The Perfect in Old English and Old Saxon: The Synchronic and Diachronic Correspondence of Form and Meaning Morgan Macleod St. John’s College This dissertation is submitted for the degree of Doctor of Philosophy. 69,684 words Declaration This dissertation is the result of my own work and includes nothing which is the outcome of work done in collaboration except where specifically indicated in the text. Acknowledgements I should like to express my thanks to my supervisor, Sheila Watts, for her invaluable help and comments throughout the research and preparation of this work; I should also like to thank my advisor, David Willis, as well as Kasia Jaszczolt and Ian Roberts, for the input that they have provided at various stages of this work. More generally, I am grateful to the faculty and students of the Department of Linguistics, who have been of help in ways too numerous to mention. I would also mention the Cambridge University Statistics Clinic, which has provided me with the opportunity for helpful discussions on statistical techniques; responsibility for the statistical analysis presented in this dissertation is, of course, my own. I should also like to say that this research was made possible through the generous support of a St. John’s College Benefactors’ Scholarship. Summary Most of the Germanic languages developed new tense forms allowing the grammatical expression of fine semantic distinctions, including periphrastic perfects and pluperfects; previously, the preterite alone had been used to express semantic content of this sort. In the absence of robust quantitative data regarding the subsequent development of these forms and distribution in the early Germanic languages, a relatively uncomplicated model has generally been assumed, in which there is little synchronic variation in their use and a steady, though not necessarily continuous, diachronic progress toward the state observed in the modern languages. The goal of this work is to provide accurate quantitative data regarding the apportionment of these semantic domains among the available grammatical forms in Old English and Old Saxon, in order to provide meaningful measurements of the synchronic and diachronic use of the periphrastic forms. Very different patterns were found in the use of these forms in the two languages. In Old Saxon the periphrastic forms are used freely, with a frequency similar to or greater than that of the preterites. In Old English there are no significant diachronic trends, but considerable variation exists synchronically among texts, with some making free use of the periphrastic forms and others preferring the preterite almost exclusively. A number of factors potentially responsible for this variation have been investigated, but none can account for the entire range of observed variation on its own. In the absence of any other account for the observed variation, the hypothesis is proposed that the periphrastic forms and the preterite differed in their perceived stylistic value, in a manner whose exact nature may be no longer recoverable; such a hypothesis would be in keeping with previous findings regarding languages such as Middle English and Middle High German. Old English and Old Saxon would therefore differ in the extent to which they make use of the potential for variation created by the absence of a paradigmatic opposition among the relevant grammatical categories. Table of Contents 1. Introduction ............................................................................................ 1 1.1 Tense and Aspect in Germanic: Origins ................................................. 1 1.1.1 Tense and Aspect: Terminology .......................................................... 1 1.1.2 Tense and Aspect in Proto-Indo-European .......................................... 2 1.1.3 Tense and Aspect in Proto-Germanic .................................................. 3 1.2 The Periphrastic Perfect and Grammaticalization ................................ 4 1.2.1 Introduction .......................................................................................... 4 1.2.2 Grammaticalization .............................................................................. 5 1.2.3 Development of Periphrastic Perfects ................................................ 11 1.2.3.1 General..................................................................................... 11 1.2.3.2 Late Latin and Romance .......................................................... 16 1.3 Periphrastic Perfects in the Germanic Languages ............................... 17 1.3.1 Overview ............................................................................................ 17 1.3.2 The Periphrastic Perfect in Old Saxon ............................................... 19 1.3.3 The Periphrastic Perfect in Old English ............................................ 20 1.4 Role of the Present Study ........................................................................ 24 1.5 Content of the Dissertation ..................................................................... 26 2. The Perfect and Pluperfect as Semantic Categories ......................... 27 2.1 Introduction ............................................................................................. 27 2.2 Events ........................................................................................................ 30 2.2.1 Introduction ........................................................................................ 30 2.2.2 Events: Nature and Representation .................................................... 31 2.2.2.1 Reichenbach............................................................................. 31 2.2.2.2 Davidson .................................................................................. 34 2.2.2.3 Kim ........................................................................................... 37 2.2.2.4 The Present Approach .............................................................. 39 2.3 The Perfect System in Modern English .................................................. 41 2.3.1 The Perfect ......................................................................................... 41 2.3.2 The Pluperfect .................................................................................... 51 2.3.3 Non-Finite Perfect Forms ................................................................... 54 2.3.4 Compositionality of the Perfect .......................................................... 56 2.4 The Perfect System Cross-Linguistically ............................................... 58 2.4.1 The Perfect ......................................................................................... 58 2.4.2 The Pluperfect .................................................................................... 60 2.5 The Perfect System in Old English and Old Saxon ............................... 61 2.5.1 Periphrastic Constructions .................................................................. 61 2.5.2 The Simple Preterite ........................................................................... 63 2.5.2.1 Adverbial Collocations............................................................. 64 2.5.2.2 Sequence of Tenses .................................................................. 65 2.5.2.3 Translation Practice .................................................................. 68 2.5.2.4 The Preterite and the Perfect .................................................... 69 2.6 Conclusion ................................................................................................. 70 3. Methodology ..........................................................................................71 3.1 Introduction .............................................................................................. 71 3.2 Data Sources ............................................................................................. 72 3.2.1 Introduction ........................................................................................ 72 3.2.2 Old Saxon ........................................................................................... 73 3.2.3 Old English ......................................................................................... 74 3.2.4 Chronology ......................................................................................... 80 3.2.5 Editorial Practices ............................................................................... 82 3.3 Selection Criteria for Verb Tokens ........................................................ 83 3.3.1 Introduction ........................................................................................ 83 3.3.2 The Simple Preterite .......................................................................... 84 3.3.2.1 Perfect-Like Preterites ............................................................. 85 3.3.2.2 Pluperfect-Like Preterites ........................................................ 94 3.3.2.3 The Preterite and Mood ........................................................... 99 3.3.2.4 The Preterite and Other Aspects ............................................ 103 3.3.2.5 The Preterite and Verbal Prefixes .......................................... 105 3.3.3 Periphrastic Forms ........................................................................... 107 3.3.3.1 Auxiliaries ............................................................................. 108 3.3.3.2 Stages of Grammaticalization ................................................ 111 3.3.3.3 Unavailability of the Periphrastic Form ................................ 120 3.4 Additional Variables.............................................................................. 122 3.4.1 Discourse Context ............................................................................ 122 3.4.1.1 Direct Speech ......................................................................... 123 3.4.1.2 Personal Discourse ................................................................ 125 3.4.1.3 Exposition .............................................................................. 125 3.4.1.4 Narrative and Indirect Speech ............................................... 126 3.4.2 Translation Practices ........................................................................ 129 3.4.2.1 Introduction ........................................................................... 135 3.4.2.2 Perfect .................................................................................... 136 3.4.2.3 Pluperfect ............................................................................... 137 3.4.2.4 Imperfect ................................................................................ 140 3.4.2.5 Present ................................................................................... 142 3.4.2.6 Perfect Participle.................................................................... 143 3.4.2.7 Other ...................................................................................... 145 3.4.2.8 Recast ..................................................................................... 146 3.4.2.9 Interchange ............................................................................. 148 3.4.2.10 Interpolation ........................................................................... 149 3.4.2.11 Expansion ............................................................................... 150 3.5 Conclusion ............................................................................................... 151 4. Results ................................................................................................. 153 4.1 Introduction ............................................................................................ 153 4.2 Old Saxon ................................................................................................ 153 4.2.1 Distribution of Forms ....................................................................... 153 4.2.2 Discourse Context ............................................................................ 156 4.2.3 Summary .......................................................................................... 159 4.3 Old English ............................................................................................. 159 4.3.1 Distribution of Forms ....................................................................... 159 4.3.2 Diachronic Variation ........................................................................ 161 4.3.3 Translation Practices ........................................................................ 166 4.3.3.1 Introduction ............................................................................ 166 4.3.3.2 Latin Periphrastic Constructions ............................................ 169 4.3.3.3 Translation Practices and Old English Verb Forms ............... 173 4.3.4 Discourse Context ............................................................................ 179 4.3.5 Textual Examination ........................................................................ 183 4.3.5.1 Introduction ............................................................................ 183 4.3.5.2 Dialogues ............................................................................... 184 4.3.5.2.1 The Perfect ....................................................................... 184 4.3.5.2.2 The Pluperfect .................................................................. 192 4.3.5.3 Bede ........................................................................................ 202 4.3.5.3.1 The Perfect ....................................................................... 202 4.3.5.3.2 The Pluperfect.................................................................. 207 4.3.5.4 Boethius ................................................................................. 212 4.3.5.4.1 The Perfect ....................................................................... 212 4.3.5.4.2 The Pluperfect.................................................................. 220 4.3.5.5 Conclusion ............................................................................. 225 4.4 Interpretation of the Data ..................................................................... 226 4.4.1 The Proposed Hypothesis ................................................................ 226 4.4.2 Stylistic Variation in Old English .................................................... 227 4.4.3 Evidence for the Proposed Hypothesis ............................................ 229 4.4.4 Stylistic Variation and its Grammatical Prerequisites ..................... 231 5. Conclusions ......................................................................................... 234 5.1 Introduction ........................................................................................... 234 5.2 Findings of the Present Study in Context ............................................ 234 5.3 Directions for Further Research .......................................................... 238 5.4 Conclusion .............................................................................................. 241 Appendix A ..................................................................................................... 243 Appendix B ..................................................................................................... 267 References ....................................................................................................... 