in the current speech-context to commit himself to the ‘reality’ of ‘him’ having been ‘there’. The non-indicative versions again signal a contrary situation: either the speaker is still unaware of the ‘real’ state of affairs, or simply deems it unnecessary to inform/confirm this. In other words, the assertion/non-assertion variation reflects the character of the overall discourse. Finally, there are verbs that render their complements ‘informatively’ dependent by indicating that the latter are presupposed background; a verb of this sort is e.g. regret.31 Not all languages code background information as irrealis. Thus, as noted, e.g. in Russian complements of presupposing verbs such as regret are indicative.32 Further, the number of genuinely presupposing verbs taking clausal object complements is small, and none are attested in Earlier Egyptian. However, speakers may also evaluate when a complement is new information and when not, as in (xxxiv) above. There the use of whether/subjunctive may have been motivated by a perceived lack of relevance of the information about ‘somebody’s being there’. That is, the utterance “He does not see whether he is there” may reflect the speaker’s assumption that the audience is already aware of the true state of affairs or that the complement information is immaterial in the current speech context, perhaps because the most salient issue is that the subject does not see something rather than what he might not see in the present context. Accordingly, whether or not an object complement is an assertion or non-assertion depends on a range of issues including the notional properties of the main verb, the discourse context and the perspective of the real speaker. In every instance the use of realis signals the complement to be believed, accepted and seen as communicatively 31 This is seen e.g. from the survival of the presupposition under negation: I regret/ do not regret that I left » I left. Presupposition defeasibility, as in “Jack won’t have to regret doing a PhD because he failed it” does not affect the choice of mood. 32 Noonan 1985, 99.
introduction to part one
49
relevant by the real speaker to whom it is assertable. Irrealis could be said to signal the speaker’s attitude and ‘epistemic perspective’ to be “I cannot/don’t want to/don’t need to affirm this”.33 In Earlier Egyptian affirmative object complements after governing predicates, ntt/wnt-introduced realis- and bare sDm=f irrealis clauses are used under the same conditions. When asserting the complement is ruled out, or the speaker opts to indicate that for whatever reason he is not fully committed to the proposition, does not accept it or intend to present it as optimally relevant information, ntt/wnt are not used. Rather, a bare sDm=f, which in certain verb-classes can be formally divided into two types of irrealis, appears instead.34 Bare sDm=f irrealis-, and ntt/wnt-introduced realis object complement clauses indicate what is the real speaker’s illocutionary intention and ‘position’ in relation to the subordinate proposition. These subjective stances are not grammatical or lexically determined constants in Earlier Egyptian any more than in other languages.
33 34
low.
Cf. Wierzbicka 1988, 159. See 0.2 above. One single lexical verb stands outside this organisation; see 2.4 be-
50
chapter two
CHAPTER TWO
AFFIRMATIVE OBJECT COMPLEMENTATION AFTER NOTIONALLY ASSERTIVE VERBS For a modal analysis of Earlier Egyptian affirmative object complement clauses and complementation more generally, the notionally assertive verbs of locution, perception and cognition are of decisive importance. Of all governing predicates, these verbs occur with the widest range of object-clause types, including ones introduced by ntt/wnt as well as bare active sDm=f forms of all sorts.1 Combining with any one of these construals is not a prerogative of the said verbs. Yet, examples of the latter are sufficiently common to allow the modal character and parameters of use to be established completely for clauses with ntt/wnt, and initially for those with bare geminating and non-geminating sDm=f and their functional counterparts. Further, with these verbs a general differentiating between ntt/wnt-introduced assertions and un-introduced non-assertions is possible even when the latter are represented by the morphologically unrevealing sDm=f of immutable roots, whose value as evidence with verbs not attested with ntt/wnt is limited. 2.1 Locution: The Verb Dd ‘Say’ and Indirect Speech 2.1.1 Core According to Gardiner, “the highly developed indirect speech found in Latin, where all the pronouns after ‘he said’ or the like are reduced to 3rd pers., hardly exists in Egyptian”.2 This is largely true for Middle Egyptian, but in Old Egyptian, complement clauses after Dd are characteristically introduced by the element wnt and assume the guise of ‘indirect speech’ of traditional Western grammars. Nevertheless, Egyptian evidently gradually gave up this sort of hypotactic linkage of 1 There are also examples of the bare sDm.n=f after these verbs; see 7.1 below for discussion. 2 GEG § 224.
affirmative object complementation
51
clauses and receded into a paratactic system of quoted direct speech after ‘say’. Indirect speech introduced by specific elements largely disappeared by the advent of the Middle Kingdom, occurring only in consciously archaising sources and the Coffin Texts, but not e.g. in the classical literature of the era, where it is replaced by direct quotes. However, as will be seen, the method of expressing indirect speech without an introducing element was not subjected to similar diachronic elimination. Yet, indirect speech complements of this variety differ from those ushered in by additional elements not only in construal, but also in pragmatic function. Proper ‘indirect speech’ was retained in the Earlier Egyptian grammatical oeuvre for a specific modal purpose only. In Old Egyptian wnt alone is used after Dd in the elite tombautobiographies and secular texts generally, ntt being reserved for other verbs attested with introduced object complements. This raises a question of the possible differences between these two elements.3 Notably, wnt appears after verbs other than Dd already in the Old Kingdom and, conversely, examples of Dd + ntt are attested in the PT.4 In the CT and later, ntt seems wholly interchangeable with wnt regardless of the governing predicate. Both ntt and wnt share the same negative equivalent iwt, and although there are no Old Egyptian examples of this element after verbs other than Dd, in the Coffin Texts iwt occurs both after Dd and predicates earlier more closely associated with ntt.5 This progressively increasing overlap of the syntagmatic properties of ntt and wnt from early on is also paralleled by their closely similar etymology.6 wnt appears to be originally a perfective feminine/neuter singular participle from the verb wnn, ‘exist’, whereas ntt is clearly a similarly feminine/neuter form of the relative adjective nty ‘what (is)’.7 Nevertheless, it is first and foremost their identical pragmatic function in complementation that suggests that ntt and wnt represent two near-synonymous elements with essentially the same role in the grammatical system whose differentiation appears to have been slight throughout Earlier Egyptian. 3
See Polotsky 1969, 480–81 and n.146 below. See e.g. PT 1102a; Nt 40–41. 5 See 5.2 below for examples and discussion. 6 For further comments on this issue, see 2.3 below. 7 See Gardiner 1920, 53; GEG §§ 233, 237; LGEC § 701; EAG § 1019; Allen 1986b, 25. Cf. also 10.3 below. 4
52
chapter two
Of all the verba dicendi of Earlier Egyptian, Dd is the only one certainly attested with ntt/wnt and indirect speech.8 Perhaps as a reflection of its diachronic and typological restrictions, the grammatical phenomenon of indirect speech in Earlier Egyptian has received relatively little special attention among Egyptologists.9 Nonetheless, indirect speech with dD + ntt/wnt is rather widely encountered; the following example could be said to represent a locus classicus: (18)
The young king Pepy I writes to his emissary concerning previous correspondence:
Dd.n=k [r] mDAt=k tn wnt in.n=k inw nb aA nfr According to this letter of yours, you have said that you have brought all (sorts of) fine and abundant tribute. (Urk I 128, 10–11)
Here the subject pronoun in the clause after wnt is altered from “I have (or “yours truly has”) brought”, what the recipient probably said ‘originally’, to conform to the real speaker’s perspective. Syntactically, wnt marks the boundary of the subordinate clause and in translation corresponds closely e.g. to English that, German dass, French que, Latin quod or at(t) of Scandinavian languages. However, this simple analysis has not been universally accepted. For example, according to Loprieno, wnt and the main verb here actually form a single ‘thematic N(ominal) P(hrase)’, with “an indefinite object (of the main verb—SU) ‘contained’ in wn.t” and followed by a ‘rhematic NP’ as a “nominalization of a complex adverbial sentence or of a jw-sentence by means of wn” (from wnt):10
8 The verbs smi ‘report’, wHm ‘repeat, relate’ mdw ‘speak’, wSb ‘reply’, Sni ‘curse’, Sdi ‘recite’ and the defective verbs i and xr are attested with direct speech only; cf. Jenni 2002, 239 n.3. 9 A partial exception to this is a study of Old Egyptian ‘quotations’ by Goedicke (1955), where it is argued that in wnt-clauses the object is actually wnt ‘what is real’, followed by an appositional proposition. This interpretation fails to take into account the thoroughly grammaticalised status of wnt; (and ntt) see 2.3 below and cf. Doret 1986, 14 n.13. Significant efforts have, in contrast, been made in study of indirect speech in Late Egyptian (see Peust 1996, with a historical summary on pp. 41–48; cf. also Junge 2001, 220–01; Peust 2005) and in Coptic (e.g. Quecke 1990). Kammerzell & Peust 2002 provides a concise diachronic survey of indirect speech from Old to Late Egyptian. 10 Loprieno 1988a, 69–70. Loprieno’s analysis shares similarities with those by Goedicke 1955, Gunn 1924, 176–77 and Lefebvre (LGEC § 701); see 2.3 below.
affirmative object complementation NP theme
S[[Dd.n=k st r mDAt=k tn]
53
NP rheme
S[wn[iw in.n=k inw nb]]
That is, the entire complex constitutes one large ‘balanced sentence’construction. Yet, there are problems with this analysis. It is not the case that wnt or any part thereof is the ‘object’ of Dd—this is clearly the content of the entire clause, which the element introduces.11 Thus, example (18) is rather an archetypal instance of reported indirect speech. Similar examples with wnt from Old Egyptian and later with ntt are:12 (19)
The king refers to his addressee’s ground plan for a ceremonial court:
sk Tw Dd=k xr Hm(=i) wnt iri.n=k sw r [Aw] mH 1000 [wsx] mH 440[...] xft wDDt n=k m stp-sA Now, you tell my majesty that you have made it to (be) 1000 cubits [in length] and 440 [+x] cubits in [width] according to (Urk I 63, 2–3) what you were ordained in the court. (20)
The king recalls what his addressee said concerning a buildingproject:
[sk Tw Dd=k]13 xr Hm(=i) wnt=k r irt S xft Ddd[t] m stp-sA [...] m xmt=k sk Tw Dd=k xr Hm(=i) wnt=k r [...] [You have said] to my majesty that you are going to construct a garden (?) in accordance with what was said in the court […] in your absence, but you tell my majesty that (Urk I 62, 1–3) you will […]
11 Loprieno’s analysis is prompted by problems encountered by Doret (1986, 81) with this particular example in explaining how the complement clause, which he assumes to be a vedette, can function as such after an ‘emphatic’ Dd.n=k. If there is anything ‘emphasised’ in the sentence, this is surely r mDAt=k tn. The king is not stressing what his addressee has said, but the fact that the latter has put his claims in writing delivered to his majesty (the translation of (18) above follows this interpretation). 12 For the reconstruction of examples (19) and (20), see Brovarski 2001, text figure 2. 13 Edel’s (EAG § 1025) restoration is undoubtedly valid, seeing what follows.
54 (21)
chapter two The young king Pepy I writes to his emissary concerning previous correspondence:
Dd.n=k r mDAt=k tn wnt in.n=k dng ibAw nTr m tA Axtw According to this letter of yours, you have said that you have brought a pygmy of divine dances from the land of the Horizon(Urk I 128, 14–16) dwellers. (22)
The god Orion says to the deceased concerning certain divine beings:
ii.n=k is Ax.ti apr.t(i) rx.n=k rnw=sn Dd.n=k n=sn ntt am.n=k dSrt Verily you have come transfigured and equipped, having learnt their names and told them that you have swallowed the Red (CT V 397m–o) Crown. (23)
The deceased asks a divinity to say to Osiris:
sDm=k mdw nTr pw StA Hr xnty s(my)t iAbtt Ddw wnt=f Hr Ts=i May you hear the word of this god whose face is hidden, one foremost of the eastern desert, who says that he is knitting me (CT VI 328f–g) together. (24)
The author tells about some information conveyed to him by a third party:
[...] Dd.n=f n(=i) wnt rdi.n=k sT.t(i) mm HqAt 10 m nxn He said to me that you have caused ten heqats of emmer seed (pTurin 54002, 8–9) to be sown in Hieraconpolis. In all these instances the original and the real speaker are different individuals. Only one example of reported speech with co-referential main clause subject and speaker is attested, in an explicitly ‘actionexplaining’ context: (25)
King Intef II sings praise to goddess Hathor:
dy(=i) rx=s Dd(=i) r-gs=s ntt w(i) Ha.kw m mAA=s I let her know, I say by her side that I am rejoicing at seeing (MMA 13.182.3, vertical 3–4) her! However, just as the shifts in pronouns show the real speaker’s orientation to be that employed in the deictic profiling of the complement
affirmative object complementation
55
situation-description, it is also the attitudes and opinions of the latter that determine its modal character. It is relatively obvious that the primary locus of information and the main assertion in the sentences above is actually the complement rather than the main clause.14 In most instances, the subject of the sentence is also the addressee, and it seems unlikely that the real speaker’s illocutionary intention is to particularly ‘inform’ the addressee that the latter has said something.15 That is, the discourse-function of these sentences is rather to focus on what was said than that it was said—perhaps in view of further commenting upon them, as is frequently the case in letters whence most of the above instances derive. However, in this capacity they represent expressions of the real speaker’s own degree of commitment to the reported statements and of his high regard of their informative value, i.e. his/her assertions. This is obvious in instances where the original speaker cannot have asserted anything, as e.g. when the governing verb refers to futurity: (26)
Coffin Texts Spell 228 begins with a declaration by the deceased:
i wr aq Dd n iab sS iry-aA n wsir ntt wi ii.kw wr.ki Ax.ki wsr.ki O Great Enterer; tell him who unites writings, the doorkeeper of Osiris, that I have come, being great, transfigured and (CT III 268/69a–270/71b) mighty.16 (27)
King Sebekhotep I says to a divine being:
i.Dd=k n Hr wnt wi Ha.kw m mAa-xrw=f m-bAH psDt17 May you tell Horus that I was delighted of his vindication before (Louvre C10, x + 9–11) the Ennead. 14 See Hooper 1975, 94–95. In fact, if some or all of the initial un-introduced sDm. n=f’s in the above examples are to be interpreted as second tenses, it may be that the governing clauses in fact constitute non-assertions. This depends ultimately on the modal analysis of second tenses; see 10.3 for some preliminary discussion. 15 Examples such as Urk I 104, 12 Dd.t(i) wnt btkw nxt m xAstw pn “It was told that there were strong troublemakers among these hill-dwellers” could be analysed as being construed with an impersonal .t(i) subject and an object wnt-clause. However, particularly in such early examples, the complement should perhaps be rather seen as the subject of a synthetic passive Dd.t(i). See 4.1 below for discussion of this problem. 16 The variant B5C has n ntt. A further similar example is the rather obscure CT VI 340j. In CT V 49c/B4C the ntt-clause also contains the particle is, (see 7.2 and example (353) below) other variants omitting ntt. 17 So Pierret (1874–78, vol. 2, 34) and all subsequent authors citing this example except Helck (1975, no. 10) who has i.Dd=k n Hr Hwntw Ha.kw m mAa-xrw=f m-bAH psDt.