271 List of Abbreviations ABL ablative NEG negative ACC accusative NEUT neuter DAT dative NOM nominative FEM feminine PA past FUT future PF perfect GDV gerundive PL plural GEN genitive PLPF pluperfect IMP imperative PPL participle IMPF imperfect PRES present INF infinitive PRET preterite INST instrumental REL relative MASC masculine SG singular MP mediopassive VOC vocative 1 1. Introduction 1.1 Tense and Aspect in Germanic: Origins 1.1.1 Tense and Aspect: Terminology The focus of the present work is on the verbal systems of Old English and Old Saxon and on their temporal and aspectual categories. Any investigation of the changing relationships among morphosyntax and semantics within the verbal system of the language must ordinarily make reference to those semantic categories, such as tense and aspect, which may be marked morphosyntactically. The grammatical categories under consideration in the present work will be discussed in detail from a semantic perspective in Chapter 2; however, even for introductory purposes it is necessary to establish definitions for certain terms, such as ‘tense’ and ‘aspect’; the exact interpretation of even such basic terms can vary widely from one author to another (for examples see e.g. Kortmann 1991). In this work, the term ‘tense’ is used to refer to the morphosyntactic representation of the temporal location of an event, with reference not to any absolute chronology but rather to a specific deictic centre such as the moment of utterance (see e.g. Reichenbach 1947, 287–98). The term ‘aspect’ is used to refer to the morphosyntactic representation of those properties of an event which may be termed its ‘internal temporal consistency’ (Comrie 1976, 1–3); for the present purposes this is taken to mean all temporal properties of an event other than those represented by tense, including duration, completion, and frequency. In this work the term ‘aspect’ is reserved for only those expressions of internal temporal consistency which have been grammaticalized to the point of receiving obligatory 2 marking in the morphosyntax of a language. The term ‘Aktionsart’ is used here to refer to semantically similar temporal properties communicated by any other means: through the use of a lexeme of whose semantic content such properties form an integral part, for example, or through forms of expression, such as verb- object collocations, from which such properties can be inferred. According to such a definition as this, the same semantic content may be expressed in one language by aspect and in another by Aktionsart (see Sasse 1991). For the sake of brevity, the term ‘tense’ may also be used occasionally in a loose sense to refer to forms in which both tense and aspect are marked, such as the English present progressive, although the finer distinction between ‘tense’ and ‘aspect’ will be maintained in contexts where it is salient. 1.1.2 Tense and Aspect in Proto-Indo-European In order to understand the verbal systems of early Germanic languages and the distinctions that they make among categories of tense and aspect, it is necessary to place these languages within their proper diachronic context and to consider the Proto-Indo-European and Proto-Germanic systems from which they descend. In Proto-Indo-European, at least at the late stage from which the Germanic languages descend, there was a rich verbal system distinguishing a number of categories based primarily on aspect and secondarily on tense (see the summary in Clackson 2007, 133–5). From a given verbal root were derived up to three stems, known as the present, aorist, and perfect stems. The perfect stem, which was ordinarily characterized by reduplication, was used to form the perfect; in some languages this category came at a later stage to have a semantic force similar to that of the perfect in languages such as English, but it is likely to have 3 originally had a stative or resultative 1 meaning (see e.g. Clackson 2007, 121–2; Fortson 2004, 93–5). Some languages also formed a pluperfect from this stem; views differ as to the extent to which the temporal distinction between the two categories was an original feature (see Ringe 2006, 25; Szemerényi 1996, 298). The function of the present and the aorist stems can best be summarized as a distinction between perfective and imperfective meaning (e.g. Ringe 2006, 24–5). From the present stem were formed the present and the imperfect tenses. The imperfect presented a past event as ongoing or incomplete; it contrasted with the aorist, which presented a past event as a complete unit (see e.g. Fortson 2004, 81). No present tense was formed from the aorist stem; the absence of such a perfective present, which has parallels in non-Indo-European languages, has been ascribed to the inherent imperfectivity of the present as a semantic category (Comrie 1976, 66–73). Although some older languages such as Homeric Greek preserve this system with little change (see e.g. Sihler 1995, 564–8), in most languages changes have taken place, which generally operate to reduce the number of inherited distinctions of tense and aspect (see Clackson 2007, 115–8). 1.1.3 Tense and Aspect in Proto-Germanic Some of the most substantial changes to the inherited system occurred in the development of the Germanic languages; the many tense and aspect distinctions made within the Indo-European system were reduced to a simple dichotomy between present and past. The Proto-Germanic present tense was derived from the Indo-European present tense, while the preterite of strong verbs was derived from the Indo-European perfect (e.g. Ringe 2006, 151–3); the origin 1 See Section 1.2.3.1 for a discussion of the semantic differences between these categories. 4 of the weak preterite remains in many respects unclear. 2 The Indo-European perfect, in its earlier stative sense, was also the basis of the Germanic preterite- present verbs which are the ancestors of most modern Germanic modals (e.g. Ringe 2006, 153–5). As a result of these developments, the Proto-Germanic verbal system provided no grammatical means for making distinctions of aspect, and allowed only the temporal distinction of past as opposed to non-past. This situation persisted in Gothic, the earliest-recorded Germanic language (see Braune 2004, 141). However, all the other Germanic languages have developed means of making further distinctions of tense and aspect morphosyntactically; of most relevance to the present study is the fact that all modern Germanic languages have developed periphrastic constructions involving the past participle and an auxiliary such as have or be, which correspond formally to the English perfect and pluperfect (for a cross-linguistic survey see Harbert 2007, 301–6). 1.2 The Periphrastic Perfect and Grammaticalization 1.2.1 Introduction The process by which such periphrastic constructions come to be available in the language as a means of expressing temporal or aspectual properties of a event has been the object of much previous study, as will be seen below, and is relatively well understood. This process can be seen as an example of the type of linguistic change known as grammaticalization. A discussion of grammaticalization as a phenomenon and of some of the terminology that has been used to describe such processes will form a prelude to a discussion of the 2 For reviews of the extensive bibliography on this subject see e.g. Tops (1974); Hill (2010). 5 grammaticalization of periphrastic perfects both as a general cross-linguistic phenomenon and in the Romance languages. The examination of similar processes elsewhere will then provide a basis for considering the history of these forms in the Germanic languages themselves. 1.2.2 Grammaticalization Although the term ‘grammaticalization’ has been used in different ways in studies embracing a wide variety of phenomena, it may broadly be said that grammaticalization is the development of grammatical morphemes, either from lexical morphemes or from other grammatical morphemes (e.g. Hopper and Traugott 1993, 2; Bybee et al. 1994, 4). It should be noted here briefly that a certain amount of terminological variation exists; some of the works cited here use the term ‘grammaticalization’ to refer to such developments while others use the terms ‘grammaticization’ or ‘grammatization’. In the present work these terms are treated as synonymous and the form ‘grammaticalization’ is used uniformly throughout (for further discussion of the terminological issues, see e.g. Traugott and Heine 1991b, 1–2). The concept of grammaticalization has its roots in the original use of the term by Meillet (1912, 131; see further Hopper 1991, 13–4). Although the theoretical assumptions behind the use of this term have varied, contemporary descriptions of grammaticalization are generally founded on concepts such as that of a given form’s progression from a semantically independent lexical item, outside the grammatical system of the language as such apart from membership in syntactic categories such as ‘noun’, to a form primarily characterized by its role in the grammatical system as an expression of abstract semantic features, which has little other semantic content (e.g. Hopper and 6 Traugott 1993, 2–8). It should be noted that such a view does not necessarily imply that each form undergoing grammaticalization passes through a fixed set of stages in a deterministic way. The process of grammaticalization is often envisioned as a cline; an example of such a cline is that given by Hopper and Traugott (1993, 7), who present a typical transition ‘content item > grammatical word > clitic > inflectional affix’. As an example of such a transition, they take the history of English let’s (1993, 10–14). This construction has progressed from the earliest stage, in which both let and us have their full value as content items (setting aside for the moment any distinction in semantic content which may be said to exist between the categories of verb and pronoun), to a subsequent stage, in which let has become generalized in a hortatory sense and has undergone a certain degree of semantic bleaching which may be said to have moved it closer to the ‘grammatical word’ status of an auxiliary; it has progressed further to a stage in which the construction is used more in the first person plural than in other persons, and in some dialects a further stage has been reached, in which let’s is no longer restricted to the first person plural, and the form ’s has presumably been reanalysed so that it is disassociated entirely from the independent pronoun us. Another well-known example of grammaticalization is the development of the Romance future from the verb habere ‘have’, from constructions in which the verb had its literal sense as a lexical verb (‘to have something to do’) to constructions in which it had become an auxiliary of obligation (‘to have to do something’), to a pure marker of the future, which in many Romance languages has become a suffix rather than an independent auxiliary; in addition to the 7 developments producing this form, languages such as French have also begun a similar process of grammaticalization that has developed new auxiliaries with future reference such as aller ‘go’ (e.g. Hopper and Traugott 1993, 42–5). Both these developments can be understood in terms of the typical grammaticalization process described above, in which forms lose their parity with other lexical items and enter a stage in which they are far more closely integrated with other aspects of the morphosyntax of the language and play a primarily grammatical role. These cases also illustrate another feature of grammaticalization, the retention in the language of earlier stages of the process; for example, let remains a lexical verb in English, as do the reflexes of habere in many Romance languages. It may be seen from these examples that the different developments encompassed by the term ‘grammaticalization’ as it is used here, involve change in several different areas of the language, including semantics, morphology, syntax, and in some cases, phonology (e.g. Andersen 2008, 15). For example, the history of let’s described above involves phonological change, seen in the cliticization and phonological reduction of the pronoun us, and morphosyntactic change, as in the dialects in which this reduced form is no longer seen as a personal pronoun and is reanalysed perhaps as an inflection or as part of the root. As a result of the many linguistic areas impacted by grammaticalization processes, there is variation in the particular types of change on which different studies focus. One effect of grammaticalization can be the syntactic reanalysis of a construction whose original syntax has become opaque, and some studies of grammaticalization have concentrated on this process (e.g. Roberts and Roussou 2003); other studies have placed greater emphasis on the boundary between 8 syntax and morphology and on the ways in which grammaticalization can traverse this boundary (e.g. Joseph and Janda 1988). Other studies concentrate on the semantic developments which are often interpreted as the cause of such morphosyntactic changes; for reasons that will be discussed in Section 1.2.3 below, it is semantic work of this sort that is most relevant to the present work, although the relationship between such semantic changes and the morphosyntax of a language must be borne in mind. One approach to the role of semantics in grammaticalization is proposed by Heine (2002); as this is based in part on the model proposed by Diewald (2002), it will be most convenient to discuss the latter first. Diewald depicts grammaticalization as a process of semantic shift involving three main stages characterized by the contexts in which the form in question occurs: the earliest stage, that of untypical contexts, in which the meaning that forms the semantic basis for grammaticalization arises as a pragmatically specific interpretation of the construction’s original sense; a later stage, that of critical contexts, in which there are no contextual cues favouring either the older or the newer interpretation; and the final stage, that of isolating contexts, in which the construction is used in a way that definitely excludes the possibility of interpretation in the original sense. Diewald’s definitions of these stages were originally made in the context of her work on German modals and the growth of differentiation between what may be called epistemic and deontic usages; the stages can best be illustrated by examples of the forms to which they were first applied (all examples adapted from Diewald 2002): 9 (1) Das muß man alles that-NEUT.NOM.SG must-3SG.PRES one-NOM all-NEUT.ACC.SG erst mal wissen. first-ACC time-ACC know-INF ‘One must first understand all that.’ This sentence is described by Diewald as exemplifying the ‘untypical context’ stage; the deontic sense, ‘one is under an obligation to understand all that’, is less likely to be intended in a literal sense, and the epistemic sense, ‘it is necessarily true that one understands all that’, can arise pragmatically through implicature. (2) Der muos se baʑ that-MASC.NOM.SG must-3SG.PRES her-ACC better gelobet hân. praise-PA.PPL have-INF The ‘critical context’ can be seen in Middle High German examples such as (2), which contains a perfect form; such constructions were originally susceptible of three different interpretations: the deontic ‘he has had to praise her better’ and the epistemic ‘he must have praised her better’, as well as the stative interpretation ‘he must have her better praised’. Such constructions, as Diewald observes (2002, 111) were ambiguous in the absence of contextual cues. However, in the actual text such cues were generally present; their frequent presence would tend to neutralize the distinction between epistemic meaning arising only as a contextual implicature and epistemic meaning expressed by the modal itself, so that the support provided by the context would become redundant. (3) Drumb haben si mussen fallen. therefore have-INF they-NOM must-PA.PPL fall.INF ‘Therefore they (have) had to fall.’ 10 (4) Drumb musz das der therefore must-3SG.PRES that-NEUT.ACC.SG the-MASC.NOM.SG heubt teuffel selbst gesagt haben. head-NOM devil-NOM self said-PA.PPL have-INF ‘Therefore the head devil must have said that himself.’ These examples typify the ‘isolating context’, in which the two senses are fully and unambiguously differentiated; the texts from which these examples are taken make it clear that sentence (3) can be interpreted only in the deontic sense, while (4) can be interpreted only in the epistemic sense. This resolution of the previously existing ambiguity into two discrete and context-independent usages is considered as indicating a stage of grammaticalization more advanced than that seen at periods where the innovative usage is restricted to certain contexts. The model proposed by Heine (2002) is similar to Diewald’s, dividing the process of grammaticalization into an initial stage, in which only the original, basic sense of a construction is present, bridging contexts, similar to Diewald’s ‘untypical contexts’, and switch contexts, similar to Diewald’s description of the later ‘isolating contexts’; Heine differs from Diewald in distinguishing a further, last stage, conventionalization, in which the grammaticalized construction not only has the new sense independently of the context but can occur in environments that are not only semantically but syntactically incompatible with the original construction. Heine discusses this model with reference to the evolution of originally reflexive constructions into passives, a development that has taken place independently in numerous languages, and suggests that in this case conventionalization is marked by the appearance of external agents. As will be shown by the subsequent discussion in this work of perfects, there is value in Heine’s recognition that further developments may take place even after a form is 11 in regular use in its grammaticalized sense; however, the question may be raised of whether clear syntactic criteria for the stage of conventionalization, such as Heine finds for passives, exist in all cases. 1.2.3 Development of Periphrastic Perfects 1.2.3.1 General Periphrastic perfects are formed using a variety of auxiliaries, generally derived from verbs with such predisposing factors as semantic vagueness and generality (see Heine 1993, 30–2); as will be seen below, the two most common verbs for the Romance and Germanic languages are those meaning ‘have’ and ‘be’. As a result of the semantic differences between these verbs, their grammaticalization as auxiliaries takes place along very different paths. Because of these differences, and because of the tendency of much work to focus on a specific auxiliary, they will be discussed here separately. The use of verbs meaning ‘have’ as an auxiliary is generally thought to arise from constructions in which they are used as lexical verbs, taking as their object a noun modified by a past participle (e.g. Bybee et al. 1994, 68–9). Over time these constructions are reanalysed, in ways that may differ in detail from one language to another, so that the temporal anteriority that was originally denoted only by the participle comes to be the primary meaning of the construction as a whole, 3 and the noun is no longer the object of have but of the verb that appears morphologically as a past participle; in this way, to state the matter simply, constructions that originally meant ‘to have something done’ come to mean ‘to 3 Nevertheless, as discussed in Chapter 2, the status of indivisible semantic unity may not be completely attained. 12 have done something’. This process has been cited as a characteristic example of grammaticalization, exhibiting traits prototypical of grammaticalization processes, such as the increasing loss of independence on the part of the verb have, which begins as an independent lexical verb and becomes a semantically weaker auxiliary that adds little content to the past participle except temporal and aspectual information, sometimes progressing to cliticization of the auxiliary of the sort seen in English (I’ve, you’ve, etc.) (see e.g. Hopper and Traugott 1993, 6– 8). 4 The semantic and pragmatic factors driving this process are necessarily complex, and various proposals have been made regarding the semantic properties of have that provide a unifying element among the different stages of grammaticalization; for example, Jacob (1995) has emphasized the experiencer role of the subject of have, while de Acosta (2006) has interpreted the process in terms of an abstract concept of pertaining. The individual stages of this process as they apply to the use of have as an auxiliary will be examined in greater detail in the following sections with reference to developments within individual languages. Verbs meaning ‘have’ are not the only source of auxiliaries for the formation of perfects; in many languages verbs originally meaning ‘be’ are used in this role, often in complementary distribution with ‘have’-auxiliaries. Whereas auxiliaries of the latter type are based on transitive constructions involving a past participle with passive meaning, modifying a noun denoting the patient of a given event, perfect periphrases using an auxiliary meaning ‘be’ are based upon past participles with active meaning, a type occurring in many Indo-European 4 However, in some circumstances clitic forms are used for have even as a lexical verb; see Trudgill et al. 2002 for a discussion of this phenomenon and its dialectal variation. 13 languages (e.g. Fortson 2006, 97–8). Predicate constructions involving such active participles express not merely a state but a state which is the result of a past action, as denoted by the participle; constructions with this semantic content may come to be seen as a distinct category within a language, and the term ‘resultative’ has been applied to categories of this sort (see Dahl 1985, 133–5). The relevant semantic processes may be illustrated in a simplified manner using Modern English. The participle in a phrase such as the fallen tree describes a situation in which the tree stands in the same relationship to the event of falling as it would as the subject of an active sentence such as The tree fell; participles of this sort can also occur in predicational sentences such as The tree is fallen, and in some languages such sentences can form the basis of a new class of resultatives. Resultatives differ from perfects in that the former necessarily entail the persistence of the relevant state at the time in question, as the following examples illustrate: (5) He is gone. (6) He has gone (and come back). In (5), the person to whom the pronoun refers must still be away at the moment of speech, while in (6), as the addition in parentheses shows, this is not necessarily the case, and his absence need not persist at the moment of speech. It should perhaps be noted that although the application of the term ‘resultative’ to English constructions such as that in (5) above follows the practice of Dahl (1985) and Bybee et al. (1994), based on the semantic equivalence of English constructions such as this with resultatives in other languages in respect of their truth condition, this should nevertheless be understood merely as an illustrative device to provide a readily understood example of the semantic properties of resultatives, rather 14 than an assertion that there is a coherent ‘resultative’ category in the grammar of Modern English. In languages where resultatives do exist as a discrete grammatical category, the notion of a persistent state can disappear from their semantic content so that they develop into true perfects (Bybee et al. 1994, 68–9); this can happen even in languages where the resultative is not formed periphrastically in the manner described above, such as with the inflectional perfect of Classical Greek (see further Sihler 1995, 564–8). It was remarked above that many languages make a distinction between auxiliaries derived from verbs meaning ‘have’ and those from verbs meaning ‘be’. The prototypical pattern for the distribution of these auxiliaries may be described broadly and neutrally as the use of ‘be’-auxiliaries with intransitive verbs denoting ‘a change of place or state’ and of ‘have’-auxiliaries with other verbs (Shannon 1995, 130). One explanation that has been proposed for the frequent occurrence of this distributional pattern is that there is a fundamental syntactic difference between the two groups involved; it has been suggested that the intransitive verbs used with ‘be’-auxiliaries form a class of ‘unaccusative’ verbs, whose subject is in origin syntactically identical with the object of transitive verbs, and that the use of verbs meaning ‘be’ as perfect auxiliaries for such verbs is thus fundamentally identical with their use as passive auxiliaries (see e.g. Perlmutter 1978; Burzio 1986). These analyses of auxiliary selection are based on the evidence that in many languages unaccusative verbs can be shown to form a discrete syntactic class; for example, in ergative languages the same morphological case, termed ‘absolutive’, is used for the subject of unaccusative verbs and the object of transitive verbs, in opposition to the ‘ergative’ case, which is used for the subjects 15 of transitive verbs and other intransitive verbs (commonly called ‘unergative’) (see Dixon 1994; Perlmutter 1978). However, the analysis of auxiliary selection on the basis of a syntactic dichotomy between unaccusative and unergative verbs raises the issue of how verbs are assigned to these classes; it is necessary to explain not only how apparently synonymous verbs in different languages may differ in their auxiliary selection but also how the same verb in a particular language may show flexibility in its auxiliary selection (see further Sorace 2000). A variety of approaches have been taken to deal with the non-binary nature of auxiliary selection. Some authors reject the syntactic analysis entirely, such as Shannon (1995), who takes a cognitive–semantic approach to analyse auxiliary selection on the basis of semantic continua related to factors such as transitivity and affectedness. Others attempt to reconcile these extremes; for example, Sorace (2000) identifies the graded semantic continua that are observed in auxiliary selection as playing a role in other syntactic phenomena that display a similar degree of cross-linguistic variability, although she does not attempt to formulate a formal syntactic model encompassing all the observed distributional patterns of auxiliary use. The issue of auxiliary selection is one that has received considerable attention in the literature (for further discussion see Aranovich 2007b), and the proposal of a new model to explain these phenomena is beyond the scope of the present work; however, an interpretation in keeping with Macleod (2008) would be that verbs can be assigned to the unaccusative and unergative classes on the basis of a number of factors, lexical determination in some cases and semantic and pragmatic factors in others, with the exact factors operating in a given case varying cross-linguistically; once such an assignment has taken place, 16 the resulting property of the verb, however conceived, can play a role in syntactic processes. For the purposes of the present study, a relatively broad terminology will suffice for descriptions of the phenomenon of auxiliary selection; the term ‘unaccusative’ will be used as a convenient label for the kind of intransitive verb that can take an auxiliary with the original meaning of ‘be’, and the term ‘unergative’ to refer to the kind that cannot, with no commitment to the nature or origin of this distinction; these terms are to be understood in a purely descriptive and language-specific sense, with no implication that the categories defined in this way are of diachronic or cross-linguistic application. 1.2.3.2 Late Latin and Romance The history of the Romance languages provides one of the best-known examples of the development of periphrastic perfect constructions through a process similar to that described above. The origins of these Romance constructions can be traced back to Late Latin, when habere ‘have’ and esse ‘be’ were already coming into use as auxiliaries of the perfect, in a distribution closely paralleling the prototypical situation described above (see Vincent 1982). Although these forms are generally considered to have arisen through a process of grammaticalization native to Late Latin, the existence of similar constructions in Greek has sometimes been noted as a possible factor contributing to the development of similar forms in Latin (see e.g. Stotz 1998, 330); however, the process of grammaticalization may be considered to have proceeded by similar stages in either language. Harris (1982) describes the preservation of different stages in this process of grammaticalization synchronically in modern Romance languages: in Sicilian, periphrastic constructions using a reflex of Latin habere as 17 an auxiliary still have a purely resultative meaning, denoting a present state resulting from a past event in the manner described in Section 1.2.3.1 above; in Portuguese, formally identical constructions are true perfects semantically, in the sense described below in Chapter 2, but remain marked expressions beside a past tense that can be used as an unmarked form to refer to the same events, while in Castilian Spanish a paradigmatic opposition has developed so that the preterite is marked as non-perfect and the two categories are no longer interchangeable. In French, these constructions have developed even further, so that they are largely divorced form the perfect as a semantic category and have become primarily a perfective past tense (see further Bybee et al. 1994, 81–7). The changes in the status of these constructions, in the course of their progression towards a greater degree of semantic abstraction and a closer integration into the grammatical system, can be seen as compatible with the previously described picture of prototypical grammaticalization processes. 1.3 Periphrastic Perfects in the Germanic Languages 1.3.1 Overview It might reasonably be expected that the grammaticalization of Germanic periphrastic perfects followed lines similar to those of the corresponding Romance process, given the semantic similarity of the two groups in regard to their past- participle morphology and the lexical verbs from which the auxiliaries are derived; however, it will be seen that the evidence for the Germanic languages is sparser in some respects and allows for greater differences in interpretation. The semantic pathways by which the grammaticalization of these constructions took 18 place are generally uncontroversial, despite some variation as to which of the senses of the polysemic verbs involved contributed most to their becoming auxiliaries (for discussion see de Acosta 2006, 1–17). However, greater controversy exists regarding the points at which different stages in this process were reached. Brinton considers the grammaticalization of periphrastic perfects to have been already in progress at the Common Germanic period; she argues that alternative explanations require ‘independent parallel variations of an unlikely extent’ (1988, 107). In contrast, others have observed (e.g. Harbert 2007, 301–2) that the use of the preterite in Gothic and Old High German as the sole translation equivalent for the Latin imperfect, perfect, and pluperfect, as well as the high degree of variation found among the Germanic languages both in the selection of auxiliaries by individual verbs and in the verbs that are used as auxiliaries of the perfect (with some languages extending beyond the basic verbs meaning ‘have’ and ‘be’ to make use of verbs meaning ‘own’ or ‘become’), would seem to suggest a certain degree of independent innovation in the individual Germanic languages. It will be seen in Section 1.3.3 below that some studies on Old English conclude that these periphrastic constructions were still at an extremely primitive stage in the earliest recorded texts; in this way, they too advocate a late date for much of the development of these constructions. Mention might also be made of the suggestion that the Germanic perfect periphrases have their origin in calques of similar Latin constructions; Drinka (2003; 2007), a recent proponent of this view, asserts that given the existence of these constructions in Latin, as well as the exposure of the Germanic-speaking peoples to Latin and their physical proximity to Romance speakers, areal diffusion is a more parsimonious explanation than 19 independent innovation. It should be noted that such a view does not preclude the possibility of different processes of borrowing in different Germanic languages, and is therefore not necessarily incompatible with the variation described by Harbert (2007). Nevertheless, there is little positive evidence to connect the Latin constructions with those found in the Germanic languages, or for the high degree of influence which Latin texts would presumably need to exert upon the vernacular language (see de Acosta 2006, 17–19). As discussed above, the evidence for the earliest stages of the Germanic languages is seldom sufficient to confirm or disprove specific hypotheses, but given the absence of any conclusive evidence against the independent development of perfects, not only within Germanic as a separate group but separately within individual languages, together with the cross-linguistic frequency of developments of this sort (e.g. Bybee et al. 1994, 68–9), it is assumed here that some degree of independent innovation in the history of Germanic perfects is a simpler explanation than one involving borrowing. The comparison of translated Old English texts with their Latin originals in Section 4.3.3.2 below will provide further support for this position. 1.3.2 The Periphrastic Perfect in Old Saxon The scant textual record for Old Saxon makes it possible to summarize what is known about the periphrastic perfect in this language quite briefly. Recent work on Old Saxon perfect constructions has focused on the Heliand, the longest surviving Old Saxon text (Arnett 1997; Watts 2001). In this text, perfects exhibiting an advanced degree of grammaticalization are found already in frequent use; the two auxiliaries in primary use to form perfects are hebbian ‘have’ and uuesan ‘be’ (Watts 2001, 125–30). For the most part, the use of these 20 auxiliaries follows the prototypical distribution pattern discussed in Section 1.2.3.1 above, so that uuesan is used with unaccusative verbs and hebbian is used elsewhere; however, some potential existed for variation in auxiliary choice, based on factors such as modality and the affectedness of the object (Arnett 1997). In addition to these clearly grammaticalized perfects, formally identical constructions occurred which unambiguously retained the original stative meaning (Rauch 1992, 178–9); as discussed in Section 1.2.2 above, the persistence of such forms after the development of the new perfect is to be expected. As will be seen in Chapter 4, the results of the present study confirm that the picture for the periphrastic perfect in Old Saxon is relatively uncomplicated, whether this is a result of the limited corpus or an accurate reflection of the state of the language in general. 