56
chapter two
These examples are, of course, not instances of reported speech; instead, the addressees are told to say something that they have not (yet) said. In the next example the speaker expresses a wish that some wholly unspecified participant would say the complement proposition:18 (28)
A myth narrates the laments of Horus when subjected to attacks by Seth:
Ad=f wi pH=f wi i.in Hr iw Hw A Dd ø n mwt(=i) tw ntt wi snd.ki wrt iw.ki m pf gs ntt wDa Ad=f wi “He rages at me and attacks me”—so said Horus, stranded. “If only someone would tell that mother of mine that I am very afraid and stranded on the yonder side, and that the Outcast (CT VI 408m–q) rages at me”. Whatever the identity of the main clause subject, the complements in these last three examples are patently unasserted by this participant. Yet, the same does not hold for the real speaker who remains wholly committed to the subordinate situations. It matters but little that the governing clause does not confirm an act of speaking or assertion having ‘originally’ occurred at all. After the verb Dd, the elements ntt/wnt can thus be analysed to mark object complement clauses as modally realis speaker assertions. Various alternative but similarly semantic-pragmatically (rather than syntactically) based analyses of the use of ntt/wnt in object complementation have been proposed earlier. Loprieno has interpreted ntt separately from wnt as ‘actualiser’ whose function is to indicate that the following proposition is ‘thetic’ in character.19 ‘Theticity’ refers to a particular strategy of information structuring whereby the described situation and its associated participants are viewed as having equal communicative salience. Thetic sentences do not serve to predicate attributes of a topical subject, but instead ‘present’ entire situations and function as ‘sentence focus-’ rather than ‘argument-’ or 18 Given the future time reference, it seems that the governing clause contains an impersonal subject omitted under relevance (cf. Collier 1990c, 88) rather than that the subordinate clause is a subject complement of a passive sDm=f with future reference. 19 Loprieno 1991a, 214. However, in Loprieno 1995 (100, 109, 200) this and the earlier arguments on wnt (see above) seem to have been abandoned and both are simply referred to as ‘conjunctions’.
affirmative object complementation
57
‘predicate focus’ propositions.20 For example, a proposition “John is singing” may represent a categorical (non-thetic) assertion if it answers the question “What is John doing?” In such a case the reply is about ‘John’, which serves as a topic, and the information about what he is doing figures more salient. If, however, the same sentence answers to “What’s that noise?”, then the utterance ‘presents’ the situation as one un-analysable whole “(It is) [John singing]”, where neither the event or its argument(s) figure larger in the information-chunk as a whole. Thus, e.g. expressions of participant-focus are diagonally opposite to thetic.21 However, ntt/wnt-introduced cleft sentence (subject focus) object clauses occur freely and show that there is no such constraint with these elements.22 It may also be asked what precisely is there in verbs of locution, cognition and perception that allow ‘thetic’ (ntt/wnt-) complements when most other matrix verbs do not. Indeed, thetic or non-thetic status of a given proposition is not determined by lexical or other grammatical factors, but arises from speakers’ shaping of their contribution in response to what sort of information they believe their audience to be seeking. Instead of indicating ‘theticity’ of the complement, Allen has suggested ntt/wnt to signal that clauses introduced by these elements express absolute tense.23 In an example such as (18) above, wnt would show that the tense of the subordinate verb is viewed from the real speaker’s temporal perspective (i.e. the time of the actual speech act) and as “relevant to his own present situation”. This proposal captures superbly the importance of the real speaker’s position, but appears to be somewhat too tempus-oriented.24 Many patterns following ntt/wnt are non-verbal, and with them the question of absolute versus relative timing does not arise.25 For Allen, the lack of an introductory element 20 For what follows, see Sasse 1987; Lambrecht 1994, 137–46. For further discussion, see 8.2 below. 21 Sasse 1987, 572–73. 22 See examples (74) and (76) below. Also ntt-introduced ‘emphatic’ (adjunct focus) second tenses seem to appear after verbs (see example (103) below) and are certainly found after prepositions (section 6.3). 23 Allen 1986a, 11–12. 24 Meltzer (1987, 148–50) raises the same objection against Allen’s thesis, particularly as regards the latter’s discussion of iw, but his criticism is based largely on the assumption of correctness of the ST system and the morphological deductions inherent thereof. 25 See below for examples with nominal and adverbial sentences. Perhaps noticing this problem, Allen suggests that with nominal sentences ntt/wnt may be merely
58
chapter two
signals the tense of the subordinate clause to be relative (taxical) to that of the governing predicate.26 However, a closer scrutiny of unintroduced ntt/wnt-less complements after Dd actually provides further support for the analysis of these elements as marking the real speaker’s stance towards the subordinate proposition. As seen, e.g. in certain negative contexts the real speaker may be denied the capacity to assert the complement, and this affects the grammatical form of the complement.27 In Earlier Egyptian there are no examples of reported speech with a negative Dd + complement clause.28 Fortunately, negation is not the only operation that may alter the speaker’s capability to assert the complement: interrogative has often closely similar effects. If the main verb of locution is questioned, in German the subjunctive is characteristically used in the complement and a whether/if-clause in English:29 (xxxv)
Sagt er, er wäre (SUB) müde? “Does he say if he is tired (or not)?”
Yet, this does not result automatically from the interrogative, but concerns its scope. The if/subjunctive shows that the complement is under the scope of the interrogation as well, i.e. the speaker is not only (or even primarily) interested in whether someone said something, but also in finding out whether that someone is ‘tired’ or not. Here again apply the usual caveats pertaining to different contexts: “Does he say that he is tired?” is quite acceptable if the ‘tiredness’ is known to be the case and all that is enquired is the saying.30 The use of a non-indicative in (xxxv) is thus a signal that the real speaker is actually ‘appositional’, and that with adverbial constructions these elements “could be considered markers of syntactic function only” (1986b, 33). 26 Allen 1986a, 7–8; this is seen to apply generally and not only to complementation. 27 Chapter 1 above. 28 There is one example with a negative hortative forbidding an addressee from saying a certain proposition, and in this unique instance the complement clause is actually not construed with a bare sDm=f, but displays a cleft sentence subordinated by the particle is; see 7.2 below for discussion. Sin B 7 is hardly to be read n Dd=i anx(=i) r-sA=f “I did not say I would live after it” or even as direct speech. anx may here be an infinitive. 29 Palmer 2000, 118, 121; see also Hooper 1975, 98–99. 30 No comparable examples without other irrealis-inducing factors exist in Egyptian with Dd, but they are to be found after rx; see 2.2.1 below.
affirmative object complementation
59
unaware of the status of the complement clause situation as well, and asks for its confirmation or denial. In instances of a similar sort, ntt/wnt is not employed after Dd in Earlier Egyptian: (29)
The ferryman of the dead asks the deceased:
in Dd=k DA=k ir gs iAbti n pt Do you say you will ferry over to the eastern side of the (CT V 103e/T1C) heaven?31 (30)
Part of a dialogue between the deceased and gods:
hA i.n=sn i.n nTrw r N pn in Dd=Tn hAw N pn r wab r-gs iSnw pi nb Htm “Go down!” So they say, namely the gods, to this N (=me). Do you say N (I) should go down to bathe beside this iSnw?32 (CT VI 316p–r)
The non-occurrence of ntt/wnt here is not due to the relative future tense of the complement situation. As is clear from examples (20) above and (51) below, future complements after Dd or in general are not restricted to bare sDm=f forms.33 Similarly, it is neither the case that ‘theticity’ is cancelled in these examples. Instead, more than enquiring whether something is/has been said, the real speakers ask whether the complement situations will occur; i.e. “will you ferry?” and “should I go down?”, with the interrogative scope extended also over the subordinate clause.34 This is particularly clear in example (30) where the matter primarily enquired is surely not whether the addressees say something, seeing that this was just heard by the speaker. Rather similar effects may also occur with conditionals. In the following example the governing predicate and its complement are both part of a conditional protasis describing a mere open possibility; again no ntt/wnt is used:35 31
Similarly in CT V 80b–c/T1C; see n.38 below. Similarly CT VI 317b–c. 33 The inverse holds as well; see below for examples of bare sDm=f with a nonfuture meaning. 34 See (41) and (73) below for instances where this is not the case. 35 A further example is CT V 30e ir Dd=k wnm=i nn “If you say that I must eat this…” However, the interrelation of conditionals and irrealis modality is not straightforward. There is a clear scale between counterfactual conditionals such as “If you 32
60 (31)
chapter two Words of a magician to a serpent:
ir Dd.n=k wd=k a r rxyt axm irt Hr hhyt gr rmT36 If you have said that you thrust (your) hand against mankind, the flaming eye of Horus will go forth, and men will fall silent. (pTurin 54003, rto. 11–12)
In all these instances the real speaker remains in various ways unaware of the actuality or eventual actuality of the complement situations and hence incapable of committing himself to them. Thus the subordinate clauses are not introduced by ntt/wnt, but are coded as irrealis with a bare sDm=f. Rather than the geminating form, non-geminating sDm=f-forms appear in the morphologically revealing examples, and include those with the ending -w.37 That is, lack of gemination seems to correlate with propositions such as these, where assertion is for some reason impossible to the speaker. These considerations allow also a better insight into the following example, which is a textual variant to (29) above: (32)
The yon ferryman says to the deceased:
Dd=k DA=k ir gs (p)f iAbti n pt You should say if you will ferry over to that eastern side of the (CT V 103e/T1Be) heaven. The editor of the CT variant T1Be seems to have understood the ferryman’s words not as a question, but as a hortative. 38
knew/had known X…” that present the protasis situation as definitely false, and sentences such as “If you know that 2+2=4, just say it”, whose ‘conditionality’ actually depends on the context in which the sentence is uttered and which are more likely to read as concessive. Moreover, the speaker’s attitude and the view of the protasis situation as more or less acceptable seem to play its role here as well. For example, “if you know all about this, why don’t you say it?” has the appearance of a ‘concessive’ conditional, but actually signals total disbelief. See Akatsuka 1986 and Sweetser 1990, chapter 5 for discussion of conditionals vis-à-vis discourse and modality. 36 The reading of the lines in the papyrus follows Vernus 1990, 32. 37 The verb DAi shows DA also in the instance quoted in n.31 above. Example (31) writes wd of the verb wdi, which apparently also has a geminating form wdd (Allen 1984 §§ 165, 170). See also example (44) below for the verb iTi ‘take’ and examples (44)–(48) and (50) for the verb ini, ‘bring’. 38 The same variation is observable in CT V 80b–c, where the version T1C has in Dd=k, whereas all the other variants have either Dd=k or i.Dd=k. The latter version appears to be the original in this instance.
affirmative object complementation
61
What differentiates this instance from the kind exemplified by (26)–(27) above is the same difference as that between English “you should say if you will ferry over” and “you should say that you will ferry over”. The first one asks for clarification as to what the addressee is planning to do, whereas the second is a request for the latter to say something, but with the speaker already aware of his addressee’s intentions to carry out the action referred to in the complement. Thus, the absence of ntt/wnt again signals the lack of ‘assertability’ of the complement proposition to the speaker. The examples (29)–(32) do not manifest his knowledge of and commitment to the complement state of affairs, but rather the ‘negative epistemic stances’ of unawareness, doubt and ignorance. Yet, there are a formidable number of examples of affirmative object complements after Dd with seemingly nothing in the grammatical context that would render the subordinate proposition beyond the ‘mental scope’ of the speaker or affect his knowledge and beliefs concerning it, but where no ntt/wnt is employed regardless. Instead, they appear to be straightforward examples of reported speech with all the characteristic modifications in deixis; for example: (33)
The deceased says to a malevolent spirit:
ii.n=i xr=k swA=i arw=k nSnS=i mDAwt=k Hr nn n mrmr Dd.n=k iry=k r=i Hr iSt=i I have come to you that I may break your pens and tear up your books because of this double ill, which you said that you would do to me for the sake of my property. (CT V 66e–h/B2Lb)
(34)
The deceased expresses his faith in the protection by Atum:
nHm=f w(i) m-a Sat Dwt Ddw rmT nTrw Axw mwt iri=sn sw r bA=i He will save me from the evil slaughter, which men, gods, spirits and the dead say that they will do against my soul.39 (CT VI 93d–e)
39 The lack of agreement between Sat and Ddw apparently results from mechanical copying from the identical CT VI 93a where the antecedent is sdb ‘impediment’.
62 (35)
chapter two Sinuhe characterises the ‘Goliath of Retenu’ who challenged him to a duel:
pry pw nn sn-nw=f dr.n=f s(y) r-Dr=f Dd.n=f aHA=f Hna=i He was a champion without equal: he had come to dominate it (his tribe) entirely, and he said he would fight with me. (Sin B 110–11)
Yet, all these examples share one crucial characteristic in common. Whether the question is of some unspecified malice, a murderous act, or an unprovoked fight, the situations described in the subordinate clauses are all most unwelcome to the speaker. Although the complements are reported as certainly having been said and asserted by the main clause subjects, they constitute scenarios to which the real speaker certainly does not want to commit himself. In other words, the issue is not that of speaker incapability, but unwillingness to express commitment and accept an unpalatable complement situation. This sophisticated conveying of information about the commitment and acceptance in Earlier Egyptian is paralleled by the use of mood in modern languages; for example, in Italian and German:40 (xxxvi) I maligni dicono che il nonno N riscuotesse (SUB) fino alla sua morte… una gabella ‘sul coito’ “Wicked people say (i.e. would have us believe) that grandfather N until his death levied a tax ‘on copulation’” (xxxvii) Sie sagen, dass mein Freund ein schrecklicher Mensch sei (SUB)
“They say that my friend is horrible person” Sie sagen, dass mein Freund ein sehr netter Mensch ist (IND)
“They say that my friend a really nice person” The speaker may decide to withhold from asserting the complement if it does not seem acceptable as a proposition, as is most clearly betrayed by 40 The Italian data from Maiden & Robustelli 2000, 326. In German the subjunctive is increasingly common in colloquial language in both sentences and the difference is consequently lost.