1.3.3 The Periphrastic Perfect in Old English For Old English, a greater range of data is available, and a greater degree of dispute exists regarding the diachronic status of the periphrastic perfect in Old English and its ongoing development. Nevertheless, there are certain points common to most previous analyses; in the absence of robust quantitative data regarding their subsequent development and distribution in the early Germanic languages, a relatively uncomplicated model has been assumed, in which there is little synchronic variation in their use and a steady, though not necessarily continuous, diachronic progress takes place toward the state observed in the modern languages. Visser (1973, IIIb, 2189–93) depicts the periphrastic perfect with have as having developed over the course of the Old English period from a stage in which it could occur only with transitive verbs to a later stage, first visible 21 around the turn of the eleventh century, in which ellipsis of the object became possible, and then to a stage in which these constructions could be used with intransitive verbs. Visser considered this last stage to have been reached only at the beginning of the twelfth century, and suggested that such constructions began to reach their modern level of productivity only within Middle English. The inaccuracy of the dates proposed by Visser for the points at which these stages were reached has often been noted; for example, Mitchell (1985, I, 289–91) cites numerous examples of periphrastic constructions involving intransitive verbs from the earliest texts. Nevertheless, other authors often view these periphrastic constructions as having developed diachronically within Old English in a manner similar to that proposed by Visser. Denison, although citing Mitchell’s discussion and providing a number of early intransitive examples from elsewhere in Visser’s own corpus, suggests that have was not available as a general auxiliary for all lexical verbs until late Old English (1993, 352). He also interprets the not infrequent coordination of periphrastic constructions with preterites having the same temporal reference, as well as the Middle English use of the perfect with definite past-time modifiers, as indicating that the semantic domains of the perfect and the preterite were entirely coextensive, suggesting that until Early Modern English the periphrastic perfect was a ‘pure tense equivalent’ (ibid.). Carey (1994), working from a different perspective in her investigation of the role of pragmatics in the grammaticalization of the perfect, reaches similar conclusions about the periods during which the periphrastic forms were undergoing significant diachronic changes. She concludes that the periphrastic constructions in early Old English had only a resultative meaning and that the modern perfect meaning was 22 not fully attained until Middle English, 5 based on the co-occurrence of adverbial modifiers with periphrastic constructions; the Old English perfects in her sample occur only with temporal adverbs referring to points in time at which the event denoted by the verb is completed, such as now and when, and are not found with manner adverbs modifying the event itself. 6 Despite the small quantitative difference between the samples of perfects taken from early and late Old English texts, Carey suggests that the absence of certain semantic categories, such as perception verbs, from the earlier sample 7 is a sign that the grammaticalization of periphrastic constructions continued to progress noticeably over the course of the Old English period. The assumption by such disparate authors as Visser, Denison, and Carey of a model in which the periphrastic perfect and pluperfect undergo perceptible progress over the course of Old English towards their modern state should indicate the widespread appeal of such a view; certainly, given the existence of a prior state before the appearance of these forms and given their continuing development in Middle English and after, which will be discussed in Section 5.3, it is plausible a priori to suppose that diachronic trends of the sort that have been proposed would be visible in Old English. In addition to the studies described above, there are other analyses of the Old English periphrastic perfects with have that consider them to be more stable diachronically across this period. As mentioned above, Mitchell (1985, I, 282– 98), although he acknowledges the existence of some diachronic trends such as 5 See Section 1.2.3.1. 6 For a counterexample see Wulf II.121.69 (shown as (41) below). 7 For counterexamples see e.g. CP LIII.413.14, GD MS C II.XIV.133.2 (shown as (167) and (168) in Appendix A). 23 the decline of inflected participles, emphasizes the lack of correspondence between such trends and any perceptible semantic distinction and demonstrates the existence at all points within the Old English period of periphrastic constructions that appear modern by any morphosyntactic or semantic criteria. Brinton (1988), like Mitchell, remarks the presence of apparently modern periphrastic perfects and pluperfects in the earliest texts and the absence of any firm correlations between the semantic content of the constructions and instances of morphosyntactic variation, such as differences in word order and participle agreement; she concludes that the development of the periphrastic forms was essentially complete by the time of the earliest texts and that these grammatical categories remained stable throughout Old English. Wischer (2002) differs with Brinton’s conclusions regarding the completeness of the grammaticalization of these constructions, considering the possibility that further conventionalization 8 was still to take place even after the attainment of the grammaticalized state described by Brinton; however, she treats the entire Old English period as a single unit for the purposes of textual analysis. Although no explicit discussion is provided of whether this is merely a methodological decision or whether it reflects a theoretical stance regarding the homogeneity of Old English in the use of these forms, it may nevertheless be inferred that the diachronic development of these constructions during Old English was not considered significant. The fact that such a wide variety of positions are held regarding the diachronic development of the periphrastic perfect illustrates one way in which further data are needed regarding the actual use of these forms. 8 In the sense of Heine (2002); see Section 1.2.2. 24 In addition to the perfects formed with habban ‘have’, Old English also used wesan ‘be’ as an auxiliary of the perfect. Unlike the situation in Old Saxon, in Old English habban could be used even with unaccusative verbs, although for these a perfect with wesan was also possible (see Mitchell 1985, I, 289–301). The differing semantic properties and diachronic paths of the two types of perfects, as discussed above, make it advisable to consider them separately; for methodological reasons that will be discussed in Chapter 3, the present study will focus exclusively on Old English and Old Saxon perfects formed with auxiliaries originally meaning ‘have’. Subsequent references to the periphrastic perfect and pluperfect may be taken as referring solely to constructions of this sort unless otherwise specified. 1.4 Role of the Present Study As has been seen, some forms of variation in the use of the Old English periphrastic perfect, such as diachronic trends, have been the object of differing views; other forms of variation, such as synchronic differences among texts, have received little attention. Much previous work has focused on the grammaticalization processes that made these forms initially available for use with the meaning found in the modern language (e.g. Wischer 2002; de Acosta 2006) rather than developments subsequent to their first appearance with this sense. In Modern English a paradigmatic opposition exists between the perfect and other categories such as the past tense, similar to that described by Harris (1982, 54–6) for Castilian Spanish; where the development of this opposition has been recognized, it has usually been seen as part of a more general process of 25 grammaticalization (e.g. Hoffmann 1934; Denison 1993, 352), and little examination has been given to the degree to which this paradigmatic opposition may have developed separately from other aspects of the grammaticalization process. What quantitative data exist on the use of the perfect in Old English are derived from studies (e.g. Carey 1994; Diewald 2002) that consider only the periphrastic constructions and ignore the semantically comparable uses of the preterite, providing no means of distinguishing effects that are caused by grammatical changes from those caused by differences in the content of texts. The goal of the present study is to provide accurate quantitative data regarding the apportionment of the perfect and pluperfect, regarded as cross-linguistically applicable semantic domains, among the available grammatical forms; in order to provide a meaningful standard of comparison by which trends in the use of the periphrastic forms can be measured, preterites semantically comparable to the perfect and pluperfect will also be examined. In order to allow cross-linguistic comparison and identify language-specific factors in the use of these forms, data from Old English and Old Saxon are included. It will be seen that these languages differ significantly in their use of the periphrastic forms and in the degree of synchronic variation among texts, with Old English exhibiting a much wider range of variation than has often been assumed. Factors that could potentially influence the choice of a particular form as an expression of perfect or pluperfect meaning have also been examined, including pragmatic context and, in the case of translated texts, the form of the original construction. It is hoped that the data provided by this study will contribute to a more accurate picture of the use of 26 periphrastic forms in Old English and Old Saxon, one that reveals complexities overlooked by much previous work. 1.5 Content of the Dissertation As suggested above, a premise of this study is that the perfect and pluperfect are cross-linguistically applicable semantic categories that may be mapped in different ways to grammatical forms. In order to identify such forms, it is necessary to arrive at a definition of these forms as semantic categories, which will be provided through the semantic discussion in Chapter 2. Chapter 3 will describe the methodology of the study and the ways in which the semantic views described in Chapter 2 are put into practice, while Chapter 4 will examine the results of the study. Chapter 5 will provide a conclusion that relates these findings to the theoretical questions discussed above. 27 2. The Perfect and Pluperfect as Semantic Categories 2.1 Introduction In order to arrive at definitions of the perfect and pluperfect as semantic categories that can be considered independently of particular grammatical forms, it is necessary to analyse the semantic content of specific grammatical forms and determine which components of this content might relate to such cross-linguistic categories. As will be seen in Section 2.4, similar approaches have been used in previous cross-linguistic studies such as Dahl (1985). Definitions of tense and aspect categories generally make recourse to other terms, such as ‘event’, which must themselves have a definition that is understood. In comparison with other verbal categories, the content of the perfect and pluperfect, particularly of the former, is complex and combines semantic and pragmatic elements to such an extent that it can be difficult to separate the two; it should be noted here that references to the ‘semantic’ content of these categories in this work are to be taken as also referring, for the sake of brevity, to any associated pragmatic content except where an explicit distinction is made between the two categories. The following semantic discussion will take as its starting point the Modern English perfect and pluperfect; once the semantic content of these categories has been analysed sufficiently for the present purposes, it will be possible to see to what extent this content is associated cross-linguistically with forms such as the periphrastic constructions of Old English and Old Saxon and to the constructions involving the Germanic preterite which these forms eventually supplanted. It will 28 be seen in Chapter 3 that these issues are closely bound up with the methodology of the present study. Considerable variation exists in the terminology used to refer to the perfect and related forms. The perfect, pluperfect, and future perfect are often described as ‘tenses’, as for example in the analysis by Reichenbach (1947) discussed below in Section 2.2.2.1; however, the extent to which these forms may be said to fall within the semantic category of tense is a complex issue, and a wide variety of semantic classifications have been proposed for these forms. In this work the term ‘perfect system’ is used to refer to the perfect, pluperfect, and future perfect, together with non-finite forms such as perfect infinitives and participles. While the finite members of this system resemble tenses in describing the temporal location of an event, they have the distinctive semantic property of expressing ‘relative’ temporal relations, relating events to a reference time not necessarily identical with either the event time or the time of speech (see Reichenbach 1947, 297). Some authors adopt definitions which explicitly stipulate that the only temporal relations indicated by the term ‘tense’ are those pertaining to the moment of speech (e.g. Kortmann 1991, 19), a definition which would automatically exclude the perfect; conversely, other authors view canonical tenses such as the present, where the temporal reference is necessarily relative to the moment of utterance, as simply a specific case of a more general principle according to which tense may express the temporal relation of an event to any point (e.g. Portner 2003, 478). The relationship of these forms to the category ‘aspect’ has been the object of similarly varied views. The prototypical form of aspect is often seen as an imperfective/perfective distinction such as is marked in 29 Slavic languages (e.g. Jakobson 1957), and some authors have preferred to restrict the term ‘aspect’ to such distinctions (see Kortmann 1991). Although the semantic content of formal categories described as aspects often relates solely to the internal structure of an event without reference to any external points in time (see Comrie 1976, 52), the additional frame of reference introduced by the perfect may play a similar role by allowing the temporal boundaries of an event to be described more specifically and with greater precision (cf. the contrast between John is resting for an hour and John has been resting for an hour). Furthermore, grammatical forms whose primary significance is prototypically aspectual may also be used to indicate relative temporality; this is the case with non-indicative forms of the verbal stems in early Indo-European languages such as Classical Greek, mentioned previously in Section 1.1.2 (see further e.g. Goodwin 1894, 275–6 and Section 2.3.3 below). The findings of the present study are dependent neither on the assignment of the perfect to a specific category such as tense or aspect nor on a particular terminological system; it may nevertheless briefly be mentioned that the view taken here is to consider the perfect as having the qualities of both a tense and an aspect, in the absence of any compelling reasons for adopting a definition of either of these categories so strict as to exclude the perfect necessarily. It should also be mentioned that some authors adopt alternative terms in place of ‘perfect’, in order to avoid any confusion that might arise with the term ‘perfective’, which is now generally used to refer to an aspectual feature unrelated to the perfect but is still sometimes used in its earlier adjectival meaning ‘pertaining to the perfect’ (see Comrie 1976, 61–4; Kortmann 1991, 16); for example, Bybee et al. (1994) use the term ‘anterior’. However, for 30 the purposes of this work, in which little reference is made to the perfective, the term ‘perfect’ is used throughout, and is substituted for alternative terms such as ‘anterior’ in discussions of works in which these are used. As mentioned above in Section 1.1.1, for the sake of brevity the term ‘tense’ may be used alone in this work to refer to the various forms of the perfect system which bear tense inflections, without any implication that these forms are purely temporal in their semantics. It should also be mentioned that the term ‘perfect’ can be used to describe either the perfect system as a whole and the semantic properties uniting its various members, or the particular form belonging to the perfect system in which the auxiliary is in the present tense, otherwise known as the ‘present perfect’; the latter term will be used here only where there is some danger of confusing these two usages. 