affirmative object complementation
63
the pair of German sentences in (xxxvii) above. Exactly the same holds for the Earlier Egyptian examples above. Non-geminating sDm=f’s, including special forms in -y, (judging from the writings iry=k/iri=k in examples (33)–(34)) are again used in these irrealis expressions of strong lack of speaker acceptance of what is said. Non-assertion occurs again, although the original speakers are indeed reported to have said the propositions. Further examples of similar sort after Dd and with the same motivation for the absence of ntt/wnt are the following:41 (36)
A mocking comment made of a defeated serpent-demon:
ia[nw] n=k imy-nhd=f... mk hwt pr m pt m-Xnw TpHt sbi Dd.n=f sbi=f r ra iri=f awA r=f sA r=f in snS wr Woe unto you, imy-nhd=f-serpent… See, burning has gone forth from the sky into the cavern of the rebel, although he had said that he would rebel against Ra and commit robbery against him; now his mouth is guarded by the Great Annihilator. (CT V 244b–245a)
(37)
An Execration Text heaps curses upon enemies specified in the following terms:
sbi.t(y)=sn wA.t(y)=sn aHA.t(y)=sn Ddw aHA=sn Ddw sbi=sn m tA pn r-Dr=f Those who will rebel, those who will conspire, those who will fight, those who say that they will fight, and those who say that (Ächt d 1–6) they will rebel—in all of this land.42 (38)
A boast uttered by a worker in a reaping-scene:
iSst pw TA Dd i.iri=f m tr(=i) Who is the man who can say that he can do (reaping) at my (Ti pl. 123, middle register) pace?43 41 See also CT IV 385e; CT V 324j; CT VI 318g (all iry=k); Tb 90/Nu pl. 22, 4 (iri=k). CT V 247b–c has sbi=f… iri=f awA, CT VI 93a iri=sn and CT VII 111f awAw=sn. CT II 51b has iri=sn but the context is obscure; Ptahh 111 appears to be corrupt. 42 Similarly Ächt h 4–5, l 4–5 and n 4–6. 43 In the bottom register, one has similarly zy pw Dd i.iri=f m tr(=i), “Who is one who can say…” etc. Here the occurrence of the i-prefix is notable (see EAG §§ 449–54; Allen 1984 §§ 11, 141). Primarily a feature of Pyramid Texts, the function of this additional inflectional feature is not altogether clear. However, for the present argument it is
64 (39)
chapter two The author of a letter denies the accusation that he is disregarding his duties because of a certain perceived injustice:
ir pA Dd wrS=i Hr rm(t) Hr=s... Now, as for the saying that I (allegedly) spend my time crying (pUC 32200, 17) about it… (40)
The deceased states that he will not partake in consumption of faeces:
bwt=i pw Hs n swri=i snt=f wsSt nTr nb Dd wnm=i bwt=i wnm=f Hna=i Excrement is my detestation and I will not drink its ‘sister’ urine. Every god who says/has said that I must eat what I detest, he (CT VI 198n–p) will (have to) eat with me. In the first of these examples the motivation for the lack of commitment is apparent, particularly as it is clear from the context that by the time of speaking the malevolent intentions of the original speaker had failed to materialise. The purpose of the text in example (37) is to annul all such intentions described, including their very likelihood, rather than to express them as accepted. In (38) the speaker is so certain of his own excellence in the task at hand that he expects no answer whatsoever, and in (39) the tone of denial is apparent and quite akin to that seen in the Italian example (xxxvi) cited above. In these last two examples the complement clauses do not refer to future, but merely describe general situations towards which little ‘faith’ is expressed. Finally, in the rather unpleasant example (40) the reasons for the lack of speaker commitment hardly need to be spelt out. Also in the following instance the non-acceptance of the complement situation is most apparent, seeing that the latter is immediately denied in the following sentence:
significant that i never co-occurs with gemination in ultimae infirmae class sDm=f, but, on the other hand, does co-occur with non-geminating forms showing the endings -y and -w (EAG § 450; see also 10.2 below). A further similar (i-less) example is Tomb of Merefnebef, pl. 21/66, right, with an exchange between two reapers A i TA Dd iri=f m tr(=i) and B ink pw nt(y) Hna(=i), “Hey, you man who says he can do (reaping) at my pace!”; “That’s me, mate”.
affirmative object complementation (41)
65
A question addressed to the deceased concerning Seth:
in smA.n=f Tw Dd.n ib=f mt=k n mt=k mk Tw irf xpr.t(i) r=f m mnw n smA Did he slay you or did his heart say that you have died? You have not died. See, you have been transformed into an enduring (CT VII 34a–e/T9C) bull against him! The denial also shows that unlike in examples (29)–(30) above, the interrogative scope does not extend over the complement mt=k; i.e. the speaker is most certainly not asking, “Have you died?” Noteworthy is also the seemingly past sense of mt. Although futurity is not wholly excluded either, past interpretation is favoured by the final sentence.44 In the next example, a yet another variation of the ‘ferryman’-theme, the speaker’s less-than-full conviction to the complement proposition is similarly suggested by what follows:45 (42)
The yon ferryman questions the deceased prior to accepting him on board:
nn i.Ddw=k DA=k ir gs pf iAbti n pt in i(w)=k rx.t(i) wAt i.Smt=k Hr=s As for this your saying that you will ferry to that eastern side of the sky; do you (even) know the road on which you will go? (CT V 104j–05b/Sq2Sq)
Finally, in the following famous passage, the speaker’s tone and rhetorical intent is close to sarcastic: (43)
The sailor has just promised to send the snake myrrh and incense:
aHa.n sbt.n=f im=i m nn Dd.n=i m nf m ib=f Dd=f n=i n wr n=k antw xprt nb sntr ink is HqA pwnt antw n=i-im(y) sw Hknw pf Dd.n=k in.t(w)=f bw pw wr n iw pn Then he laughed at me and at these things I had said, which were foolish in his mind, saying to me: 44 Of course, given that the example derives from the CT, n mt=k may also be read as future. The variant T1C inserts n rdi(=i) mt=k n mt after the initial question. This may similarly be interpreted either as a future “I will not let you die” or as past “I did not allow you to die” (cf. n.47 below). The third variant B10C has an additional n mt=k in the same position. 45 Cf. also the rather similar CT V 115a–d.
66
chapter two “You don’t have much myrrh or any sort of incense, whereas I am the ruler of Punt, and myrrh is mine. And that incense which you said would be brought;46 it is the mainstay of this (Sh.S. 149–52) island”.
In every case the speaker does not accept the complement proposition, and no ntt/wnt appears.47 2.1.2 An Example Text The most extensive source of examples displaying the full richness of motivations for lack of speaker commitment after Dd, and also one that contains an example with a bare non-geminating sDm=f in close proximity to a wnt-clause, are Coffin Texts Spells 38–40.48 The text is of considerable complexity and furnishes a case for testing the hypothesis presented above in at least equal measure as it provides further examples. The spells present a remarkable account of a court-case between a deceased father and his son before a tribunal of divinities. The son is about to enter the yon realm himself and eager to usurp his father’s position there. He employs refined rhetoric to persuade the judges of the righteousness of his intentions. The father’s words betray clear reluctance to bow to the pressure and hand over his status. The tribunal plays a passive role throughout and makes no contribution, although the contestants address it as much as they do each other, and arguments, protests and accusations are hurled to and fro. The debate opens with the son addressing the tribunal and exhorting it to confirm that the transition of rights has their blessing and that the gods have specifically summoned him: 46 Here the alternative would be to read int=f as an infinitive “of whose bringing you spoke”, but then one should perhaps expect the preposition r or Hr to appear before int=f. 47 Also the PT provide the following particularly fine examples: PT 944a–b/MN ii.n=f ir=k Dd.n=f smA=f Tw n smA=f Tw Twt smA=k sw “Although he has come against you and said that he would slay you, he will/has not slay/slain you; it is you who will slay him”. PT 1477a–b/P has in ir.n=Tn ir=f Dd.n=Tn mt=f n mt=f “Have you acted against him and said that he has/will die(d)? He has/will not die(d)”. As usual, the timereference of the negations is not entirely clear. PT 481a–b is an earlier version of example (41) above. 48 See Faulkner 1962; Grieshammer 1970; 1975/76 and De Jong 1994 for previous studies on one or all of these spells. The versions primarily followed here are the variants B4L and B12Cb.
affirmative object complementation (44)
67
mTn it=i pf aHaw=i pf mHy=i pf HAy=i pf hA.n=i n=f pf imy imnt imy Xrt-nTr siwy n=f wi m DADAt Dd n=f int wi tp-r=Tn sar.tw hrw=i m tA pn n anxw nt(y) wi im=f Dd n=f iT=i st=f nHm=i saH=f m tA pf Dsr nt(y)=f im=f See that father of mine; that attendant, guardian and champion of mine for whom I have descended; him of the West and of the necropolis! Announce me to him in the tribunal. Tell him that your mandate49 has brought me (now that) my days in the land of the living where I was are over. Tell him that I may take his seat and assume his status in this holy land where he (CT I 158a–159c) is.50
By not phrasing the complements int wi and iT=i as assertions, the son cleverly avoids presenting their propositional content as something to which he himself is committed or wishes to present as a fact.51 This is because he is seeking the tribunal to accept culpability of the looming usurpation by announcing this to the father, as well as responsibility of his being brought to present his claims, which undoubtedly is not the case.52 ntt/wnt appears in neither instance and int seems to have past reference. The son then continues by asking the tribunal whether they have been told why he has been brought before them:53
49 tp-r, ‘Ausspruch’ (Erman & Grapow 1926–31 (henceforth Wb) V, 287; Hannig 2003, 1423) clearly refers here to an authorising pronouncement. 50 Faulkner, (1962, 36; 1973–78 vol. 1, 30; 1979, 2) Grieshammer (1970, 117) and Silverman (1980, 9) understand both instances of Dd as sDm.n=f’s reporting the father having said that his son may usurp his status. This misses the point of the subsequent exchange of arguments quite fundamentally. 51 I.e. the motivation for irrealis is the same as in German or French ‘journalistic style’ (see chapter 1 n.26 above). 52 Compare the latter instance with example (228) below where the complement situation is similarly ‘false’ but the clause is introduced by iwt, the negative equivalent of ntt/wnt. The difference seems to be based precisely on what the speaker wishes to present as his own committed stance. 53 However, the construction in this sentence in fact belongs elsewhere in the discussion. The subordinate clause is not an object- but a subject complement of the passive Dd; see chapter 4.
68 (45)
chapter two in Dd n=Tn in.t(w)=i n=Tn r sAr n=i it=i pf r iwa=i st=f r nHm=i saH=f in Dd=Tn in.tw=i n=Tn (n)54 tp-r=f Have you been told that I am brought to you so that my father might be dispossessed55 in my favour and that I may inherit his seat and assume his status? Will you (rather) say that I am brought to you because of his pronouncement? (CT I 159d–h)
The son’s point is clearly that if the tribunal has been told that his intentions are so negative, this is not to be believed; his words show a denial of that suggestion. The intent of the son’s second question is less clear. He seems to suggest that the tribunal may consider announcing the transfer of rights in the name of the father—without hearing the latter.56 At this juncture the father puts forth a rebuttal whose central claim is rather obscure. Next the son addresses his father and attempts to persuade him to testify that it is his will that the son succeed his father. His intentions are the same as when addressing the tribunal at the start of the debate, and again he uses non-assertion: (46)
in Dd=k in.tw=i r tA pw Dsr {rdi.n.t(w)=f}57 nt(y)=k im=f r st=k imt Xrt-nTr...r pXr rf n=i Axw=k r nHm=i n=i iAwt=k r Dd(=i) rf Axw(=i) mitw=k r=k Will you say that I be/am brought to this holy land where you are to your position in the necropolis…so that58 your might
54 The variants B12Cb and B16C have no n before tp-r=f. However, the damaged B13Cb omits the dative n=Tn but has […] in.t(w)=i n tp-r=f, which must mean “I am brought because of his pronouncement”. It seems thus that in the two first-mentioned versions the n is lost in a haplography with the final n of n=Tn. 55 So Silverman 1980, 9. 56 Faulkner (1962, 36; 1973–78 vol. 1, 30) and Grieshammer (1970, 117) interpret the first question as “have ye said/habt ihr gesagt?” (i.e. as in Dd.n=Tn with a sDm.n=f). As the tribunal is not indicated as saying anything at any point, this is unlikely to be correct. Faulkner’s rendering of the second question as “Do you say that I may bring you his utterance?” (see also Faulkner 1979, 2) makes little sense and ignores the w after int in B12Cb. Grieshammer’s “Habt ihr gesagt, dass ein Ausspruch von ihm mich wegführt zu euch?” (1970, 117) is better, but the tense of Dd=Tn and the translation of in.tw=i (the verb ini is translated thus recurrently) seem incorrect. 57 This clause, due to B16C, appears wholly spurious. 58 Because of the particle rf, Faulkner (1962, 36; 1973–78, vol. 1, 30) translates as imperative “Transfer to me thy power”, “despite the preceding r” (1962, 38 n.o). Gries-
affirmative object complementation
69
may revert to me, that I may take for myself your offices and also that I may say thus: “My power is equal to yours”? (CT I 160g–161b)
The father replies to this by claiming that the usurpation plotted by the son will result in a harmful situation where the father’s enemies will be given a free rein to exult over him and his house will face destruction. The son seems amazed of this attack and launches upon a forceful denial:59 (47)
Ax nTr(y) ir=k m imnt m tA pw Dsr nt(y)=k im=f bA=k n=k Axw=k Hna=k mr.n=k bA=k im=i tp tA in Dd=k in.t(w)=i r=i r tA pw Dsr nt(y)=k im=f r sD rf pr=k r whn rf arrwt=k r wSr rf iwa=k r nHn xftw=k im=k isT wi aA m tA pn Hr DbA nst=k Hr sAq bdSy=k You have divine might for yourself in West, in this holy land where you are; you have your soul and your powers are with you. You wanted your soul in me upon earth. Do you (really) say that I am brought to this holy land where you are only to destroy your house, to break up your gate and despoil your inheritance, so that your enemies will exult over you? Why, I am here in this land for the sake of reasserting your seat and (CT I 162d–163d) for pulling together your weakness!
The son is not asking whether his father said the complement proposition since he himself has just heard it, but rather what is his justification for doing so. He appears genuinely offended by the accusations and shows that he does not accept them at all. There then follows a lengthy section where the son protests his righteousness as an heir, after which he again turns to the tribunal and asks could they not60 acclaim his rights, seeing that he is now part of the community of the blessed dead:
hammer has “damit dann die Verklärten, deinesgleichen, über dich sagen…” (1970, 119). 59 Grieshammer’s translation in particular is notably different here (1970, 120–01). 60 See Silverman 1980, 88–93 on the particle rr as marking the question as rhetorical and expecting an affirmative reply; but cf. Junge 1983, 548.
70 (48)
chapter two in rr Dd=Tn in.t(w)=i r st nt it=i pf... isk rx.n=i qd=Tn mA.n=i Xnw=Tn Should you not say that I be brought to the status of my father…for I know your nature and have seen your abode? (CT I 168d–169a)
Once again, his words show careful avoidance of presenting the complement situation in such a way as to imply that he himself is in a position to assert it. Then, without expecting an answer the son continues with a further denial of his father’s accusations: (49)
Dd n=Tn Ssp=i saH=f nHm=i st=f r nHn xftw=f im=f... Dd n=Tn whn(=i) rf msxnt=f m iw nsrsr... isk ink is irr DbA=i nst=f tpt tA nt (sic) anxw m iw nsrsr imi wS=s m tp tA You have been told61 that I would receive his status and take his seat only that his enemies may exult over him… and you have been told that I would overthrow his abode in the Island of Fire… But I am, rather, one who acts, and I will replace his seat upon the land of the living and on the Island of Fire lest it be destroyed upon earth. (CT I 169d–170f)
Here the son presents a report of his father’s accusations and signals that he is not the least committed to their correctness; quite the contrary. After some further argument there follows the final example from this most extraordinary text, where a bare non-geminating sDm=f and wnt appear closely together after Dd. First the son appears to say grudgingly to his father that at least his right to be brought to be aside his father should be recognised,62 but then reminds him of what he had said earlier: 61 Again the construction involves not object- but subject complementation after a passive Dd. Faulkner (1962, 40; 1973–78, vol. 1, 32) and Grieshammer (1970, 123) have “you have said/Ihr habt gesagt”, which entails an assumption that the tribunal said the negative accusations. It did not—the father did, and the son now seeks to denounce them. 62 This seems to imply that the son actually concedes defeat and pursues the second-best option, which is to be with the father in the hereafter, although not in a superior position.
affirmative object complementation (50)
71
mk Dd.t(w) in.t(w)=i n=k r wnn Hna=k m tA pw Dsr nt(y)=k im=f Dd.t(w) wnt mdw.n=k r int=i Ds=k r DbA nst=k Ssp=i saH=k r nS=i Tw m st=k tw m tA pw Dsr nt(y)=k im=f Look, may it be said63 that I am brought to you to be with you in this holy land where you are, although it is said that you yourself have spoken against bringing me to occupy your throne, as I would assume your status only that I may oust you from this your seat in this holy land where you are. (CT I 174j–175a)
Assuming that the interpretation of the son’s intentions is correct, the complement of the first Dd parallels closely the Spanish use of the subjunctive in the following example:64 (xxxviii) Admito que sea (SUB) inteligente, pero qué pesado! “I admit that he’s intelligent, but what a bore!” In Spanish, the verb admitir ‘admit’ is normally used to assert the complement and the indicative is employed. But this can be overruled if the speaker is ‘forced’ to admit something that he does not really accept, even when this may be something very real and ‘true’. Similar ‘grudging acceptance’ seems to be in question in the first Dd-complement in example (50) above. But in the second instance the son is merely recalling what he is told that his father has said,65 and recognises the fact that his father indeed expressed reluctance to hand over his position, referring to the various excuses cited. It is only here that wnt makes its appearance. Accordingly, the text does not present all its indirect speech without this element. Yet, in all but one instance the speaker is keen to disclaim responsibility of his words and avoid 63 Here, given the date of the text, analysing the complement as an object of the main verb Dd with an indefinite subject .t(w) ‘one’ seems legitimate. See 4.1 below for discussion. Faulkner (1962, 42; 1973–78, vol. 1, 33) has “it is said” and Grieshammer (1970, 127) “es ist gesagt worden”. 64 See Klein 1975, 360, 364 n.8; cf. Bergen 1978, 224. 65 By whom? The reference is less than lucid, but the interpretation offered here appears better than that suggested by Faulkner’s (1962, 42; 1973–78, vol. 1, 33) translation “See, it is said that I am brought to you to be with you… a word to you yourself is indeed spoken to bring me to occupy your seat…” which is dubious both semantically and grammatically. Grieshammer (1970, 127) has “Es ist gesagt worden, dass du selbst gesprochen hast…”
72
chapter two
presenting them as something to which he is committed, or there are factors mitigating his commitment and acceptance of the complement propositions, for which reason they appear as ntt/wnt-less irrealis. Another instance of a wnt-introduced complement and one with a bare sDm=f appearing close to each other after Dd in reported speech occurs in the following example, where the context is unfortunately both damaged and obscure, and the motivation for the variation unclear: (51)
Ra is evoked on behalf of the deceased:
i ra iw Dd.n x[ftw] i[p]f nw wsir N nHm=s[n] HDt aAt tp=k Atfw imw wp=k iw Dd.n=sn wnt=sn r HDt tpw r Xnn r[...] m-bAH iw Dd.n=sn wnt=s[n] r Xnn mAat O Ra, those enemies of Osiris N have said that they will take away the great White Crown on your head and the Atef-crown upon your brow. They have said that they are going to destroy heads and to confuse […] previously. They have said that they (CT VI 277q–278d) are going to disturb Order. Nevertheless, ntt/wnt-introduced and un-introduced complements after Dd do appear side by side, and there seems to be a definable set of pragmatic parameters for their respective uses, which relate directly to the real speaker’s knowledge, commitment and acceptance of the complement state of affairs. Un-introduced non-geminating sDm=f’s do not ‘substitute’ for realis ntt/wnt-complements or vice versa. Their distribution is complementary. The un-introduced sDm=f forms clearly function as ‘echter Modus’ in Earlier Egyptian indirect speech and are used for non-assertion under exactly the same conditions as in other languages.66 It is of some interest that the sole form of indirect speech after Dd that survived after the Old Kingdom and the stage of language represented by the CT is the one without ntt/wnt. Thus, in Classical Middle Egyptian, with its system of quoting speech directly, the Latin-style ‘highly developed indirect speech’ referred to by Gardiner is an indication of irrealis modality and speaker non-assertion of the complement. This same situation pertains also to the other
66
Contra Polotsky 1964, 272.
affirmative object complementation
73
potentially ntt/wnt-combining predicates of Earlier Egyptian, whose complements similarly display a split between asserted/assertable and irrealis complements and with the same grammatical division into introduced and un-introduced object complements respectively. 2.2 Verbs of Cognition and Perception 2.2.1 Seeing is believing and other truisms: realis, irrealis, knowledge and perception Besides locution, in Earlier Egyptian the division between ntt/wntintroduced and bare suffix-conjugation object complement clauses is primarily found after predicates that describe mental states and perception. Unlike with Dd, this feature remains constant throughout the stage of Egyptian language discussed in the present work. The verbs in question are rx ‘know’, mAA ‘see’, siA ‘perceive’ (visually) and sDm ‘hear’. There are certain differences between the said verbs and verbs of locution with respect to the indication of speaker commitment and acceptance. Nevertheless, the principles governing the grammar of their object complement clauses are for the most part the same. The differentiation between introduced and un-introduced complements is, again, reflective of their status as assertions and non-assertions respectively. Although there is a considerable quantitative bias in favour of rx,67 affirmative non-future verbs of cognition and perception with ntt/wnt– introduced object complements are common in Earlier Egyptian: (52)
Iyni-hor says concerning his tomb:
iw rmT nb rxy wnt iri.n(=i) s[w] Everybody knows that I made it.
(EAG § 1022)
67 Further similar examples of rx + ntt/wnt not quoted elsewhere are CT I 140g (wnt=f sr/wnt nts srt); CT IV 84j (ntt iT.n N Hw); HPB III, pl. 6b (wnt sp.n n=f […]); MFA 04.2059, vso. 6 (nt(t)=s[…]); pBerlin 10036, 15 (ntt st SA); Urk IV 429, 5 (nty (sic) wbA Hr=i); Urk IV 592, 1 (ntt sA=f pw); Urk IV 993, 9 (ntt anx=f); Tb 78, 37/Nu pl. 41, 42 (ntt it.n=f). CT VI 382e–f has rx=i ntt wn di=i sA n it, which may mean “I know that I (used to?) give son to a father”, but this sounds rather strange.
74
chapter two
(53)
It is said of certain divine beings:
iri.n=sn rn n N pn m bik nTr(y) sk sn rxy ntt wa im pw They have made the name of this N as a divine falcon, because (CT VI 312d–f) they know that he is the one therein. (54)
Hapdjefa reminds officials that cancellation of funeraryarrangements is unacceptable:
mTn rx.n=Tn ntt ir sr nb nDs nb dd tp n Smw=f r Hwt-nTr n nDm.n n=f xtxt im Look, you are aware that as for any official or any ordinary citizen who gives the top-ration of his harvest to the temple, (Siut I 310) reversal therein is not pleasant to him.68 (55)
The author reminds her dead addressee of the protection promised by another dead person:
i(w)=k rx.t(i) ntt Dd.n idw r sA=f ir wnnt wnt im nn di(=i) nqm=f n nqmt nbt iri m n(=i) mitt irt You know that Idu said of his son: “Whatever there might be in the hereafter, I will not allow him to be afflicted by any mishap”. Please do the same for me! (Haskell Museum 13945, 1–2)
(56)
A reply to an inquiry concerning the addressee’s sister:
mk tw rx[.ti] ntt sy mn.ti Hr wpt [gs]-iAb Look, you know that she is fixed on the listing of [Ges]-Iab. (pUC 32126, fragment ii, 3–4)
(57)
The king is the lord of Egypt, thus:
nn tA nb r ib=s n rdi(.n) n=s Hr=sn rstw n Dar.n st mHtyw rx.ti ntt nxy=s wn mi min qA-a nsw-bity mn-xpr-ra There is no land against her will: southerners do not aspire for such, northerners do not seek such, for they know that her protector exists like Min, one high of arm, the dual king (Buto stela of Thutmosis III, 6–7) Menkheperra.
68
Rather similarly Siut I 280–01.
affirmative object complementation (58)
75
A legend under the cartouche of Thutmosis III
qmA.n=sn m Hwt-aA m nfrw Ha=sn rx ntt iri=f nsyt wAHt xr nHH One whom they (the gods) created in the palace from the splendour of their own flesh, and they know that he exercises (Urk IV 593, 4–5) kingship that lasts beyond eternity. (59)
An instruction concerning the vizier’s conduct in land-disputes:
ir sprty nb n[t]y r Dd mnmn tAS=n xr mAA.t(w) ntt st Hr xtm n sr iry xr=f Sd=f Sdwt n tA DADAt smnmnt st As for any petitioner who shall say: “Our boundaries have been moved”—when it has been seen that69 they (i.e. land-register documents) carry the seal of the relevant official, then he (the vizier) can confiscate the Sdw-lands of that council which caused (Urk IV 1111, 9–13) them to be moved. (60)
The postscript of Tb Spell 148 tells of its discovery by prince Hardjedef:
in.n=f sw mi biAw n nsw xft mAA=f ntt StA pw aA He brought it (away) like a royal marvel when he saw that it (Tb 148, 21–22) was a great mystery. (61)
In the debate between father and son, the son claims the primeval god having been informed of his acceptance to the hereafter:
sDm.n=f m r irrw wnt Ts.n wi it=i pf imy imnt He has heard from the mouth of the rite-performers that my yonder father who is in the West has elevated me. (CT I 167a–b/B12Cb)
In these examples the subordinate clauses report what the main clause subjects know or perceive. However, as seen, they do not amount to assertions from the subject’s part inasmuch as the subject is not necessarily saying, or has not said, anything. The same may also happen with main verbs of locution, but is always the case with other 69 Not ‘whether’ as e.g. in Van den Boorn 1988, 185, 190. The prerequisite for the seizure of lands by the vizierate is that official records, which the local authorities have ignored, support the petitioner’s claim.
76
chapter two
verbs.70 This matter is less of an issue in the numerous instances of rx and the verbs of perception where the speaker indicates his own knowledge or sensory experience in the first person and is the complement ‘original speaker’:71 (62)
The king tells his addressee what he thinks of the latter’s skills as an organiser:
sk Hm Hm(=i) rx wnt Haw nb Hr nfrw=f Verily, my majesty knows that every ship is on an even keel.72 (Urk I 61, 9)
(63)
Queen Hatshepsut explains why she has favoured the temple of Amun:
iw=i rx.kw ntt Axt pw ipt-swt tp tA I know that Karnak is a horizon upon earth. (Urk IV 364, 1–2)
(64)
Ptahmose says his conduct in life was flawless:
iri.n=i m mAat mr nsw rx.kw ntt anx=f im=s I acted in righteousness, as the king would desire, because I knew that he lives of it. (Urk IV 1531, 15) (65)
The king tells why he has chosen Ichernefret to carry out a mission on his behalf:
iw is [hA]b Tw Hm=i r irt nn siA.n [Hm]=i wnt nn ir.t(y)=f(y) st Hr-xw=k Now, my majesty sends you to do this because my majesty has perceived that there is no-one who will (be able to) do it, except you. (Berlin 1204, 8–9)
70 Cf. chapter 1 above; this is particularly clear in instances such as example (52) where the subject is generic. 71 Further similar examples: Red Chapel 188a, 13 (ntt mrt.n=f xpr); Urk IV 835, 16 (ntt Htp=f Hr=s; see example (105) below); Urk IV 1074, 12 (ntt wpwt aSA). Urk IV 346, 3–4 is corrupt. See example (226) below for sDm + iwt, the negative equivalent of ntt/ wnt. 72 Following Wente’s (1990, 19) splendid rendering. For the text, see Brovarski 2001, text figure 1.
affirmative object complementation
77
Yet, it is not necessarily immediately clear whether the complement clauses here, and particularly when the real speaker and the main clause subject differ, are truly asserted (rather than simply reported) by the former participant either. With the verb Dd, this could be shown to be the case by studying examples where the complement was not accepted or believed by the real speaker, even though it was reported as having been said by the original speaker. As seen, in such instances ntt/wnt turned out to be systematically absent, signalling the nonasserted irrealis status of the complement. Here, however, the situation is somewhat different. When occurring in affirmative reporting main clauses, verbs of cognition and perception make a strong claim for the ‘reality’ of their object clauses, regardless of possible differences between the original and the real speaker.73 Consequently, unlike with Dd, instances where the real speaker could ‘dismiss’ the reported complement information as unacceptable or unreliable hardly occur with these verbs. Nevertheless, various modifications of the context that remove the main clause from being a report again show that the real speaker’s knowledge and commitment to the complement situation are key factors in deciding the latter’s grammatical form. As with Dd, the context in which the verbs of cognition and perception appear are often such that it is impossible for the main clause subject to have perceived or been aware of the situation described in the complement clause prior to the uttering of the sentence itself. Yet this in no way affects the employment of ntt/wnt, whose use pertains to the real speaker’s modal stance: (66)
Sinuhe begs understanding for his long hiding from the king:
nb siA siA rx(y)t siA=f m Hm n stp-sA aws wnt bAk im snD Dd st O lord of perception who has insight with plebs; may he perceive in the majesty of the palace l.h.p. that this servant was afraid to say it. (Sin B 214–15)
73
See chapter 1 n.30.
78 (67)
chapter two Horus quotes what Ra told him to say concerning certain deities:
iw=sn Hna=i kA=k Dr.kA=sn Hna=k r rxt stX wnt=sn Hna=k nx=f “They are with me”—so you shall say. Then they will end up with you well before Seth knows that they are with you and (CT II 359c–60a/S2P) complains. (68)
Sabni relates what he did prior to his journey to the hostile hillcountry:
iri.n(=i) [ig]r mDAwt r [r]di[t] rx.t(i) ntt w(i) pr.k(i) r int it(=i) pf I wrote some letters to let it be known that I had gone to bring back my late father. (Urk I 136, 9–11) (69)
The writer informs his superior of a failure in work-assignments:
Dd bAk-im di rx (i)r(y)-mDAt nt(y) m DADAt wnt iqdw n-sprt=f r rwDt r irt wAt HqA n dmi-iw Yours truly is writing; inform the courier who is in the council that, as for the potter, he has not yet arrived at Rewdet to prepare the journey of the chief of Demiy. (Balat-aAyn Asil 3686, 1–3)
In the epistolary expressions akin to those in examples (68)–(69), the complement of rx is most commonly introduced by ntt or by r ntt. The latter is originally a particular idiom of these contexts, but its use spreads also to other object complement contexts in early XVIII dynasty texts:74
74 r ntt will be further discussed in 6.3 and 10.2 below. Further examples of this type with ntt/wnt are Urk I 126, 10 (broken); Urk I 128, 8 (wnt=k hA.t(i); Urk I 137, 12 (broken); Heqanakhte III, rto. 4 (ntt rdi.n(=i); Meketra letter, 5–6 (ntt N iy); Sin B 181 (ntt dbn.n=k); pBrooklyn 35.1446, insertion c, 4 (ntt spr.n N); Urk IV 80, 8–9 (ntt Hm=i aws Ha). Examples with r ntt are Michaelidis Lahun letter, 1 (r ntt hAw=k nb aD wDA); pBerlin 10003, 1 (r ntt xpr prt-spdt); pBerlin 10016, 1 (r ntt wAgi r xpr); pBerlin 10022, 3 (sDm. n=i); pBerlin 10023A, 1 (spr.n N); pBerlin 10033, 1 (r ntt sDm.n=i); pBerlin 10036, 1 (r ntt sDm.n=i); pBerlin 10038B, 1 (r ntt rdi N, a passive sDm=f); Lesestücke 98, 6 (r ntt rdi.n N). See also 6.3 below. An example of r ntt after siA is Urk IV 1381, 4 r ntt TAty pH=f Tni “that the vizier reaches senility”, with an anticipated subject TAty; see 2.4 below.
affirmative object complementation (70)
79
The king tells what he told his followers when he was about to enhance the temple of Amun:
di=i rx=Tn r ntt ib n Hm=i r mnx [...] I will let you know that my majesty’s mind will be fixed (‘to (Urk IV 181, 10–11) build this temple’ or the like) The only participant who is constantly wholly aware of the situations described in all the complement clauses above is the real speaker. In the following example the speaker and the subject of the complement clause are the same individual, but what is actually said is that the speaker now (at the time of writing) knows the complement state of affairs to obtain. Hence the clause is introduced by wnt:75 (71)
In a letter from the king to one of his officials:
[iw mA.n Hm(=i) mDAt=k tn irt.n=k] r ist r rdit rx Hm(=i) wnt in n=k wD n nsw [My majesty has seen this letter which you made] for the palace to inform my majesty that a royal decree has been brought to (Urk I 61, 17–18) you. In the next instance the main clause subject is referentially generic and semi-hypothetical, but the actuality and assertability of the complement situation remain the same for the speaker, and ntt appears: (72)
The king smites his enemies ignorant of his might; however:
mar sp r Tnw rxw ntt sA=f pw mAa Fate is lenient to multitudes that know that he is his (Amun’s) (Urk IV 1293, 1–3) true son. Instances where the real speaker would be rendered not ‘entitled to assert’ the complement due to lack of knowledge about its validity are rare with verbs of cognition and perception. As for interrogatives, with these verbs there are no examples of the type “Do you know whether Jill is there?”, i.e. where the complement situation would 75
Cf. chapter 1 above. The situation is rather complex: the king is writing back to his addressee to inform him that the latter’s previous letter, in which a reception of royal decree is confirmed, has arrived at the palace. The king now knows that his addressee has received his orders and tells him that. The restoration of the beginning follows Urk I 60, 16–17, another letter from the same king to the same recipient.