2.2 Events 2.2.1 Introduction A description of the semantic content of verb forms must necessarily make reference to the types of entities to which verbs refer. Different authors have used different terminology to describe these entities, as will be seen in Section 2.2.2 below; for example, Reichenbach (1947) refers to the ‘situations’ denoted by verbs while Davidson (1967) distinguishes between verbs referring to ‘events’ and those referring to ‘states’. Although the following discussion of previous work on this subject will make use of the authors’ own terminology, elsewhere in the present work the term ‘event’ is used to denote the referent of any verb, without regard to distinctions of aspect or Aktionsart such as underlie classifications into 31 ‘events’ and ‘states’; where distinctions of this sort are relevant, they will be described as distinctions among different types of events. As discussed in Section 1.1.1 above, tense and aspect are morphosyntactic means of describing the temporal location and consistency of events; views regarding the semantic content of tense and aspect categories depend in part on conceptions of the nature of the events on which they operate. There is a substantial body of literature on the semantics of events, and an exhaustive survey of the work done on this topic would necessarily encompass much material not directly relevant to the present study; however, a review of some of the previous work in this area will provide an illustration of how different approaches to the semantics of events relate to issues regarding tense and aspect. To illustrate these questions, the theoretical models and systems of formal representation advocated by Reichenbach, Davidson, and Kim will be outlined, and the contributions of these differing theoretical positions to the present study will be discussed. 2.2.2 Events: Nature and Representation 2.2.2.1 Reichenbach Reichenbach’s Elements of Symbolic Logic (1947) addressed, among other topics, the representation of natural language within the framework of symbolic logic. As a prerequisite for the logical analysis of language, Reichenbach provided a formulation of the logical representation of events, including a method of representing tenses and describing these grammatical categories in semantic terms. In order to provide an integrated picture of Reichenbach’s semantic theory, it may be most useful to review these areas together, beginning with a discussion of his work on tense before addressing his interpretation of events. Reichenbach’s 32 analysis of the semantics of tense and aspect takes these categories to be expressions of the relative position of different points in time (1947, 287–98). In his system three points are defined: E, the time of the event; S, the time of speech, and R, the reference time; the differences among tenses are related to differences in the relative position of these points. The application of this system to the English tense system can be seen in the following table, in which the dash (—) indicates precedence of the point on the left over that on the right, and in which the comma (,) indicates simultaneity (adapted from Reichenbach 1947, 297): Structure Traditional Name E — R — S Past perfect E, R — S Simple past E — S, R Present perfect S, R, E Present S — E — R Future perfect S,E — R E — S — R S — R, E Simple future Table 1: Reichenbach’s Analysis of English Tenses It can be seen from Table 1 that in this system tense schemata fall into two groups, those in which the point R is identical to one of the other points and those in which it is separate. The latter type are ‘absolute tenses’ in the terminology of Comrie (1985), and relate the time of an event only to the time of utterance, without reference to any other point in time. It is for the ‘relative tenses’, the forms from the perfect system, that an additional point is necessary; this distinctive semantic property of the perfect system has long been noted (for bibliographical discussion see Klein 2010, 1222–3) and will be discussed in 33 greater detail below. 1 This system of analysis also provided foundations for incorporating additional distinctions of aspect; it was suggested that for progressives and imperfectives E could represent an extended span of time rather than an atomic point, and that for iteratives there might be multiple points E (Reichenbach 1947, 290). Although Reichenbach himself did not explore these possibilities in great detail, subsequent analyses of tense semantics that draw on his work have often expanded on this implicit potential (e.g. Huddleston 2002 for English; Curat 1991 for French). As part of the same model for the logical representation of natural language, Reichenbach also discussed the logical representation of events. The proposed approach used standard propositional and predicate logic as the basis for a system in which a ‘situation’, defined as the referent of a proposition of any kind, could be represented symbolically by means of predication (Reichenbach 1947, 251–74). In this system, sentences are analysed as describing a situation, about whose existence an assertion is made; thus the logical form of a sentence such as (7) below can be represented in natural language by (8) (from Reichenbach 1947, 270–1): (7) Amundsen flew to the North Pole in May 1926. (8) A flight by Amundsen to the North Pole in May 1926 took place. This analysis also allows for the possibility of multiple symbolic representations of a natural-language utterance, which may vary depending on which function is used, without any commitment to the determination by syntax alone of the 1 Although instances exist such as Reichenbach’s description of the simple future as being ambiguous between ‘S — R, E’ as shown above and the alternate representation ‘S, R — E’ (1947, 297), the perfect system is distinct in explicitly marking the temporal separation of R. 34 primacy of one such formal structure over another; thus, sentences (8), (9), and (10) are considered to be transformations of one another (adapted from Reichenbach 1947: 270–1): (9) A flight by Amundsen to the North Pole took place in May 1926. (10) A flight of Amundsen’s took place at the North Pole in May 1926. Although the full implications of this analysis are not explored in great depth, this approach suggests that these sentences are viewed as sharing a fundamental logical form, rather than merely being truth-functionally equivalent, and that they can be interchanged by relatively superficial operations. As the following sections will show, such an analysis of event semantics has implications which can be seen as undesirable. 2.2.2.2 Davidson A different approach to the semantics of events is taken by Davidson (1967; 1969), who builds on previous work in formal semantics, including that of Reichenbach, and undertakes to address some of the limitations inherent in previous systems for the symbolic representation of events. In contrast to the propositionally defined situations of Reichenbach’s model, events as conceived by Davidson are singular entities of the class to which actions belong (1967, 105–6). This approach was devised in order to address issues such as entailment; for example, in natural language, sentence (7) above entails (11): (11) Amundsen flew to the North Pole. In many previously employed systems for the formal representation of natural language, including that used by Reichenbach, these sentences are expressed as 35 predicates that differ in their number of places, as respectively in (12) and (13) below: (12) Flew(Amundsen, North Pole, May 1926) (13) Flew(Amundsen, North Pole) However, following Kenny (1963), Davidson observes that this mode of expression provides no formal means for showing the entailment of the latter by the former; moreover, attempts to overcome this difficulty by assuming that places for such modifiers as those for time are always present, either explicitly or implicitly, create the difficulty of having an indeterminate and potentially infinite number of such places in any given predicate (1967, 107–17). His approach to this issue is to interpret predicates representing verbs as containing one additional place for the event, a singular entity whose existence can then be asserted; it is then possible to predicate additional properties, such as spatial and temporal modifiers, separately of the event variable. The application of this approach to sentence (11) above can be seen in the following example (after Davidson 1967, 119; 126–7): (14) (∃x)(Flew(Amundsen, x) & To(North Pole, x)) Not only does this approach provide the basis for a formal system that more clearly reflects the inferences obtaining in natural language, but it allows a distinction to be drawn between those elements of a sentence which are essential to the action itself and those which are merely incidental adjuncts; in his discussion of this distinction Davidson does not attempt to provide an exhaustive set of criteria by which the two groups can be distinguished, adopting instead a relatively intuitive approach. A significant difference between Reichenbach’s approach and that of Davidson is that the former represents propositions by means 36 of a single predicate that is not readily decomposable, while in the latter propositions may be composed of multiple predicates that are linked by the occurrence in each of a singular event term; as a result, it is possible to make finer distinctions regarding the scope and semantic class of different modifiers (see Reichenbach 1947, 256–74; Davidson 1967, 115–9). It should be noted that Davidson’s original theory was formulated with reference to ‘events’ as a category distinct from, and in opposition to, that of ‘states’. The question of whether this distinction is essential to the validity of Davidson’s approach has been explored by subsequent authors (for discussion see Pianesi and Varzi 2000, 25–7), but as stated above in Section 2.2.1, it is held that for the purposes of the present study the maintenance of such a distinction is not generally necessary. Another issue with which Davidson’s work is concerned is the identity relations between events and the circumstances under which identity may be said to exist (see Davidson 1969, 163–4), an issue customarily described as the question of how ‘finely-grained’ events are. One aspect of this issue is addressed by the method of symbolizing event relations described above, in which the event is represented as an entity whose existence is asserted independently of the predication of other properties such as temporal modifiers; in example (14), therefore, the existence of the event is asserted separately from such predicates as ‘To(North Pole, x)’, and the presence or absence of the latter is not essential to any description of the event’s identity relations (see Davidson 1970, 185–7). A further issue in the determination of identity relations among events is the conflict between the formal necessity for the presence of singular terms in such expressions of identity and Davidson’s view that the events denoted by verb 37 phrases are not singular terms (see Davidson 1969, 164). One suggestion made by Davidson regarding this problem is that identity relations among events should be defined in terms of identity among their causal relationships, so that if all the causes and effects of event x and event y are identical, the events themselves may be said to be identical (1969, 179). However, in subsequent work he accepted Quine’s views regarding the circularity of such a criterion and endorsed an alternate approach, that the identity of events should be determined on the basis of the identity of their extent in space and time (see Quine 1985, 166; Davidson 1985, 175). A common example illustrating the latter approach is that of a sphere simultaneously rotating and heating up; the rotation and the heating would be considered identical because they occupy the same space-time location (Davidson 1969, 178–9; for further discussion see Pianesi and Varzi 2000, 18–22). 2.2.2.3 Kim Among the many theories on the semantics of events which differ from those discussed above, special mention may be made of the approach proposed by Kim (e.g. 1966; 1973). Theories of event quantification may broadly be classified in terms of the ‘thickness’ of events, the extent to which they are seen as resembling concrete entities (e.g. Pianesi and Varzi 2000, 5); whereas Davidson’s analysis may be seen as exemplifying ‘thick’ event quantification, Kim’s model exemplifies ‘thin’ quantification. Kim’s definition of events is based on the exemplification of properties by objects at a particular time (1973, 222); accordingly, in this theory the identity of events is dependent on the identity of the extensions of their properties (see Davidson 1969, 170). To simplify somewhat, an example of the implications of this approach is that the description of an action 38 as a stabbing must refer to a different event from the description of the same action as a killing, since the words ‘stabbing’ and ‘killing’ express different properties and there is no entailment of either by the other (see Kim 1973, 226– 36; also Pianesi and Varzi 2000, 9–13). Conversely, one case in which different verbs might be taken as referring to the same event is that of ‘giving’ and ‘receiving’, under the assumption that these two verbs differ only in the permutation of their arguments (Kim 1973, 225). Kim also addressed the issue of causality, in a way that employed a distinction between individual events and the ‘general events’ of which the individual events were instances; this distinction makes it possible to differentiate between those properties of an event which constitute it as such (for example, the properties that make a particular action a stabbing) and those which are merely contingent (for example, the location of a specific stabbing). According to this view, causality between individual events consisted in a law-like constant conjunction of the general events which the individual events instantiated (Kim 1973, 226–8). Kim’s theories have been subject to different interpretations by later authors. For example, it has been questioned to what extent the properties on which this definition of events depends must be determined by the semantics of the lexical items used; in other words, whether describing an event as a stabbing means that the event is defined by only those properties expressed by the word ‘stabbing’ or whether the event may be said to have other properties beyond those entailed by the term used (e.g. Bennett 1988; see further Pianesi and Varzi 2000, 10). While such variation in the elaborations of the different theoretical stances may in some cases tend to reduce the differences among them (see Bennett 1985), 39 in general it may be said that Kim’s model and Davidson’s stand at opposite poles in their approach to the determination of identity relations among events, the former admitting fewer cases of identity and the latter admitting more. 2.2.2.4 The Present Approach It is possible to view some of the dispute regarding ‘thick’ and ‘thin’ events as resulting from the use of the term ‘event’ in two distinct senses. When ‘event’ is used to refer to a phenomenon in the real world, something that happened at a particular place and time, a given event may be seen as having a large number of properties (to take the previous example, the event’s properties would include whether or not a knife was used, whether or not the consequences were fatal, etc.); considered from a strictly objective perspective, there would be few criteria for deciding which of these properties could be considered essential, and a ‘thick’ view of identity would be appropriate. However, in framing a linguistic representation of this phenomenon, it is necessary to mark some of these properties as more salient than others, in Kim’s terminology to choose a type of ‘general event’ to