80
chapter two
also be questioned and fell under the interrogative scope, in Earlier Egyptian. In yes/no-questions involving these verbs, the interrogative scope invariably invests the main clause only:76 (73)
An utterance of a workman to his fellow whilst manufacturing a vessel:
in iw=k Hr mAA ntt n aHa.n pA mnw Do you see that the cup cannot stand up? (74)
(Meir I, pl. 5)
The author asks if his dead correspondent is ignoring something essential in his failure to protect a member of the household:
(i)n-wnn n rx.n=k ntt in tA bAkt irr pr=k m rmT Could it really be that you do not know that it is this maidservant (Cairo Bowl 7–8) who runs your house among the people? In both these examples, the first of which is a fine instance of ntt n, a later isomorphic version of iwt, (the negative equivalent of ntt/wnt77) the matter inquired is whether the addressees see or know something. Yet, the speakers are not at all unaware whether or not this something holds and are not asking anything about it. They themselves clearly do see and know that the situations in the complement clauses obtain, and ntt/wnt mark this. Also in case of negations, in most instances the negative scope invests only the main clause. What is denied in the following examples is that the subjects knew or are allowed to know about the state of affairs described in the complements, whereas it is of course not the case that the real speakers are, or ever were, unaware of their ‘truth’: (75)
The king’s superiority in battles against foreign foes is attributed to higher forces:
ii.n=sn m HHw n rx=sn ntt imn-ra Hr mw=f Although they (the foreigners) came in millions, they did not (Urk IV 1291, 1–3) know that Amun-Ra is his ally. 76 Another example (with some Late Egyptian influence) is Tb 153A/Nu pl. 58, 29– 30: in iw=Tn rx.twny ntt {t}wi rx.kw rn n N “Do you know that I know the name of N?” This sentence is repeated 11 times (spelled r ntt {t}wi rx.kw) in Tb 153B, 2–11/Nu pl. 59, 3–11. For a detailed study of example (74), see Uljas 2004. For further comments on example (73), see 5.2 below. 77 See 5.2 below.
affirmative object complementation (76)
81
The king instructs the high steward concerning the treatment of the peasant:
wnn=k Hr rdit di.tw n=f aqw nn rdit rx=f ntt ntk rdi n=f st You are to keep on making sure that he is given rations, but without letting him know that it is you who has given them to (Peas B1, 114–15) him. Yet, there is one unique example where the real speaker states that at the time of speaking he does not know whether the complement situation holds or not, and, as expected, ntt/wnt is not employed (the version BH3C which provides the first person is badly damaged and presented here in conjunction with the better preserved T1L):78 (77)
The deceased says: T1L: N pn Htm=f n rx N pn hAy N pn i(w)=f Ha BH3C: ink Htm n rx=i [...] iw(=i) Ha.ki
I am the destroyer, but I do not know whether I must descend, (CT VI 260b–d) as I have appeared in glory. Here the real speaker is not in a position to assert the complement—he in fact says this explicitly. The subordinate clause is semantically an indirect question, as is the whether-clause in the English translation, and remains un-introduced. Rather than the geminating sDm=f, in this expression of ignorance about the complement state of affairs use is made of a non-geminating form, (which also here displays the distinctive writing hAy) just as after the verb Dd.79 Somewhat similar 78 In Urk IV 365, 10–11 one reads sAw Dd=Tn m n rx=i sp-sn iri.n=Tw nn Hr m. However, the clause headed by iri.n.Tw is not a complement of n rx=i, but an independent appositional second tense sentence: Take care lest you say: “I don’t know, I don’t know; why has this been done?” 79 Here a mention should be made of non-verbal indirect questions after rx. Strangely, in these instances the complement consists of a bare noun + interrogative adverbial. Examples are of extreme rarity, but the following instances after interrogative and conditional rx may be quoted: (Sin B 126–27) Sinuhe expresses his belief in preordination with a rhetorical question: in iw nTr xm Sat.n=f rx nt pw mi m Is god ignorant of what he has ordained or aware of what the facts are like? (Admonitions 5, 3) The sage quotes the words of a hypocrite: ir […]=i rx.n=i nTr tn kA iry=i n=f If I… knew where god is, I would serve him.
82
chapter two
are instances where some contextual factor cancels the speaker’s ability to assert. The following examples, which similarly represent indirect questions, are particularly instructive: (78)
Neferty describes Egypt in turmoil:
iw ra iwd=f sw (r) rmT wbn=f wn wnwt nn rx.tw xpr mtrt Ra separates himself from mankind; he rises in due time, but one will not know whether noon has occurred.80 (Neferty XI d–e)
(79)
An instruction on determining the effects of an ulcer on the health of a bull from the discharge:
siA=k snb=f Hr iw HsA You will be able to perceive whether he is healthy on the basis (pUC 32036, 1.x+14) of how semi-solids emerge. (80)
An instruction concerning the treatment of a certain type of skullfractures:
srwx=f pw Hmst iri n=f mkAty nty Tbt r rx=k spr=f r xt His treatment is rest (lit. ‘sitting’); provide two brick-supports for him so that you may establish whether he attains critical (pEdwin Smith 2, 7–8) condition.81 Although the complements here are not from verbs with mutable roots, it is nonetheless noteworthy that no ntt/wnt is employed, as is the clear relative past sense of xpr in example (78). In this said example the speaker’s key idea is that in the circumstances described, nobody will know whether or not the noon has arrived. In examples (79)–(80) the texts describe conditions in which it can be ascertained whether some state of affairs is the case and procedures to be carried out to (Note also the use of sDm.n=f in a counterfactual sense in the second example and see 7.1 below). The nt pw in the first example is a crux. Seeing that it occurs also after the verb gmi after which ntt/wnt are not licensed, (2.4 below) nt cannot be an abbreviated form of ntt, but neither can it be a lexical expression ‘what exists/fact’ serving as a predicate of pw since mi m could then hardly be accommodated in the sentence. 80 Somewhat similarly Faulkner in Simpson 1972, 238 and Parkinson 1997, 138; Allen (2000, 376) has “that noon has happened”. 81 Following Wb IV, 102, III; the same expression appears also in pEdwin Smith 2, 10; pEdwin Smith 2, 23; pEdwin Smith 3, 8; pEdwin Smith 3, 15; pEdwin Smith 4, 9; pEdwin Smith 8, 9 and pEdwin Smith 12, 18–19.
affirmative object complementation
83
verify whether some situation results, but in both instances the latter are mere possibilities among various others. A somewhat later text provides an example of this with the distinctive form msy:82 (81)
Title of a birth-prognosis:
ky mAA msy st nn msy=s Another (method) of ascertaining whether a woman will give (Berlin medical papyrus rto. 2, 2) birth or not. In none of the above instances does the ‘speaker’ indicate any commitment or certainty regarding the situation described in the subordinate clauses, which are consequently not introduced by ntt or wnt and whether/if is used in translation. It thus appears that in object complement clauses of cognition and perception verbs, the employment of ntt/wnt follows the same modal parameters as after Dd. These elements are used whenever the conditions for speaker assertion of the complement are fulfilled—primarily when the information therein is vouched for through knowledge and commitment.83 When these prerequisites are not fulfilled for whatever reason, the complement is un-introduced and modally irrealis. In the latter case the morphological evidence is rather weaker with verbs of cognition and perception than with Dd, but weak verbs use non-geminating sDm=f, including forms with the ending -y. Again these complement types do not stand in any mutual substitutionrelationship, and the role of the speaker dominates the grammatical coding. Incorporating the data from examples with Dd, the respective ‘positions’ of the real speaker and the subject of the governing clause are charted in the diagram below. It can be seen that the use or non-use of ntt/wnt reflects the knowledge, acceptance etc. of the real speaker in every instance:
82 As pointed out in WGMT § 204, the relation of the negation nn msy=s and the verb mAA is unclear. 83 As discussed, speaker acceptance plays only a minor role here. For the third parameter involved, relevance, see next section.
84
chapter two
asserted / assertable Main clause speaker +
Complement speaker subject + +
→ ntt/wnt: “You said that X”
-
+
-
→ ntt/wnt: “If only someone said that X”
-
+
-
→ ntt/wnt: “Do you see that X?”
-
-
+(?)
→ no ntt/wnt: “If you have said X”
-
-
-
→ no ntt/wnt: “I don’t know whether X”
+
-
+
→ no ntt/wnt: “He said he would fight with me”
2.2.2 The ‘Evaluative’ Use of Irrealis In the instances above the decision between assertion and non-assertion has been conditioned by the speaker’s possibilities and willingness to express commitment towards the complement proposition. However, languages use modality also as a discourse-organising device to indicate degrees of information relevance. As seen, negligible information-value is one of the basic reasons for non-assertion. Also in Earlier Egyptian complement clauses, irrealis and realis are mobilised to signal differences in the relative salience of the information conveyed as evaluated by the speaker. The principles of this use of modality are apparent in objectclauses after verbs of cognition and perception. Here both geminating and non-geminating sDm=f’s and their functional counterparts occur extensively in opposition to ntt/wnt, and indicate that the complement proposition consists of information that belongs more to the discourse background. To understand the motivations for assigning these values to complements it is necessary to take into consideration the wider context, both the immediate sentence frame and beyond. Purely numerically, of all morphologically revealing object complement clauses after verbs of cognition and perception, those with the geminating sDm=f are most common, and this use of the form has prompted many a syntactic and semantic hypothesis. Of the former, one may mention e.g. the analysis by Loprieno according to which the form after the said verbs signals a particular syntactic conversion
affirmative object complementation
85
of “substantivization of a jw-sentence” having taken place.84 Thus, e.g. in an example such as (82)
The king tells his henchman of the high regard in which the latter is held:
rx.n(=i) Hm mrr w(i) ra Hr rdit=f n(=i) Tw I truly know how Ra loves me because of his giving you to (Urk I 180, 7) me.85 the underlying form of the sentence would be †rx.n(=i) [iw mr wi ra]. However, as with the ‘thetic’ hypothesis of ntt, there does not seem to be any principle restricting the use of such ‘embedded iwsentences’ to the ‘object-slot’ of verbs of perception/cognition only, or, conversely, prevent them from following other predicates.86 The derivative mechanism postulated must similarly be viewed with some suspicion, particularly as it seems to be at work only in complement clauses.87 Furthermore, bare geminating sDm=f object complements after verbs of cognition and perception are often best translated in a manner never applicable to either iw- or ntt/wnt-clauses. Rather than informing that something is known or perceived—as is always the case in ntt/wnt-introduced object complements—often the most suitable rendering of a complement geminating sDm=f after the said verbs corresponds to a description of the character (manner, degree etc.) of the complement situation.88 For example:
84
Loprieno 1991a, 214 and particularly n.55; cf also Gilula 1971, 16. Similarly in Urk I 62, 12 (see Brovarski 2001, text figure 2). 86 See also Uljas 2000, 129. Loprieno analyses the use of the geminating sDm=f after verbs other than locution, cognition and perception to signal a different clausal conversion; see 3.1 below. 87 Indeed, it may be asked should clauses with a bare initial geminating sDm=f, that is, second tenses, be similarly analysed as somehow derived from iw-sentences in turn? 88 Contra Allen 1984 § 231. Further examples after mAA are Urk I 39, 1 (irr.t(i); Urk IV 267, 17 (mss sw N). CT I 169f shows iw=f and pRamesseum I, Bii, 10 the interesting form iy of the anomalous verb ii, ‘come’. There are no certain examples of the geminating sDm=f after siA or sDm in the corpus studied for the present work. A possible instance of the latter is CT VII 32c/T1C, but whether the prr=f is a complement of the preceding sDm is unclear. However, examples occur in the PT; e.g. PT 1775b sDm. n=sn dd N mAat [m st izft] “they have heard how king N places right in the place of injustice”. Further examples after rx not quoted below are Urk I 61, 14 (mrr(=i); Urk I 63, 11 (mrr(=i); Urk IV 1270, 1 ([A]bb=i, restored); Urk IV 1673, 9 (wnn=sn); Urk IV 1676, 12 (xaa N); Urk IV 1830, 10 (Haa=f); Urk IV 1833, 9 (wnn=f). 85
86 (83)
chapter two A caption of a scene depicting a man delivering a cow:
mAA mss Hmt Monitoring how a cow gives birth (84)
(Deir el-Gebrawi II, pl. 28)
A magician suggests a form of entertainment guaranteed to delight the king:
Hwy A wDA Hm=k r S n pr-aA aws apr.n=k bAw m nfrwt nbt nt Xnw aH=k ib n Hm=k r qbb n mAA Xnn=sn Xnt m xd m xnt I propose your majesty proceeded to the lake of the Great House l.h.p., having equipped a barque with all the pretty ladies of your palace; your majesty’s heart will be refreshed through seeing how they row hither and thither. (pWestcar 5, 1–5)
(85)
Ahmose tells of his valour in battles against foreign foes:
nn tnwt m sqr-anxw in.n Hm=f m nxwt=f ist wi m tp n mSa=n mA.n Hm=f qnn=i Countless were the living captives whom his majesty brought from his victories; I was at the helm of our army, and his majesty (Urk IV 9, 14–16) saw how brave I was. (86)
The king tells his correspondent:
iw Hm Hm(=i) rx mrr=k Dd xt nb mrrt Hm(=i) My majesty most certainly knows just how you love to say (Urk I 180, 1) everything my majesty desires. (87)
Irtysen describes his skills as a sculptor, including:
iw(=i) rx.kw... dgg irt n sn-nwt=s I know… how an eye looks at its fellow. (88)
(Louvre C14, 9–10)
Sinuhe relates how he gained his host’s respect through warlike exploits:
Ax.n(=i) m ib=f mr.n=f wi rx.n=f qnn=i89 He came to see me as useful and grew to love me, having learnt (Sin B 106–07) how brave I was. 89 The second n is written after the determinative; the H-version (pHarageh I) has it correctly before. Ashmolean Ostracon has qnn{n}.