which to relate an ‘individual event’; in this way, a speaker might choose to class a real-world phenomenon that is both a stabbing and a killing either with other stabbings or other killings. Because the criteria determining membership in such sets vary among different sets, ‘thin’ identity criteria may be more appropriate for ‘events’ in the sense of referents of such linguistic forms. As stated in Section 2.2.1, in the present work the term ‘event’ is used in such a linguistically defined sense, and therefore a ‘thin’ approach would be more appropriate; however, it should be understood that this is not a 40 commitment to a specific position regarding the ontology of ‘events’ in the sense of real-world phenomena. From the foregoing discussion, the relevance of event identity to the present study may not be immediately apparent; the data analysed here consist of individual verb tokens, and the identity between the events denoted by different tokens is not directly relevant to questions regarding the distribution and development of the grammatical categories to which they belong. However, semantic issues such as those described above are often implicit in theories on the semantics of verbs and temporal categories. For example, Klein (2010) proposes an approach to tense and aspect based on a rich semantics in which verb phrases assert the existence and temporality of one or more logically and causally related events; according to this approach, a phrase such as to have felled a tree has semantic content making explicit reference to a state in which the tree is upright, a state in which it is fallen, an action in which an agent causes it to progress from one state to another, and a point in time at which the latter state obtains (see Klein 2010, 1225–42). For such an approach issues of the causal relationships among events, such as those discussed by Kim (1973), would need to be taken into account methodologically in the semantic analysis of tense and aspect categories. Conversely, the analysis adopted in the present work, which will be based more conservatively on a modified version of Reichenbach’s approach, avoids the need for such analysis to form part of the methodology used here. It is not the aim of the present work to provide a detailed critique of the relative merits of these semantic theories; rather, it is hoped that the descriptive accuracy of the semantic analysis discussed in the following sections will warrant its validity for the 41 purpose to which it is put. However, it is important to recognize the extent to which even simple questions of semantics may require a commitment to theoretical positions on a wide range of subjects. 2.3 The Perfect System in Modern English 2.3.1 The Perfect The semantic analysis of the perfect presents certain complications not found with other forms from the perfect system, such as the pluperfect; however, discussing these complications in connection with the form for which they are most at issue will make it clear the extent to which the same considerations may recur to a lesser extent for other categories. The perfect is a grammatical category that can be put to a wide variety of functions; among the functions commonly distinguished are those of the ‘continuative perfect’, ‘experiential perfect’, ‘resultative perfect’, and ‘perfect of present relevance’, which are illustrated respectively by the following examples (examples and categorization adapted from Huddleston 2002, 141–6): (15) She has lived in Berlin ever since she married. (16) His sister has been up Mont Blanc twice. (17) They’ve gone away. (18) I’ve discovered how to mend the fuse. It should be emphasized that this list of functions is not exhaustive and that for reasons that will be discussed below the above examples do not illustrate the full semantic range of the English perfect; nevertheless, classifications such as these provide a useful starting point for discussions of the semantics of the perfect. Analyses of the perfect have differed in the extent to which they treat these 42 categories as semantically heterogeneous domains that happen to be subsumed in English under a single grammatical form (e.g. Michaelis 1998; Iatridou et al. 2001) or as deriving from a single semantic principle and its interactions with other factors (e.g. Portner 2003). In the present discussion of these issues the latter position will generally be assumed; the extent to which such an approach explains or fails to explain the phenomena under discussion may be taken as indicating the extent to which such an approach is justified. The present perfect, like the simple present tense, relates the temporal position of an event to the present time; however, variation exists regarding the exact interpretation of this connection. One analysis of the relationship expressed by the perfect between the event time and the present can be seen in Reichenbach’s schematic representation of the perfect as ‘E — S, R’ (1947, 297), showing an event time E prior to a reference time R, which coincides with the moment of speech. It has sometimes been suggested that a definition of the perfect formulated in this way does not fully reflect continuative perfects such as (15) above, which refer to events beginning in the past but continuing into the present (for a review see Portner 2003, 460–2), and different approaches have been taken to deal with this issue. One possibility is to take a compositional approach to the semantics of the perfect, treating the temporal relations expressed by the auxiliary separately from those expressed by the participle; this approach will be discussed in more detail in Section 2.3.3 below, in connection with the non-finite forms of the perfect system and their semantics. Although Reichenbach himself did not explore the implications of this suggestion in any great depth, this approach has been pursued in greater detail by subsequent authors such as Moens 43 and Steedman (1988) and Portner (2003). Portner’s model incorporates the concept of the Extended Now, first proposed in its present form by McCoard (1978; for the further history of this concept see pp. 123–36), which reflects the fact that the present time to which the perfect makes reference is not a single point but rather a broader span of time, which includes the present moment and whose exact extent is determined by the pragmatic context; the use of such a concept allows not only E but R to represent an extended span of time, rather than a single point, so that the two can be said to have the possibility of overlapping, instead of merely being either simultaneous or separate (see Portner 2003, 481). Such a perspective seems promising in its ability to subsume the various types of perfect discussed above under a single notion of present relevance, without defining the latter concept in an excessively strict way: an event whose origins lie in the past can be relevant to the present in terms of its current persistence (continuative perfect), its contribution to the present sum of the subject’s experiences (experiential perfect), its causal relationship to a subsequent state (resultative perfect), or its connection to a discourse context assumed to be of current interest (perfect of present relevance), among others (see further Portner 2003, 459–61). However, providing a formal definition of the perfect which reflects the full breadth of this semantic range is a complicated task that has been approached in many different ways; phenomena often discussed in this context include the interaction of the perfect with other semantic properties, such as telicity, and the specific ways in which pragmatic context can influence the interpretation of perfect forms. 44 One criterion often used in defining subcategories of the perfect in English is that of continuativeness, the degree to which the event denoted by the verb can be interpreted as persisting at the moment of utterance. It has often been observed (see Portner 2003, 461–2) that continuativeness interacts with other semantic features such as Aktionsart; for example, a continuative interpretation is possible with stative constructions, such as (19), but not with eventive (i.e., non-stative) constructions, such as (20) (adapted from Portner 2003, 462): (19) Mary has understood the issue. (20) Mary has run a mile. In other words, in (20) it cannot be the case that Mary is still running the mile, while in (19) it would normally be the case that Mary still understands the issue. Moreover, it has been observed that where ambiguity exists, the non-continuative interpretation generally obtains, as in (21) as opposed to (22) (see Huddleston 2002, 141–2): (21) Mary has lived in Berlin. (22) Mary has lived in Berlin ever since she married. Such effects of Aktionsart are not restricted to the perfect; similar interactions have been observed with other tense forms in contexts such as indirect discourse and narrative sequences, as the following examples show (after Portner 2003, 481–2): (23) John said that Mary understood the issue. (24) John said that Mary ran a mile. In (23), the time of understanding may be taken to overlap with the time of saying, whereas the possibility of overlap is excluded in (24); moreover, although pluperfects could be substituted in both these sentences for the simple preterite 45 forms, the availability of continuative readings would be unaffected by such a change. Because interactions between continuativeness and Aktionsart are found in other contexts besides the perfect, Portner (2003) considers this interaction to belong not specifically to the definition of the perfect but to the broader domain of the sequence of tenses, an issue whose interaction with the perfect system will be examined in Section 2.5.2.2 below. The preference for non-continuative interpretations where ambiguity exists, such as in example (21), can then be explained in pragmatic terms; if the event persisted into the present, the present tense would be more informative in that it would indicate this explicitly, and therefore the avoidance of the present tense can create the implicature that the event is no longer happening (Portner 2003, 490–1). Another way of expressing the priority of non-continuative interpretations in such cases is to consider expressions of time as implicitly present when no overt specification has been made, resulting in default indefinite readings such as ‘for a certain period’; theories differ in the extent to which they attribute such default readings to syntax, semantics, or pragmatics (see Portner 2003, 491–3). In the present study, the pragmatic approach outlined above is preferred as being more parsimonious; given that language users have known pragmatic reasons for their interpretation of perfect sentences, there seems little reason to postulate the existence of entities such as a ‘phonologically null adverbial’ (Portner 2003, 492) to explain these interpretations. The interpretation of the perfect is influenced not only by Aktionsart- related properties, but by interaction with other grammatical categories such as the progressive. In English, the perfect and the progressive are two separate 46 categories which can be expressed independently of each other and which can thus interact. One effect of this interaction is that progressive perfects (and pluperfects) have a continuative reading, as can be seen from the following examples (adapted from Portner 2003): (25) Mary has been eating dinner. (26) Mary has been living in London. As a result of this continuative property of progressive perfects, they provide the normal means of expressing continuativeness in the perfect for eventive verbs, which do not normally have this interpretation in non-progressive forms; this can be seen in example (20) above (see Comrie 1976, 62). This similarity in interpretation between progressive forms generally and stative verbs has sometimes been ascribed to semantic similarities between the two categories (Portner 2003, 463); as with the continuative interpretation of the perfect of stative verbs, what influences the choice of grammatical form is that the beginning of the event precedes the present reference point, not whether the end of the event has taken place before this point (Comrie 1976, 62). It should also be noted that although the event denoted by a perfect progressive normally continues into the present, this is not invariably the case: (27) Mary has been eating dinner, but she’s just finished. A property that sentence pairs such as (25) and (27) have in common is that in neither case does the endpoint of the event of eating precede the Extended Now, as it does in the unmarked interpretation of (20) above; in (27) this endpoint coincides with the extended now, while in (25) it would normally be taken as subsequent. 47 The combination of the previously discussed semantic range of the perfect with the additional meaning that is provided by the inclusion of the progressive forms means that in languages such as Old English, which do not regularly distinguish either the perfect or the progressive, a simple past-tense form may correspond to any of six Modern English forms: simple past, past progressive, perfect, perfect progressive, pluperfect, or pluperfect progressive; accordingly, past-tense forms can be used to refer to events that were still taking place either at the time of utterance or at another reference point. Under such circumstances, the identification of those past-tense forms denoting events that in Modern English would need to be expressed using a perfect progressive raises the question of how many of these categories can validly be considered as potential cognitive entities for Old English speakers; the methodological implications of these formal disparities will be discussed in greater detail in Chapter 3. The use of pragmatic criteria such as those discussed above in definitions of the perfect raises questions regarding the pragmatic context such as how long a period of time can be considered as ‘present’ and what constitutes ‘relevance’ to this period. Analyses of the perfect such as that of Portner (2003), which view pragmatics as playing a prominent role in determining its signification, present the perfect system as differing from other tenses such as the simple past in the extent to which the explicit marking of pragmatic features forms an essential part of its meaning; according to such an analysis, the use of the present perfect automatically entails certain presuppositions. One such presupposition is that of present relevance, which Portner defines in modal terms such that the present perfect predicates the existence of some state which is of present relevance and 48 which is necessarily entailed by factors derived from the combination of the utterance itself and the world of discourse (Portner 2003, 496–502). According to this interpretation, sentence (17) above would be interpreted as asserting not only that the people in question went away but that there exists some situation, perhaps their current absence, which is of present relevance and to which their going away stands in a causal relation; similar analyses can be made for all the other examples of the perfect cited above, with a wide variety of hypothetical contexts possible for each. Another presupposition is that the event falls within the Extended Now, a property which can be related to the present tense of the auxiliary used. The establishment by the perfect of presuppositions such as these can then be used to explain the unacceptability of sentences that would violate them, such as the following (see Portner 2003, 464–98): (28) ??Gutenberg has invented the printing press. The unacceptability of this sentence has been taken as evidence that the content of the perfect is not purely temporal in nature and that the relationship expressed between the event and the Extended Now is more complex than the assertion of a salient result state (Portner 2003, 464). Consequently, sentences such as this are problematic for many theories of the perfect, such as the simple analysis by Reichenbach described above; although the event referred to may be of present relevance, it can scarcely be interpreted as falling within any period qualifying as the Extended Now. Conversely, there are some sentences whose acceptability is due not to the Extended Now, but to the presence or absence of pragmatic context (examples adapted from Portner 2003, 463–4): (29) Mary has lived in London for five years. 