affirmative object complementation (89)
87
Ptahhotep advices on successful strategy in debate:
ir gm=k DAisw m At=f m Hwrw n-is mitt=k m Ad ib=k r=f rx.ti Xss=f If you come across a disputant making his case, but one who is a dilettante and no match to you; do not show hostility against him even when you know how feeble he is.90 (Ptahh 74–76) (90)
Instruction concerning the use of certain ingredients in preparing a potion:
ps Hr qd r mnx rx=k pss=sn Hr axx mw iry Hr Sww=sn Boil completely and thoroughly. You may tell how cooked they are on the basis of how water evaporates and how dry they are.91 (pEdwin Smith 21, 17–18) (91)
A further instruction following the previous one:
ir isk st qb rdi.xr.tw ø r anDw r iat st Hr itrw iaw.xr.tw ø r mnx rx.tw iaa=sn Hr dp.tw dpt nA.n mw nty m pA anDw nn dhr nb r=s Now, when they have cooled off, they are to be put into an anDw-jar so that they can be washed in the river. Then (they) have to be washed thoroughly. One may tell how clean they are by tasting the taste of the water in the said anDw-jar: there should be nothing bitter therein. (pEdwin Smith 21, 19–22, 1) (92)
Queen Hatshepsut claims to have abided to Amun’s will at all time:
n mh=i Hr sp n Sat.n=f iw Hmt=i rx.ti ntrr=f I did not neglect a matter of his ordering; my majesty knows just how divine he is. (Urk IV 363, 5–6)
90
So pBM 10509; the version in pPrisse has xft Xss=k. Rather similarly in pEdwin Smith 22, 3, but the subject of pss is omitted. Reintges (1997, 381) analyses the pattern in this and the next example as ‘intensive stems’. 91
88 (93)
chapter two Ramose extols his virtues before Osiris:
iw iri.n=i mAat tp tA iw=i rx.kw Hss=k mAa-ib tm irt spw n Dat I behaved justly upon earth, for I know how you favour righteous-minded who do not commit acts of evil. (Urk IV 1776, 14–16)
However, the apparent suitability of this translation here cannot arise simply from the mere simultaneity of the governing verb and the complement situation or the generality of the latter.92 There are examples after rx where the geminating sDm=f is neither generic nor simultaneous with the main verb,93 and, conversely, ntt/wntclauses may also describe simultaneous or generic instantiations. The apparent suitability of translating ‘how’ here formed the cornerstone of Callender’s thesis of all ‘nominal forms’ as actual ‘manner nominalisations’ specialised to indicate “the way an action is performed, or the circumstances under which it occurs”.94 Callender’s theory has not received wide acceptance among Egyptologists, and for good reasons.95 The ‘manner’ sense does not suit non-geminating sDm=f complements and its applicability even with the geminating form is often a matter of degree.96 Thus the sense is clearly not an 92 Nevertheless, there is no doubt that this is indeed often the best way of rendering geminating sDm=f complements of the above sort. For example, translating example (84) as “that they are rowing” would seem to miss an instance of Ancient Egyptian humour. Certainly the main attraction for the male king is watching how the semi-naked girls row, rather than ‘that’ they do it! 93 See example (100) below. 94 Callender 1975, 47–48, 72–77. This idea is anticipated in Assmann 1974, 63 and Schenkel 1975, 56. See further Callender 1977, 306 n.3; Depuydt 1983, 36–37. 95 Satzinger (1987, 620) dubs Callender’s thesis ‘unfortunate’ but against his later claim that the sense arises from the ‘nominal form’ being embedded in an ‘Adverbialsatz-Konstruktion’ (1993, 205 n.27) one may refer to the arguments concerning Loprieno’s theory discussed above. Junge (1978b, 32–33) appeals similarly to the role of the geminating sDm=f in a ‘subject position’ of an adverbial sentence, but his statement that the concept ‘manner nominalisation’ alone does little to distinguish the patterns used after verbs if applied to all forms appearing there without taking into consideration time-reference etc. is sound. The same holds also for his remark that the term ‘manner nominalisation’ suggests a curious hybrid between syntax and semantics whose precise nature is difficult to grasp. For example, Callender speaks of what he calls ‘prospective’ as a “nominalization of the optative mood” (1986, 10). This begs the question is ‘optative mood’ a ‘manner’ and how can it be ‘nominalised’. Polotsky (1984, 119) notes the impossibility of ‘manner’ reading in second tenses where the vedette is not a manner adverbial. 96 See Uljas 2003, 396 n.52.
affirmative object complementation
89
inherent property of any of the bare suffix-conjugation forms found in complement environments. Callender’s hypothesis confuses what is actually a mere translation strategy of one form in one syntagmatic environment with its wider functional profile.97 Yet, the said translation here is nevertheless a manifestation of the more general character of the geminating sDm=f, and more particularly of its value as expressive of a standard type of irrealis modality.98 Several commentators have noted that the difference between the geminating sDm=f ‘how’- and the ntt/wnt-introduced ‘that’-translated complements is that communicatively it is not the situation per se that is at stake in the former case, but merely its internal character.99 This notion not only provides a fine characterisation of the pragmatic contour of these complements. It also reveals them as examples of the phenomenon of using irrealis modality to indicate that the information presented is considered as background and low in relevance. In this instance, this information is the actuality of the situation predicative nexus. From a discourse-perspective, nothing hinges on the reality of the complement situation in e.g. “his majesty saw how brave I was”, (example (85) above) which is uncontested information taken for granted and already assumed in the context. That is, the speaker’s interest centres rather on making a judgement and assessment on the degree and character of the bravery, and in this [I was brave] represents merely a necessary point of departure. Any subsequent discussion would centre on how brave the speaker truly was, rather than on the acceptability and reliability of his claim of actually having displayed such qualities. This contrasts sharply with e.g. “his majesty saw that I was brave” where the issue presented for attention and comment is precisely whether [I was brave] is true or not. The information communicated by bare non-geminating sDm=f complement propositions above thus approaches presupposed, which occasionally is very nearly at issue, e.g. due to the nature of the preceding co(n)text: 97 However, this does not imply that the same strategy cannot be used to translate the geminating sDm=f elsewhere: as noted by Polotsky (1984, 119) and Borghouts, (1985, 35) it is often quite suitable after prepositions with a ‘correlative’ sense—see 6.2 below. 98 The hypothesis presented here appears in an embryonic form in Uljas 2000 and 2003, 396–97. 99 See Borghouts 1985, 37, particularly n.34; similarly Reintges 1997, 107 and already EAG § 494–95.
90 (94)
chapter two From a dialogue between the deceased and a god:
wDa n(=i) sbA pry(=i) im=f di mA=i prr=k di mA=i aq[=k] D: A gate has been opened for me that I may exit through it. G: Let me see you exit; let me see you enter. (CT VII 228k–l) In this and the examples above, the geminating sDm=f thus signals that the actuality of the complement situation, or the predicative nexus between the verb and its subject, is not intended as the communicative focus of what is said. Here, then, we have the other basic cross-linguistic motivation for irrealis modality besides lack of commitment in complement clauses after verbs: the relative lack of optimal discourse relevance of the information presented.100 This sort of ‘evaluative’ use of irrealis/ non-assertion is a wide phenomenon both typologically and crosslinguistically.101 As seen, “speakers do not have to assert assertable information”102 if e.g. they disagree with it, but also if they assess it to be of limited interest and usefulness to the hearer. In many languages irrealis can be used when there is nothing ‘objectively’ barring assertion and the ‘unwillingness’ to do so does not relate to rejection or nonacceptance of the information but rather to speaker assessment of it as somehow lying “at the bottom of the scale of ‘relevance’”.103 In Earlier Egyptian the geminating sDm=f after verbs of perception and cognition is similarly a mode of refraining from asserting for this reason. The unquestioned ‘reality’ of the situation in such instances is less than optimally relevant background. For such information languages typically employ “forms associated with a lower degree of assertiveness, and even forms designated as irrealis”.104 Also the non-geminating sDm=f forms are mobilised for this general 100 The word ‘optimal’ signals that in communicative terms the information is located, as it were, somewhere on the continuum between maximal relevance and total irrelevance; see Lunn 1989a, 251. 101 See Wallace 1982, 209; Lavandera 1983, 232–33; Lunn 1989a, passim and 0.1.2 above. 102 Lunn 1989b, 695 (emphasis in the original); see chapter 1 and Lunn 1995, passim. 103 Lavandera 1983, 231, referring to Spanish subjunctive. 104 Hopper 1981, 216 (italics by SU).
affirmative object complementation
91
function, and as such play a very significant role in object complements after many notionally non-assertive verbs.105 However, after verbs of cognition and perception, mutable forms occur only in ‘future-inpast’ clauses after rx, and nearly all these examples show the same complement verb iri:106 (95)
Wepwawetaa says he never failed to satisfy royal expectations:
xd.n=i xnt.n=i m Xnw rx.n=i iri=i HAw Hr mdwwt Whenever I travelled north or south from the residence, I knew I would exceed what had been asked. (Lesestücke 72, 20–21) (96)
Amun expresses his satisfaction with Queen Hatshepsut’s performance as a ruler:
ist Tn m imt sS=T rx.n(=i) iri=T n(=i) mnw When you were still in your nest, I knew that you would make monuments for me. (Deir el Bahri pl. 131, right) (97)
Having poured out her woes, the author of a Letter to Dead turns to her addressee:
srs it=k iy r bHzti Tz Tw wn Tw r=f i(w)=k rX.t(i) ii(=i) n=k aAy Hr wDa-mdw Hna bHzti AAi sA a-nanxi Awaken your father Iy against Behezti; rouse yourself and hurry against him! You must have known that I would come to you here for the sake of litigating against Behezti and Aai’s son Anankhi. (Cairo Linen 9–10) Here the speakers report (and perhaps make assessments) of their own or someone else’s past commitment to some then future situation. The complements are future vis-à-vis the main clause state of affairs 105
See 3.2 below. Further similar examples are White Chapel 170 (rx.k(i) iri=k); Urk IV 1293, 7 (rx.n=f it=i); Urk IV 1425, 13; (rx.n=f iri=i); Urk IV 1468, 1 (rx.n=f iri=f) and Urk IV 1579, 13 (rx.n=f iri=f). See also example (106) below. In example (97) the reading of the penultimate word as ii(=i) (cf. Gunn 1930, 149) seems better that analysing it as a participle “the one who comes to you” (Gardiner & Sethe 1928, chapter 1, 1). If the former reading is adopted, the form contrasts with writings such as the iy noted in n.88 above and showing two i-reed-leaves. In this example also, a translation as epistemic necessity “you must have known” seems to suit the overall sense particularly well. 106
92
chapter two
and there is no temporal overlap between the two. This will be seen to be a characteristic temporal feature of non-geminating sDm=f complements, but it is not the sole motive for the choice of form, nor for the absence of ntt/wnt, and thus for non-assertion. This is rather the lack of discourse relevance of the complement propositions in the current speech situation. The latter does not equal the time of the main verb, but again refers to the same point of perspective from which all modality is assigned, namely the actual time of speaking, i.e. the ‘here and now’ of the real speaker. Due to the inherent factivity of the verb know, future-in-past complement situations of this verb must have occurred, and are known to have occurred, prior to the time of reporting.107 That is, e.g. “I knew that you would make monuments” entails “you made monuments”, which is a non-issue and already clear by default to all participants involved in the discourse context in which the report is made.108 Thus, in examples (95)–(97) above the subordinate clauses are future relative to the main verb, but in the current speech context they also contain no newsworthy information whatsoever. For this reason they remain non-asserted and non-geminating sDm=f forms appear. The same motive for the use of irrealis as in examples (95)–(97) above pertains also to the following instances of negated perceptionverbs with immutable, but, notably, ntt/wnt-less complements: (98)
Amenemhat I reflects posthumously on his assassination to his son:
mk stAw xpr iw=i m-xmt=k n-sDmt Snyt swD(A)=i n=k See, the mishap took place when I was without you and when the entourage had not yet heard that I would hand over to you (the kingship). (Amenemhat VIIIa–b) 107
See chapter 1 n.20 and n.30 above for ‘factivity’. This is clear from the impossibility of cancelling this entailment: †“I knew that I would do it but I didn’t”. The same does not, of course, hold with verbs such as e.g. ‘order’ or ‘say’: “I said that I would do it, but I didn’t” and “He ordered me to do it, but I didn’t” are perfectly possible, since ‘say’ and ‘order’ do not in any way entail that the thing said or ordered was realised somewhere between the time of saying/ordering and the time of reporting. Similarly, the slight strangeness of “?I knew that I would do it and I did it” results from the obviousness of [I did it]. Additional confirmation of the latter is communicatively redundant and it could be said that there is never any need to further verify the realisation of the situation in the current context of discourse. As will be seen in chapter 3 below, this constitutes one of the key uses of bare non-geminating sDm=f complements. 108
affirmative object complementation (99)
93
The deceased says that his rebirth is a mystery even to the divine:
mA.n wi nw xpr.k(w) n rx=f bw xpr.n=i im n mA=f xpr=i m Hr=f Nu has seen me after I came into existence, but he does not know where I came to be because he did not see with his own eyes that I had come into existence.109 (CT I 334/335a–c) Due to the negations, in these examples there is no similar ‘automatic’ entailment involved as in (95)–(97) above. But again the speakers need not phrase the complement situation descriptions as optimally ‘informative’ to their audience if the latter can be assumed to be already quite familiar with them. This can also depend on the context. In example (98), at the time of speaking it is of course well known to the addressee (Senwosret I) that the transition of government took place, seeing that he himself was the person upon whom it was bestowed.110 In example (99) the complement xpr=i is past in relation to mAA, with the speaker reporting that the matrix verb subject did not personally witness the ‘coming into existence’. Again, this shows most clearly that relative futurity time-reference is not the sole factor conditioning the choice of complement construal. Yet, that the event described took place is obvious from the context to everyone involved: the speaker, the subject of mAA and the audience. There is thus no need to assert and, consequently, no ntt/wnt introducing the complement. For the present analysis, examples (95)–(97) above with nongeminating sDm=f forms are of some importance. In these examples the ‘irrelevance’ of the complement information arises from a ‘logical’ necessity and is not based on the decision and assessment of the speaker, as is the case with the geminating sDm=f. The situations described are, of course, currently both accepted and realised— or better, completed.111 However, exactly because of its ‘automatic’ 109 So the variants S1C, M3C, M5C, M20C, M28C and M-Ann. B1Bo and B3C have n mA.n=f xpr=i/N tn whereas T3C, G1T, A1C and BH2C have understood xprw ‘forms’. M4C, B1C and B2C have xpr.n=i. See 7.1 below for further discussion of this example. 110 An alternative translation could be “whether I would hand over”, which leaves these matters open, but can be used in similar circumstances when there is no need to assert. 111 This issue and its significance to the modal profile of the non-geminating sDm=f forms will be discussed in 9.2 below.
94
chapter two
obviousness, as information their ‘reality’ is also maximally lacking in discourse relevance. Accordingly, the use of non-geminating forms in (95)–(97) does not represent a violation of their functional profile proposed earlier, quite the contrary. It shows that just as complements with bare forms of this sort are associated with the lower degree of acceptance and commitment of the situation described vis-à-vis the bare geminating sDm=f, the mutual relationship of these two types of Earlier Egyptian complement non-assertion is identical also in what pertains to the relevance of the information conveyed. The role of nongeminating sDm=f complements as descriptive of maximally ‘obvious’ information is relatively marginal after notionally assertive verbs. However, it will be seen to be a major feature of their use elsewhere, particularly in clauses after the verb rdi, ‘give, cause’.112 In sum, all the ntt/wnt-less complements after verbs of cognition and perception discussed in this section represent information and “describe ‘states of affairs’ that could easily be denied or affirmed, but is instead left unasserted”113 due to lack of need to assert. They do not ‘substitute’ for ntt/wnt-clauses, nor does e.g. the geminating sDm=f simply describe ‘indicative’ or ‘objective fact’.114 Yet, as seen, the speakers’ decisions as to how much and what sort of information they think the audience requires are mostly based on highly subjective criteria.115 The marking of complement propositions as realis or irrealis provides merely a general sign “about the presence or absence of ‘assertiveness’… whereby the hearer is instructed on how to interpret the content expressed in a message”.116 For this to succeed “it is necessary to know quite a lot about the discourse context”, as “the sentences themselves do not
provide enough context to explain these choices”.117 As an illustration of this complexity, one may consider the following excerpt from the CT, which also furnishes a rare example of wnt after a verb other than one of locution, cognition or perception:118 112
See 3.2 below. Lavandera 1983, 211. 114 Contra Gilula 1971, 16; Allen 1984 § 364; Doret 1986, 23, 40, 41, 49; Malaise & Winand 1999 § 908. 115 Cf. e.g. 0.1.2 above for the Spanish subjunctive. 116 Lavandera 1983, 232. 117 Lunn 1989a, 250. 118 See Griffiths 1960, 52–53; Faulkner 1968; Polotsky 1969, 481; Gilula 1971 and Allen 1986b, 25. The interpretation here is rather similar to that by Griffiths, Gilula and 113
affirmative object complementation
95
(100) Isis has claimed that her unborn child is ‘Osiris’ seed’ and divine; a dialogue ensues:
ihi i.n ra-tm sA ib=T Hmt i.rx Tn rf mi iSst nTr is pw nb iwa psDt... ink Ast Axt Spst r nTrw iw nTr m-Xnw Xt=i tn mtwt wsir pw Dd.in ra-tm iwr=T sdx=T Hnwt pw mss=T iwr=T r nTrw wnt mtwt wsir is pw imi iw rqw pw smA it=f sD=f sw(H)t m-Xnw nxn=s “Well”, said Ra-Atum; “Be prudent, woman! How do you know if he is a god, lord and heir of the Ennead?” “I am Isis, more spiritual and noble than (other) gods; a god is inside this my womb, and he is Osiris’ seed”. Then said Ra-Atum: “Since you are pregnant, young lady, you should conceal from the gods that you are pregnant and about to give birth, and that he is Osiris’ seed, lest that enemy who would even slay his father come and break the egg in its early (CT II 215b–217g) stage”. Here the variation between the introduced and un-introduced complements is again based on commitment and the evaluation of information salience. Ra-Atum asks how Isis can know that her unborn child is of divine offspring, and she answers this.119 He then suggests that she conceal her pregnancy and delivery, and chooses this incontestable state of affairs as his starting-point “since you are pregnant”. When this presupposed situation, and Isis’ imminent delivery occur again in the complement, they appear un-introduced/as geminating sDm=f for the same reason.120 It is noteworthy that the situation mss=T ‘you give birth’ is not ‘generic’ or ‘simultaneous with the main verb’, but lies still in the (apparently immediate) future. The geminating sDm=f seems to have been chosen instead of a perhaps more ‘expected’ nongeminating one due to the different degrees of information relevance Polotsky. Faulkner (ibid, 40 & 42 n.15) translates wnt as “seeing that(?)”, viewing it as being used “after the manner of n ntt”. 119 Gilula (1971, 17 n.14) assumes the first utterance to be Ra-Atum questioning other gods “on Isis’ side.” This seems unlikely; more probably, and as assumed by both Faulkner (1968, 42 n.11) and Griffiths, (1960, 53 n.1) the question is addressed to Isis, although the pronoun Tn is strange. 120 The presupposition does not arise from the verb ‘conceal’; matters ‘concealed’ can be asserted.
96
chapter two
expressed by these two alternatives. That is, modality appears to have guided the complement selection over considerations of tempus. But the divinity of the unborn child is not similarly taken to be obvious and accepted by everyone involved—in fact, it was initially subject to doubt by the first speaker (Ra-Atum). Notably, wnt does not appear in “how do you know if he is a god” when the speaker was not yet committed to the complement proposition, but only when he has accepted it in “you should conceal… that he is Osiris’ seed” (=god).121 Both these complements also involve nominal sentences, which shows clearly that the use of ntt/wnt does not depend on the following construction, but on what is or is not acceptable as, and worthy of assertion. Deducing the speaker’s motives for the complement form choices and following his intentions requires meticulous attention to the context from the audience. Yet, this is an inseparable part of interpreting modality; modal expressions are not self-contained ‘meaning-chunks’ to be studied in isolation. The ‘evaluative’ use of irrealis in Earlier Egyptian is the context-sensitive class of employment par excellence and ultimately serves the same discourse organising and information structuring function as in all other languages. There are also other such strategies available in complementation. In the following example the geminating sDm=f appears in variance with ‘object-raising’, where the complement clause actor is both syntactically and pragmatically ‘foregrounded’ as the primary object and first-order focus of interest, followed by an oblique object describing the action in which it is engaged:122
121
Cf. Griffiths 1960, 53, who notes “Re-Atum’s doubt appears to be removed”. See Givón 2001 vol. 2, 272; Collier 1991a, 48–49; Uljas 2003, 395. ‘Object raising’ is in Earlier Egyptian restricted to verbs of perception involving visual contact between the governing clause subject and the complement object, namely mAA, siA and xpi ‘encounter’. Other verbs of perception such as sDm do not allow this construal. Also rx and gmi ‘find’ belong to the ‘raising’ group, but with the latter the situation is rather more complex; see 2.4 below. The verbs rdi and iri may also be followed by object + adjunct (GEG § 84) when direct manipulative contact is suggested (see 3 n.52 below). In PT there are examples that look like ‘raising’ after wD ‘order’, see Allen 1986b, 15. 122
affirmative object complementation
97
(101) The deceased says of hostile serpents in the underworld: S1C,123 B1Bo, M23C, G1T: iri=sn n=i wAt nfrt mAA=sn prr=i m kAr S2C, T2C:
iri=sn n=i wAt mAA=sn wi pr=i m kAr They make (a good) way for me when they see how I come out of the shrine/me coming out of the shrine. (CT I 390c–91a) This does not indicate e.g. some mutual syntactic ‘equivalence’ [prr=i + adverbial] and [wi + adjunct clause]. Instead, both the use of irrealis and ‘object raising’ alter the shape of the information conveyed in the complement. They reflect the speaker’s perspective to the situation described and his decisions on what to highlight and what to leave to the background in the process of structuring the information communicated. 2.3 Interim Summary Evidence of assertion with ntt/wnt is sparse after verbs other than locution, cognition and perception. Apart from sdx in example (100) above, there is only a single instance of an ntt-complement after some other predicate in the corpus studied, namely swD, ‘inform’: (102) After a directive outlining offences for which individuals are entered to the criminal dockets, it is decreed:
[ir iw] sp=sn ky sp xr.t(w) smi.t(w) swD.t(w) ntt st Hr Sfd n xbnty sSr mdwt wAH st Hr=s Hr pA Sfd r DAt qnw=sn [Should] a case involving them [arise] again, then a report is to be made and it is to be informed that they are on the criminal docket, with a statement on the matters because of which they were entered on the said dockets, in accordance with their offence. (Urk IV 1109, 5–8) However, this is presumably merely a matter of survival; there is one 123 This is the variant followed; the versions differ slightly in their phrasing of the passage.
98
chapter two
possible instance of the verb ib ‘think’ followed by iwt, the negative equivalent of ntt/wnt.124 In the PT there is also an example of nttclause after the verb sr ‘advice’,125 which further confirms that the use of these elements is not restricted to complements of Dd, rx, mAA, sDm and siA.126 Indeed, it would be strange if this were so, seeing that the pragmatic parameters determining the use of object complements introduced by ntt/wnt and of those without these elements are of very general character and may now be summarised. In Earlier Egyptian, the basic division between ntt/wnt-introduced and ntt/wnt-less complements is organised as follows (S = Speaker, H = Hearer, p = proposition and the information it conveys):127
ntt/wnt
gem. sDm=f/ non-gem. sDm=f/128 no ntt/wnt
It is possible for S to assert p
S is willing to assert p
→ S knows p
→ S believes p → S accepts p → S thinks p is relevant to H
It is impossible for S to assert p
S refrains from asserting p
→ S does not know p → S knows p is not relevant to H
→ S does not believe p → S does not accept p → S does not think p is relevant to H
128
Starting from the bottom half of the above diagram, non-assertion is an expression of the speaker’s ignorance, lack of commitment and acceptance of the complement proposition as well as of its known or perceived lack of discourse relevance. The overall formal signal of such 124
See example (123) below. PT PII, 1055 +30 sr n=k n ra ntt P [iw=f] “Advice Ra that king P comes”; Nt 40– 41 has changed this into Dd n=k n ra ntt N iw=s, “Say to Ra…” etc. 126 Anticipating the discussion in chapters 4 and 6, it should also be noted that the same elements are used to mark assertions also after prepositions and in subject complements of verbs. 127 Cf. here Searle 1969, 66. 128 The labels ‘gem. sDm=f’ and ‘non-gem. sDm=f’ should be understood as also inclusive of the functional counterparts of these forms of the ult. inf. verb-class, such as the doubling form of doubling roots in the first instance. 125
affirmative object complementation
99
a modal status of these clauses is the absence of the elements ntt/wnt and the use of bare sDm=f forms. However, in clauses involving verbs with mutable roots a further distinction is made. Complements with the geminating sDm=f, or some of its functional counterparts from verb-classes other than ultimae infirmae, describe situations that are both ‘objectively’ (mostly in terms of current ‘reality’) and/or ‘subjectively’ (accepted with commitment) ‘near’ to the speaker. Yet, asserting their reality by presenting this as optimally relevant information is not the speaker’s illocutionary intention. As opposed to its functional counterparts, which also have other values, the specialised or marked formal representative of this proximal irrealis is the geminating sDm=f. As a typological class the proximal contrasts with the kind of non-assertion expressed by non-geminating sDm=f forms. These demonstrably cluster to complement clauses describing situations that are both temporally separated from the speaker—i.e. are future or past—and whose assertion is most clearly blocked by the ‘objective’ factor of lack of knowledge. But lack of acceptance, rejection and denial, i.e. when p is most strongly ‘subjectively’ un-assertable to (and not just un-asserted by) S and where there is the greatest motivation for S to distance himself from p, seem also to be domains of non-geminating sDm=f forms. The same holds also for complements known to convey information most lacking in discourse relevance, whose assertion is not merely deemed unnecessary, but actually blocked. The marked forms of this distal irrealis are the sDm=f forms of mutable verbs displaying the endings -w and -y. Both these hierarchically organised categories of irrealis contrast with complements introduced by ntt/wnt, whose distribution is complementary to the former two. Semantic-pragmatically, in such clauses p is presented as information of whose reliability the speaker is prepared to take full responsibility, which he accepts and which is offered to the consideration of the audience as something new, relevant and worthy of their full attention. Grammatically, the status of such realis and asserted p is clearly signalled by the introducing element. Here it may be noted that just as the presence or absence of ntt/wnt is neither random nor based on syntactic criteria, the role of these elements as modal operators and markers of assertion provides a fine example of the principled manner in which grammaticalisation of such function-words occurs. The etymological origin of ntt/wnt in
100
chapter two
expressions referring to existence and reality is not coincidental. It has been postulated that ntt and wnt may have originally represented abstract lexical expressions meaning something like ‘what is/exists’, i.e. ‘fact’, serving as the actual object or subject of the governing verb and followed by a paratactic or appositional clause.129 Later they were probably reanalysed as functional elements forming a clausal unit with what follows in a manner known from other languages.130 If true, the later functional role of the erstwhile lexical ntt/wnt not only as ‘syntacticised’ clause subordinators but also as ‘pragmaticised’ indicators of realis modality seems perfectly natural.131 Nevertheless, the expression of realis is more restricted in Earlier Egyptian object complementation after verbs than irrealis. The speaker’s motives for non-assertion are extremely mixed and only one of them suffices to render the proposition unasserted. In contrast, assertion only takes place when all the relevant conditions and prerequisites are fulfilled and even then it is ultimately the choice of the speaker whether or not he wishes to profile the complement proposition thus. More common overall is distal or proximal non-assertion with the geminating and non-geminating sDm=f’s (or, more generally, without ntt/wnt) which is ubiquitous after nearly all governing predicates. Hence assertion is not the ‘default case’ in the modal system of Earlier Egyptian object complement clauses after verbs, but non-assertion.132 Atypically, but in accord with the previous point, asserted complements are also the more marked option, and non-asserted the less marked. Interestingly, this relationship between ‘markedness’ and assertion/non-assertion seems to have an exact parallel in main clauses. It is noteworthy that
129
See Goedicke 1955; LGEC §§ 701, 705 and Gunn 1924, 176–77. See Heine, Claudi & Hünnemeyer 1991, 205–06; Hopper & Traugott 1993, 14– 16, 185–89; Heine & Kuteva 2002, 211–12, 295; Cristofaro 2003, 96–97 for discussion and examples of such a process. However, it must be stressed that throughout the historical period ntt/wnt are clearly grammaticalised elements and are not be treated as lexical ‘objects’ or ‘subjects’. This honour belongs to the clauses that the elements introduce, following the assumed ‘re-bracketing’ of e.g. the hypothetical prehistoric *rx=f [ntt] [sDm.n=f] as rx=f [ntt sDm.n=f]. 131 Yet as is typical for such function words as ntt/wnt, some remnants of the original lexical use persist: bare ntt functions as a lexeme for “what exists/is”; cf. e.g. the expression ntt-iwtt “what exists and what does not” (GEG § 203.4). 132 As will be seen, this state of affairs finds an exact parallel also in the sphere of subject- and predicate complements as well as in complement clauses of prepositions, and is thus a general principle of all complementation in this language 130
affirmative object complementation
101