49 (30) Mary has been ill. (31) ?Mary has lived in London for five years. She has been ill. These examples were intended to be evaluated within a hypothetical context in which Mary has been living in London for the past five years, and during that time was ill only once, three years ago. In this scenario, the perfects in (29) and (30) are each more or less acceptable in isolation, but when they are combined into a connected discourse as in (31), they may seem pragmatically odd. However, as Portner suggests, this combination would become more acceptable in a context in which it is supposed that anyone of whom these things are true is now medically at risk. The effect that the existence of a pragmatically relevant present result can have on the acceptability of the perfect in contexts such as (31) indicates the extent of the role played by extralinguistic factors alone in the distributional patterns of the perfect; it should, however, be noted that examples such as this are more difficult to create than those such as (28), due to the greater variability of pragmatic factors and the tendency for the use of the perfect to create an implicature that some relevant context must exist whether or not it is known to the listener (see Portner 2003, 502–4). Nevertheless, the dependency of the perfect on context in determining the acceptability of sentences in which it is used, a dependency which is moreover independent of truth conditions (see e.g. Mittwoch 2008), is arguably greater than that of other grammatical forms and suggests that the degree to which the perfect interacts with pragmatic factors may be similarly greater. The prominent role played by pragmatic and extralinguistic factors in the meaning of the perfect complicates attempts to produce an exhaustive formal 50 definition of this tense. Portner’s view of the perfect is that there is a fundamental duality in its nature, that its truth-conditional semantics assert temporal relationships in a manner similar to Reichenbach’s model and that the acceptability of sentences involving the perfect is further constrained by the pragmatic factors outlined above. In order to formalize the pragmatics of the perfect, he makes use of concepts deriving from the possible-worlds approach used in some discussions of modality, such as the concept of ‘conversational backgrounds’, multiple sets of propositions which are selected according to the field of discourse and assumptions of shared knowledge; the number of possible conversational backgrounds is necessarily infinite (Portner 2003, 479–80). This approach is typical of many that have been found to deal with this component of the meaning of the perfect in that it regards some components of the meaning of the perfect as essentially ambiguous or vague, in a way that may not be resoluble through formal analysis (see Portner 2003, 488–9). As a result, formal definitions of the perfect can explain the acceptability or unacceptability of sentences occurring in the perfect, but are less suited to predicting whether a particular event will be represented by a perfect verb; although there are some contexts in which the representation of a particular event by the perfect or by another tense can be predicted with relative ease, there are many more for which judgements as to which of the various result states, if any, that might be produced by an event may be considered as relevant to a particular discourse context would be largely subjective. In contrast, choice between other grammatical categories, such as that between present and past tense, can generally be predicted with reference to a small number of easily measurable properties such as temporal location, and 51 sharper distinctions can be drawn between the typical uses which fit these criteria and the exceptional uses (such as historic presents) which do not. It should also be noted that other, more intuitive, criteria, such as pragmatic felicity, also form part of many theories of the perfect and continue to be used as tests to evaluate and compare the merits of different theories. As a result, the application of semantic work on the perfect to the analysis of data must be carried out, at least in part, on such an intuitive basis. As will be seen in Chapter 3, the lack of a formalized definition of the perfect with predictive power has important methodological implications for the study of perfect forms. 2.3.2 The Pluperfect In comparison with the perfect, as suggested above, the semantics of the pluperfect are in some ways less complicated; however, pragmatic factors also play a part in determining the interpretation of pluperfect forms. The semantics of the pluperfect tense are often defined in terms of the positioning of a past event prior to a reference point which is itself in the past (e.g. Comrie 1985, 64–6; Dahl 1985, 144–9); such a definition underlies Reichenbach’s representation of the pluperfect schematically as ‘E — R — S’ (1947, 297). The centrality of the reference point to these definitions places the temporal relations expressed by the pluperfect within the category of ‘relative tense’ discussed above. There are fewer restrictions on the circumstances under which pluperfects can occur than on those for present perfects; for example, pluperfects can co-occur with past-time adverbials, even those referring to times prior to R (example adapted from Portner 2003, 468): (32) On Tuesday I learned that Mary had arrived two days before. 52 However, defining the location of an event relative to a reference point raises the question of how such reference points are chosen and in what circumstances a point is eligible; in the case of the present perfect the reference point is the present time, or at least the Extended Now, but for the pluperfect selection of a reference point is determined pragmatically. This can be seen in the following examples: (33) John arrived at seven. Mary arrived at six. (34) John arrived at seven. Mary had arrived at six. These examples show that variation can exist as to whether a particular point or event should be taken as a reference point for the purposes of tense marking. Example (33), without the pluperfect, is an adequate linguistic representation of the events described by these sentences; although the inverted temporal sequence is somewhat unusual, some contexts, such as alphabetical ordering, would make even this perfectly felicitous. Example (34), in contrast, explicitly marks the latter event as anterior with reference to the former; this not only provides an additional indication of the temporal relationship between the events, but has the effect of linking the two events within a continuing discourse context (see Portner 2003, 484–8). The fact that both these sentences are acceptable and that any preference for one over the other would depend largely on context shows that the introduction of a particular point in time as a reference point R in the temporal semantics of the verb phrase is not determined solely by the nature and position of the events themselves or by grammatical considerations. In some cases the effects of context can be weak enough to produce something close to free variation, as the following examples show (adapted from Visser 1966, II, 757–60): (35) John was surprised that he broke the window. (36) John was surprised that he had broken the window. 53 In sentences such as these, the context serves to determine the temporal relations of the events sufficiently to make any overt grammatical marking superfluous (e.g. Huddleston 2002, 141–6); the optionality of such marking, which was a characteristic feature of Old English (see Mitchell 1975, 159–66), has persisted to a certain extent into Modern English. In some contexts, however, such variation may not be entirely meaningless, but may instead result in slightly different interpretations. An example of such variation can be seen more clearly from the following examples (adapted from Comrie 1976, 56): (37) Bill had arrived at six o’clock. (38) At six o’clock, Bill had arrived. The sentence in (37) is ambiguous; it can be interpreted as meaning that six o’clock was the time of Bill’s arrival, prior to some other unspecified point in the past, or it can have the meaning expressed unambiguously by (38), that at six o’clock it was already the case that Bill had arrived previously. Such ambiguities of scope are not restricted to the pluperfect; similar phenomena have been described in the case of the perfect (e.g. Iatridou et al. 2001; Portner 2003, 490–2). An ambiguity parallel to that found in (37) can also be seen in (36); the sentence can mean that John was surprised either by the preceding event of breaking the window or by the ongoing state of having broken it, while (35) can have only the former meaning. 2 From a methodological perspective, however, even where such nice semantic distinctions exist it is rarely possibly to identify them consistently in historical texts, especially given the role of pragmatic context; if a construction is 2 From a diachronic perspective, these two readings of (36) may not always have been present to the same extent; the event reading would presumably occur only after the pluperfect had evolved beyond its original stative meaning. 54 known to have a specific meaning in the context in which it is found, it can be difficult to determine whether the same form might have had a different meaning in a different context. As with the perfect, the similarity between the pluperfect and the preterite in terms of their truth conditions makes it difficult to determine the extent to which these may be perceived as separate semantic categories in the absence of explicit marking. 2.3.3 Non-Finite Perfect Forms The perfect system includes non-finite forms in addition to finite tenses such as the perfect and pluperfect. In Modern English non-finite perfect forms comprise the perfect infinitive (to have done) and the perfect participle (having done), to which may be added the past participle (done) as a form in its own right. The first two categories do not play a prominent role in the present study, being marginal at best in Old English (Mitchell 1985, I, 388; 413). In contrast, the past participle, which itself predates the periphrastic forms, can be used alone in constructions that possess the notion of temporal anteriority which is common to all members of the perfect system, both through its adjectival use and in absolute constructions. Absolute participial constructions are already found in Old English texts from an early period; although there has been some dispute regarding the extent to which such constructions are Latinisms, it seems probable that their use to render Latin absolute constructions is the result of a pre-existing semantic and syntactic compatibility between the two languages (for discussion see Mitchell 1985, II, 926–30). The following example and its translation illustrate the form and meaning of absolute participle constructions in Old and Modern English: 55 (39) Þa him swa gecigdum þær wæron then he-DAT so call-PA.PPL-MASC.DAT.SG there be-3PL.PRET eac oþre VII broðru be naman also other-ACC.PL seven brother-ACC.PL by name-DAT gecigde. call-PA.PPL-NOM.PL (GD MS C I.21.52.25) ‘Then, he having been summoned thus, there were also seven other brothers summoned by name.’ Participial constructions such as that in the example above illustrate one of the most salient semantic traits of non-finite verbal forms, the expression of relative time. It may be seen in this example that the past participle gecigdum ‘called’, occurring by itself in a dative absolute construction in the absence of any auxiliary marked for tense, is sufficient to express the temporal relationship between the two events of summoning, and that this relationship is the same as would be expressed if a pluperfect (after he had been summoned) had been used. The expression of such anterior meaning by the past participle and other non-finite perfect forms is parallel to the expression of simultaneity denoted by the present participle and related progressive forms; such a use of the present participle can be seen in the following Old English example: (40) Þa him gebiddendum seo then he-DAT pray-PRES.PPL-MASC.DAT.SG the-FEM.NOM.SG sawl þæs cnihtes gehwearf eft soul-NOM the-MASC.GEN.SG knight-GEN turn-3SG.PRET back to þam lichaman. to the-MASC.DAT.SG body (GD MS C I.21.52.25) ‘Then, with him praying, the soul of the boy returned to the body.’ As suggested above in Section 2.1, relative temporality of the sort expressed by these participles is characteristic of non-finite forms that are marked for aspect, supporting the notion of some semantic affinity between the perfect and aspectual 56 categories. Similar distinctions in meaning between finite and non-finite forms are found in other languages in which the categories involved are undoubtedly aspectual in nature; for example, in Classical Greek the distinction between the present stem and the aorist stem is fundamentally the opposition between imperfectivity and perfectivity, but the participles of these stems can additionally express the relative temporal relationships of simultaneity and anteriority respectively (e.g. Goodwin 1881, 275–6).3 The possibility of expressing relative temporality by means of the past participle alone might suggest that the tense marking of the auxiliary in perfect constructions is what produces the absolute time reference that makes it possible to relate the time of the event denoted by the perfect construction to the time of utterance; in the next section different theories regarding the extent to which the perfect is compositional in this way will be discussed. 2.3.4 Compositionality of the Perfect Compositional semantic analyses of the perfect have formed part of some proposals for the formal representation of tense relationships, especially those constructed along Reichenbachian lines. The system adopted by Huddleston (2002, 125–41) draws a distinction between ‘deictic’ and ‘non-deictic’ tenses; the former tenses, excluding the perfect system, relate the time of the event directly to the time of utterance, while the latter tenses, comprising the perfect system, make use of an extra set of Reichenbachian-type points; for example, the pluperfect is described as locating the time of the event prior to an ‘orientation point’, and it is 3 For a discussion of the semantic and pragmatic relationships between perfectivity and anteriority see Comrie (1976) 57 the use of a preterite form of the auxiliary which specifies that this orientation point is identical to a reference point located in the past, prior to the moment of speaking. Such compositional analyses have even been applied to formally similar constructions that differ semantically from the English perfect, such as the French passé composé; for example, Curat (1991, 239–263), whose adaptation of a Reichenbachian framework reflects the work of Guillaume (e.g. 1929), analyses the role of the past participle in this construction as making explicit reference to the endpoint of the event, as opposed to its beginning or any intermediate points, while the auxiliary is seen as locating this endpoint relative to the moment of speech. A slightly different approach is taken by Portner (2003); his analysis follows Reichenbach (1947) in representing the perfect with only a single set of points E, R, and S, and deals with the greater semantic complexities of the perfect by allowing these three points not only to occupy separate positions but also to represent overlapping spans. Nevertheless, his analysis identifies specific semantic and pragmatic contributions of the auxiliary, to which is ascribed the origin not only of absolute temporal reference but of the Extended Now restriction in the present perfect, whose pragmatic nature may vary cross-linguistically (2003, 495–6). Yet another type of compositionality is proposed by Klein (2010), who draws a distinction between the absolute temporal reference that is due to the finite inflection of the auxiliary and the relative temporal relationships that result from the combination of the auxiliary and the participle. However, not all analyses of the perfect view it as compositional; for example, McFadden and Alexiadou (2010) interpret the periphrastic perfect as simply a spelling out of a single verb with an abstract ‘perfect’ feature. While it is not the goal of the 58 present study to adjudicate among the various analyses that have been proposed, it might be observed that a non-compositional semantic analysis of the perfect requires commitment to the view that the seemingly transparent formal compositionality of the perfect is ignored in the acquisition of the language and in the comprehension of its semantic content. The question might therefore be raised of whether any semantic compositionality of the perfect is determined by the use of a periphrastic grammatical form, so that each part of the periphrasis retains some vestige of a distinct meaning, or whether this compositionality is an abstract property of the semantic domain with which these forms happen to be associated, so that semantic compositionality and formal compositionality might be independent phenomena with separate origins. Dahl (1985, 129) found that from a cross-linguistic perspective perfect meaning is significantly, but not universally, correlated with periphrastic form; this might suggest the existence of a degree of semantic complexity, which of course need not be represented by means of an equally complex grammatical form. Regardless of the extent to which the perfect is semantically compositional, it is nevertheless possible to treat the perfect system as a single, unitary category; the justification for doing so will be discussed in the following sections. 