sDm=f main clauses with a seemingly ‘indicative’ profile are preceded by some auxiliary element such as iw, whereas clauses of most nonassertive sort (wishes, exhortations) appear un-introduced.133 The hypothesis above is—and could only have been—a result of an examination to the motivations for the use and distribution of the different Earlier Egyptian object complement patterns of verbs as a whole, and, in case of ntt/wnt-clauses, without regard to their internal build-up. However, as noted in the introduction, Egyptological researchers have hitherto preoccupied themselves rather with the question of what sort of forms and constructions are to be found after these elements. This issue must now be addressed, because closer examination of the paradigm after ntt/wnt actually supports the modal analysis of these elements proposed here. The phenomena to be observed allow one to define the modal character of the bare forms and constructions following ntt/wnt and provides an initiative for studying them from this vantage point also in other grammatical environments of Earlier Egyptian. 2.4 The Paradigm after ntt/wnt: Form and Function Revisited In addition to their role as markers of realis modality, the elements ntt and wnt also serve to identify the clauses they introduce as complements. Consequently, besides modal operators they may also be characterised as complementisers.134 Rather in agreement with the earlier Egyptological analyses of ntt/wnt, in linguistics the defining feature of elements assigned to this grammatical category was traditionally taken to be the syntactic function of a subordinator of sentential objects (and subjects).135 More recently, however, complementisers have been increasingly interpreted also as one of the means in which languages encode modality.136 For example, the variation in Japanese of the 133
See conclusion for further remarks on this issue. Barta (1986) argues that ntt/wnt and iwt introduce adjunct clauses in the grammatical idiom of New Kingdom Unterweltsbücher. No comparable uses are observable in earlier material. 135 Generative linguistics forms a notable exception here. 136 In Egyptology this point is most clearly expressed by Ritter 1995, 97–98, who speaks of complementisers as bringing “die innere Einstellung des Sprechers zu seiner Aussage (das sogenannte commitment of the speaker towards the truth value of the 134
102
chapter two
complementisers to and koto has been analysed as expressive of whether or not the complement proposition is presented as ‘psychologically distant’, i.e. untrustworthy, unacceptable, representing someone else’s beliefs, or seen from a ‘wholly objective’, ‘detached’ perspective.137 In Earlier Egyptian, the dual role of ntt/wnt as both syntactic and modal markers is most apparent from the distribution of the clauses that they usher in, but the same holds also with the paradigm that follows them. There are clearly but few restrictions on what forms and constructions can actually occur in this syntactic position. On basis of the examples cited thus far, at least the following patterns are combinable with ntt/wnt:138
The nominal sentence The adverbial sentence Subject + stative Subject + Hr + infinitive Subject + r + infinitive Cleft sentence Existential sentence Ante-position with/without ir + sentence sDm.n=f Active sDm=f with or without subject ante-position (ntt sDm=f/ntt=f sDm=f ) Past passive sDm=f n sDm.n=f
ntt can also introduce second tenses, although in complements after verbs no examples headed by geminating sDm=f are attested. Instances of sDm.n=f apparently in this function are late and far and wide apart:139 proposition) zum Ausdruck”. See also e.g. Palmer 1986, 148–49; Ransom 1986, 87– 91; Wierzbicka 1988, 164–65; Langacker 1991, 446–47; Frajzyngier 1995 and ibid pp. 474–75 for a bibliography on past research on this topic. Cf. also Givón 2001 vol. 2, 72–74. 137 Suzuki 2000; cf. also Palmer 1986, 148. 138 Cf. Johnson 1984, 81–83; Collier 1991a, 29–30 (on ‘preposition-ntt converters’); Uljas 2000, 130. In addition, the negation n sDm=f is attested in ntt-introduced complement clauses of prepositions—see 6.3 below. 139 Cf. Polotsky 1944, § 28 end. Allen (1979, 8) and Silverman (1985, 272; 1986b, 317) argue that the element is is required for second tenses to be able to appear as object
affirmative object complementation
103
(103) Thutmosis III is told that Amun favours him: rdi.n=f n=k tA nb xrp n=f sw rxw ntt pr.n=k xnt=f He gave to you all the land. Lead it for him, for he knows that you have come before him. (Urk IV 1241, 5–6) However, abundant examples of unambiguous second tenses with geminating sDm=f occur in ntt-clauses that serve as complements of prepositions.140 Contrary to claims made by various researchers when attempting to find a syntactic explanation for the paradigm above, it is obvious that the use of ntt/wnt does not depend on whether the construction following these elements is verbal or non-verbal,141 or inherently ‘initial’ or ‘non-initial’. 142 For Junge, ntt/wnt is required if the construals subordinated are to retain their status as sentences rather than clauses.143 However, this claim entails the assumption that e.g. a bare non-geminating sDm=f cannot alone constitute a ‘sentence’, a view to which Junge himself does not readily subscribe.144 Further, the occurrence of ntt/wnt with verbal clauses is not governed by whether or not the verb stands first in the following construal or, more narrowly still, in case of sDm=f, whether or not there is an ante-posed subject.145 Much ingenuity has also been applied, particularly among the ST, on attempts to segment the paradigm internally on basis of the assumption of ntt and wnt as ‘subordinate equivalents’ of the initial elements iw and mk, and their paradigms as somehow mutually correlative.146 complements, even when ntt is present (see 7.2). Yet, in view of the evidence from preposition complements, this claim seems to be incorrect (see 6.3 below). A further example of a sDm.n=f second tense in an object clause of a verb is Urk IV 1531 11/14, where the E-variant has rx.kw ntt wD.n.tw n=f st “knowing that it has been ordained to him” with a sDm.n.tw=f passive. The damaged A-version agrees, whereas the late XVIII dynasty variant D has ntt wD.tw. Wente (1969, 3 n.18) states categorically that the sDm.n=f of verbs of motion (henceforth VOM) is not used in Old Egyptian after ntt/wnt. 140 See 6.3 below for examples. 141 Frandsen 1975, 13; cf. also Allen 1986b, 33. 142 Silverman 1985, 272; 1986a, 38. 143 Junge 1979, 83–84. 144 See Junge 1989, 84; cf. also 3 n.21 below. 145 EAG § 1018. 146 Thus, for Polotsky (1969, 480–81) wnt was originally (in Old Egyptian) a counterpart of iw and followed by the same paradigm as the latter. ntt, compatible with all construals, corresponded to mk, with the later appearance of e.g. wnt-introduced nominal predicates being explained as reflecting the merging of functions of ntt/wnt. Gilula (1971, 16) added ntt to be confined mainly to “adverbial sentences… which are
104
chapter two
Sometimes these speculations have resulted in certain quite correct characterisations of the semantic-pragmatic properties of e.g. wntclauses, which have been dubbed ‘indicative’ due to their assumed parallelism with iw-sentences.147 In general, however, hypotheses of this sort have been plagued by numerous anomalies arising from the impossibility of dividing the patterns following ntt/wnt as the respective paradigms of iw and mk.148 Also where no such detailed divisions have been attempted, the idea of ntt/wnt as ‘syntactic counterparts’149 of iw and/or mk has nevertheless been often raised. This is somewhat surprising in view of the fact that syntactically these pairs have nothing to do with each other: iw and mk introduce independent initial main clauses, whereas ntt/wnt usher in dependent non-initial subordinate clauses. That is, the most basic syntactic characteristics of these two sets of elements are in fact polar opposites. The last point seems to have been recognised in the more recent accounts of ntt/wnt where no correlations with iw and mk are proposed.150 However, this has not disturbed the wider consensus on the syntactic character of the paradigm following the former two elements. As seen, the ST and post-Polotskyan approaches alike have embraced the assumption that it consists of patterns ‘converted’ or nominalised by ntt/wnt.151 One of the more persuasive arguments seemingly favouring this interpretation is the conspicuous absence of the bare geminating of the pattern of non-verbal sentences with iw”. wnt was used for this function in Old Egyptian, where it could also occur in “verbal constructions with iw”, but not in sentences “without iw” (cf. also Gilula 1978, 47). For Frandsen, (1975, 12) ntt subordinates non-verbal sentences with iw, whereas the particle is (see 7.2 below) serves as a subordinator of such sentences without iw. 147 E.g. Malaise & Winand 1999 § 912. 148 For instance, Polotsky’s hypothesis raises the old problem of how can the allegedly ‘adverbial’ ‘pseudo-verbal’ or ‘circumstantial’ patterns substitute for e.g. nominal sentences after elements such as mk/ntt (see Collier 1990a; 1991a). In Gilula’s case, Middle Egyptian examples of ntt preceding “verbal constructions with iw” (even excluding ‘pseudo-verbals’) are common, as are non-verbal sentences without iw (contra Frandsen). Second tenses also belong to this latter group in orthodox ST, but seem to follow ntt regardless. Johnson (1984, 81) argues that with ‘circumstantial’ patterns, iw is deleted after ntt, but still performs its function of assigning the forms/construals a ‘predicative’ value. It may be asked, pace Collier (1990b, 87) why should iw be deleted here if its function is so decisive? 149 Allen 1986a, 11; cf. also Satzinger 1986, 299, 307; 1989, 216. 150 Indeed, Malaise & Winand (1999 § 912) speak merely of a ‘resemblance’ between iw- and wnt-clauses. 151 See 0.1.1.
affirmative object complementation
105
sDm=f, the quintessential ‘nominal form’, from the paradigm. Yet, there is no need to assume that this restriction is syntactically founded. There are also other absentees from the paradigm, and these share features in common mutually and with the geminating sDm=f. It is notable that similarly missing from the set of forms and constructions after ntt/wnt is the thoroughly modal imperative. This constraint is unaccountable if the function of these elements is simply to ‘convert’ main clause construals into complements, particularly as there hardly is a verbal pattern less inherently ‘subordinate’ than the imperative.152 Further, there are no examples of any sort of interrogative constructions after ntt/wnt either. This again would represent a strange gap in the ‘converting’ facilities of the said element—particularly as formally many Earlier Egyptian interrogatives are in fact nothing but nominal sentences, which of course otherwise do occur after ntt/wnt.153 Yet, from the perspective of modality these omissions make perfect sense. Imperatives and interrogatives do not assert, but command and inquire, and this sets them into the same irrealis category with the geminating sDm=f, which as a group is incompatible with the realis-markers ntt/ wnt. The absence of the geminating sDm=f after ntt/wnt thus finds a straightforward explanation from the respective modal profiles of this form and of the said elements. Both indicate the modal status of the clause in which they appear: the geminating sDm=f that it is unasserted and irrealis, ntt/wnt that it is asserted and realis. Although speakers may withhold from asserting almost at will, the reverse is not possible.154 In Earlier Egyptian, the complementisers ntt/wnt are a device of modal marking, just as are particles and, in case of the geminating sDm=f, inflection. Marking a bare [hAA=f] non-assertion 152 However, it is not the case that imperatives can never be subordinated; e.g. in Late Egyptian they occur in adjunct clauses introduced by the subordinator iw. 1erný & Groll (1993, 352) note that in such instances iw “cannot be classified as a subordinizer”, which seems to contradict their own assertion “every initial main clause (both nominal and verbal) may be preceded by the dependent iw, which converts the initial main clause to a non-initial subordinate clause” (ibid. 422). 153 One thinks e.g. of sentences such as pty rxt=f “What is its amount?” (pRhind no. 46, 1) or sy ty pw “Who is he?” (CT IV 191c). 154 This asymmetry follows from the most basic principles of modality. For example, Lyons notes that “it may very well be the case that in the vast majority of the world’s languages it is impossible for a speaker to assert the objective existence of either epistemic or deontic possibilities” (1982, 112). The reason for this is obvious: a mere ‘possibility’ of something lacks precisely the properties that render that something assertable as a proposition.
106
chapter two
as an assertion with ntt/wnt is ungrammatical because the resulting complex would be modally contradictory and un-interpretable. The same holds, of course, also with imperatives and interrogatives, and not only in Earlier Egyptian, but in all languages. The bare geminating sDm=f does not need, nor allow ntt/wnt. On its own this form is capable, when required, of forming complement clauses, but the character of these is exactly the opposite of those with the latter elements. The relationship between ntt/wnt and the other types of active sDm=f is also modally based, but here again arise the same problems of morphology discussed in the introduction (section 0.2). According to some commentators, at least one of the forms encountered after ntt/wnt is what they call the ‘subjunctive’ or ‘prospective’ sDm=f.155 Yet, there is not a single example of these elements introducing a sDm=f displaying the endings -w and -y, which most researchers (including the present author) would probably accept as some sort of ‘prospective’ forms.156 Unless these morphological characteristics were for some unknown reason systematically omitted after ntt/wnt, this speaks strongly against the use of any ‘prospective’ form(s) after these elements. In addition, their inclusion in the paradigm would in effect demolish the argument that the latter consists of forms and patterns that need to be ‘nominalised’ by ntt/wnt, seeing that the ‘prospective(s)’ can and do occur as complements also without the latter. Arguing that the issue is not that of ‘nominality’ plain and simple but rather that the ‘converters’ introduce forms that are merely not specialised for nominal use does not remove this problem either.157 This would, of course, account for the ‘adverbial’ final ‘so-that’-clause use of the assumed no-ending ‘prospective’ or ‘future’ sDm=f, but would still not answer why should the ‘converters’ ntt/wnt be used with this form that nevertheless could form complement clauses alone and without their help. Similarly, the complete absence of forms with the endings -w and -y after ntt/wnt does not support the hypothesis of the presence of some ‘unspecialised future sDm=f’ any more than it does that of a ‘nominal prospective’ after these elements. Yet, as with the geminating sDm=f, their absence is not at all peculiar from the vantage point of 155 Johnson 1984, 81 and Collier 1991a, 29, on basis of examples after r ntt (see 6.3 below) and Collier, ibid, 29 n.35 specifically for ntt/wnt. 156 But cf. chapter 6 n.72 and see section 9.2 for discussion. 157 See Collier 1991a, 29–31; 1999, 54.
affirmative object complementation
107
the present analysis. Forms with these morphological characteristics are similarly specialised for non-assertion: they are marked for distal irrealis function through their inflection and are thus modally incompatible with the assertion-markers ntt/wnt. Other researchers have argued for the appearance of the alleged ‘circumstantial’ sDm=f, a truly ‘non-nominal’ form, after ntt/wnt.158 In support of this claim one could refer to example (58) above showing ntt iri=f, example (28), where ntt is followed by an anticipatory subject, (ntt wDa Ad=f wi) and the following instance in which wnt hosts a subject-anticipating suffix-pronoun in a fashion similar to iw:159 (104) A damaged text narrates the courtiers’ reaction to Wašptah’s sudden seizure before the king:
[...] Dd=sn xr Hm=f wnt=f dbAx=f […] with them saying to his majesty that he was unconscious. (Urk I 42, 11)
However, arguments against the entire category of ‘circumstantial’ sDm=f have already been put forth earlier,160 and these examples give no reason to revise them. As for the writing iri in example (58), this is not distinctive of any ‘circumstantial’ sDm=f but has exactly the same appearance as iri e.g. in examples (95)–(96) above. The former instance can just as well be interpreted as involving a ‘circumstantial’ or a ‘prospective’ and is useless as evidence if these terms are seen as referring to forms rather than functions. The problem is further exacerbated when ntt/wnt-introduced and ntt/wnt-less clauses with immutable verbs are contrasted: (105) Thutmosis III explains why he has built a temple for Amun: rx.kw ntt Htp=f Hr=s (Urk IV 835, 16) I know that he is content with it.
158 E.g. Doret 1986, 34 n.264; cf. also the works cited in n.146, n.148 and n.156 above. 159 Cf. Polotsky 1969, 481; one may also again cite Nt 40–41, ntt N iw=s and Urk IV 1381, 4 with r ntt TAty pH=f tni. 160 See 0.2 above.
108
chapter two
(106) Amun gives thanks to Thutmosis III for his building enthusiasm: rx.n=f Htp(=i) xnt=f He knew that I would be content before him. (Urk IV 883, 14)
Here there is no morphological (or phonological) support whatsoever for the claim that Htp in (105) is a ‘circumstantial’ rather than a ‘prospective’, ‘subjunctive’ or in general a different sDm=f form than Htp in