2.4 The Perfect System Cross-Linguistically 2.4.1 The Perfect The foregoing discussion of the semantics of the Modern English perfect would have little relevance to the study of perfects in Old English and Old Saxon 59 unless there were a cross-linguistically valid semantic category which was manifested by formally similar means in these different languages. Evidence for the existence of such a cross-linguistic category was provided by Dahl (1985); starting from a division of the functions of the perfect into resultative uses, experiential uses, continuative uses, and recent-past uses, similar to the classification illustrated by Examples (15)– (18) in Section 2.3.1 above, he found evidence of grammatical categories that combine these functions similarly to the English perfect in at least 24 languages (1985, 129–33). Despite a certain degree of functional overlap, it was possible to distinguish perfects semantically from pure resultatives and experiential markers (Dahl 1985, 133–44; see also Section 1.2.3.1 above). The perfect categories in the different languages are not all identical in their distributional patterns; for example, although the English perfect cannot occur with definite past-time reference, perfect constructions of this sort can be used in restricted contexts in Swedish and more generally in Bulgarian (Dahl 1985, 137–8). This variation has been interpreted in various ways; Dahl proposes that different languages may adhere to a single Reichenbachian schema with varying degrees of strictness, while Lindstedt (2000, 369–71) suggests that the variation may reflect different degrees of association between the perfect and pure evidentiality as a semantic category, and Portner (2003, 495–8) suggests that languages may differ in the extent to which their present tenses impose a pragmatic Extended Now requirement. If this cross-linguistic variation is fundamentally pragmatic, regardless of the precise nature of the differences involved, then given the prominent role of pragmatics in the content of the perfect, as discussed in Section 2.3.1 above, this would provide a way of reconciling the 60 observed degree of cross-linguistic variation with a view of the perfect as a recurring semantic category. Further support for the notion of the perfect as a meaningful cross-linguistic category comes from diachronic research. Bybee et al. (1994, 63–105) found that the perfects have developed in many unrelated languages through numerous paths of grammaticalization, including that described previously in Section 1.2; additionally, the perfects has often been an earlier stage in the development of categories that now have another meaning, such as the perfective sense seen in the French passé composé. There is a lack of consensus regarding the extent to which categories that have moved away from the prototypical perfect usage retain ties to the perfect system; for example, Lindstedt (2000) sees periphrastic past tenses of this sort as semantically divorced from the perfect, while Curat (1991) and Klein (2010) propose analyses for French and German respectively that suggest underlying semantic differences between the periphrastic and simple tenses, related to the compositionality discussed in Section 2.3.4 above, although these differences may be almost neutralized pragmatically so that the different forms can be used in similar contexts. Grammatical forms of this sort may make it easier to identify the centre of the perfect as a semantic category than to delineate its precise boundaries. 2.4.2 The Pluperfect Less attention has been paid to the pluperfect as a cross-linguistic category than the perfect. Pluperfects are usually formed using the past tense of the auxiliary used in the present to form the perfect, or a tense diachronically descended from the perfect; this is the case for virtually all the pluperfects identified by Dahl (1985, 144–5). The persistence of the pluperfect even in the 61 absence of a prototypical perfect may suggest a degree of semantic independence between the two categories; alternatively, a semantic duality has been ascribed to the pluperfect in languages such as English, representing as it does an anterior shifting of both the past and the perfect (see Comrie 1976, 56), and it may be only the former sense that survives the loss of the perfect. The degree of formal interrelationship between the perfect and the pluperfect may make it difficult to determine the extent to which the semantic content of the pluperfect can be seen as a unitary, cross-linguistically valid semantic category independently of the perfect, and questions of this sort may also be raised by the optionality of pluperfect marking even in a language such as Modern English in which the pluperfect exists as a distinct formal category. However, the approach of the present study is to treat the pluperfect as a category in its own right, permitting its development to be studied independently from that of the perfect; it is hoped that this approach is borne out by the significantly different distributional patterns of the two categories, which are discussed in Chapter 4. 2.5 The Perfect System in Old English and Old Saxon 2.5.1 Periphrastic Constructions The existence of periphrastic constructions in Old English and Old Saxon which are formally identical to the Modern English perfect and pluperfect has been described in Section 1.3 above. The task of discriminating between those constructions of this sort which retain their original stative meaning and those which represent grammaticalized perfect forms is a complex issue, as is the extent to which the distribution of these two groups may vary diachronically; these 62 topics will be discussed in detail in Chapters 3 and 4. For the present, it is hoped that the expression by these constructions of content belonging to the domains of the perfect and pluperfect will be illustrated sufficiently by the following examples: (41) Ure Drihten […] wile þonne our-MASC.NOM.SG Lord-NOM will-3SG.PRES then witan […] hu we urne know-INF how we-NOM our-MASC.ACC.SG cristendom gehealden habban. Christendom-ACC hold-PA.PPL have-3PL.PRES (Wulf II.121.69) ‘Our Lord will then know how we have kept our Christianity.’ (42) Se feond […] þe on þa the-MASC.NOM.SG fiend-NOM REL on the-FEM.ACC.SG frecnan fyrde gefaren hæfde bold-FEM.ACC.SG campaign-ACC fare-PA.PPL have-3SG.PRET (GenB 13.225.688, cited by Mitchell 1985, I, 289) ‘The fiend that had gone on that terrible campaign’ (43) Sô hue sô iu than antfâhit[…], sô so who-NOM so you-ACC then receive-3SG.PRES so haƀad mînan forð uuilleon have-3SG.PRES my-ACC forth will-ACC geuuarhten[…]. work-PA.PPL-MASC.ACC.SG (Heliand XXIII.75.1957) ‘Whoever then receives you has thus carried out My will.’ (44) Thea liudi forstôdun, that he the-NOM.PL people-NOM understand-3PL.PRET that he-NOM thar habda […] godcundes huat there have-3SG.PRET godly-NEUT.GEN.SG what-ACC forsehen[…]. see-PA.PPL (Heliand III.13.187) ‘The people understood that he had seen something there from God.’ 63 These examples, the first two from Old English and the second two from Old Saxon, show the availability of periphrastic constructions as expressions of perfect and pluperfect meaning. In this instance it is relatively easy to identify the association between formal and semantic categories; the only other sense that periphrastic constructions of this type can have, the stative sense, is sufficiently different in its semantics from the perfect and the pluperfect that in some contexts at least the two are readily distinguishable. 2.5.2 The Simple Preterite As discussed in Chapter 1, prior to the development of the periphrastic forms the simple preterite was the only means available in the Germanic languages for referring to past events, and even after the introduction of these forms it continued to be used in contexts where Modern English would require a perfect or a pluperfect. It may be asked whether this breadth of usage indicates a polysemous formal category whose range of meaning encompasses a number of distinct semantic categories or else a monosemous formal category whose semantic content is so underspecified as to be applicable in a wide variety of contexts. Even if the latter alternative were the case, the identification of certain occurrences of the preterite as equivalent to the perfect or the pluperfect would still be possible from a functional perspective, but it might be argued that the many contexts in which a given event can be described with equal truth and felicity by a past tense or a perfect, the two forms differing only in their pragmatic connotations, would make such a functional distinction of limited validity. However, there is a certain amount of evidence for the existence of semantic differences between perfect-type preterites and other preterites. The following 64 discussion will deal primarily with the Old English preterite, as this has received greater attention in the literature than its Old Saxon counterpart, but the criteria discussed may be taken as usable for both languages. 2.5.2.1 Adverbial Collocations In Old English the simple preterite could occur with adverbs such as nu ‘now’ in contexts where the adverb had a temporal sense, which moreover clearly referred to the moment of speaking rather than to another time taken as a deictic centre (see Mitchell 1985, I, 246–7). The use of nu with the preterite can be seen in examples such as the following: (45) Ic nu gyta ne geopenode minne I-NOM now yet NEG open-3SG.PRET my-MASC.ACC.SG muþ to Godes lofe[…]. mouth-ACC to God-GEN praise-DAT (GD MS H I.XXIII.62.20) ‘I now have not yet opened my mouth in praise of God.’4 (46) *I now did not yet open my mouth in praise of God. In contrast to Modern English sentences such as (46), in which now can only have the meaning ‘at that time’ rather than ‘at the present moment’, (45) illustrates the use in Old English of nu to refer to the time of speaking while modifying a preterite verb. The use of now with a perfect verb is quite easy to explain; in Reichenbachian terms it makes explicit the position of the reference point R, which in the perfect is situated at the moment of speaking. If the preterite in examples such as (45) had only the ‘absolute tense’ semantic structure shown for the past tense in Table 1 above, it would be difficult to understand the reference of 4 See the discussion of verbal prefixes in Section 3.3.2.5. 65 nu; it could not refer to the time of the event, which is in the past, 5 and it would be redundant as a description of the moment of speaking. In contrast, if such sentences involve a notion that the present is not only the moment of speaking but is also relevant in another sense, this would seem to suggest semantic content closer to the perfect than to the past. 2.5.2.2 Sequence of Tenses The term ‘sequence of tenses’ refers to a phenomenon in which the temporal location of events and the previous use of specific tenses in the discourse interact to determine the tense form used in a given context to express the temporal location of the event to which reference is made; the existence of such phenomena also affects the temporal relations that can be interpreted as existing given the use of a specific tense form in a particular context. The term ‘sequence of tenses’ has its ultimate origins in classical grammar; in grammatical descriptions of the classical languages, a twofold distinction is often made between ‘primary tenses’ and ‘secondary tenses’, which can be broadly described as non-past and past, respectively; in the case of Greek the former set includes tenses such as the present, perfect, and future, while the latter set includes the imperfect, pluperfect and aorist (e.g. Goodwin 1881, 271–2). This primary/secondary distinction can be applied to English examples such as the following: (47) He says that he has seen her. (48) He said that he had seen her. 5 It is assumed here that negative sentences are identical in their temporal semantics to the corresponding positive sentences. 66 (49) He thinks that he will come. (50) He thought that he would come. In these examples the first member of each pair shows a primary sequence, while the second member shows the corresponding secondary sequence; it can be seen that the events described stand in the same temporal relationship to each other in each case, and that the tense of the second verb depends on that of the first. From (47) it can be seen that the perfect is a primary tense in Modern English as it was in Classical Greek. The simple primary/secondary dichotomy described above is adequate to describe much of the observed distribution of the different tenses; however, there are certain phenomena that this distinction is less equipped to address, both in the classical languages and elsewhere (see Gildersleeve and Lodge 1885, 314–24). One common example is the possibility of combining secondary and primary tenses, as seen in the following example (adapted from Abusch 1997, 40): (51) John said two weeks ago that Mary is pregnant. Sentences such as this are not merely reflections of the temporal relationship between the two events as it was reflected in John’s original utterance, but convey the additional information that the state of pregnancy continues at the present time. Certain sequences of tenses can also have the converse property of collapsing distinctions that are otherwise formally separated, as can be seen in the following example (from Ogihara 1995, 668): (52) John said that Mary was sick. Examples such as this can have either a simultaneous reading (‘John said, “Mary is sick.”’) or a ‘shifted’ reading (‘John said, “Mary was sick.”’); a similar ambiguity can be seen in (23) above. Various theories have been proposed to 67 account for tense-sequence effects of this more complex sort, such as the suggestion that the past-tense morphology in the subordinate clause is semantically null (Ogihara 1995) or that it can lose its absolute temporal reference and express temporality relative only to the time of the main verb (Abusch 1997). Within the framework of the present study, it would be possible to view the ambiguity of examples such as (52) in terms of the optionality of giving morphosyntactic expression to the temporal anteriority normally associated with the pluperfect, a form of variation which was described above as existing in certain contexts for English; in other words, this ambiguity is the result of free variation between past-tense sentences such as (52) and pluperfect sentences such as John said that Mary had been sick. However, such a tentative hypothesis as this is not intended as an exhaustive explanation of all tense-sequence phenomena in English (for further discussion and bibliography see e.g. Giorgi 2009, 1838– 40). Although it is apparent that the division of tenses into primary and secondary tenses is not a full explanation of all the observed phenomena related to the sequence of tense, the labels ‘primary’ and ‘secondary’ will be retained in this work to make reference to a semantic distinction that is valid in itself. It is possible to use the sequence of tenses as a diagnostic for semantic differences that may not be reflected morphologically. In Latin a single grammatical category, known as the perfect, could express both perfect and perfective past meaning (e.g. Gildersleeve and Lodge 1885, 159–60); however, these two senses could be distinguished by their distribution, the former occurring in primary sequence and the latter in secondary sequence (e.g. Gildersleeve and 68 Lodge 1885, 314–6). This difference can be illustrated by the following examples: (53) Dixit quid eventurum sit. say-3SG.PF what-NOM happen-FUT.PPL-NEUT.NOM be-3SG.PRES.SUBJ ‘He has said what will happen.’ (54) Dixit quid eventurum esset. say-3SG.PF what-NOM happen-FUT.PPL-NEUT.NOM be-3SG.IMPF.SUBJ ‘He said what would happen.’ In this way, the sequence of tenses in which a verb occurs can provide cues for the resolution of morphological ambiguities, if not absolutely then at least with regard to the default, unmarked interpretation. There is some evidence that the Old English preterite displayed a similar duality, functioning both as a primary and as a secondary tense; although stressing the non-deterministic nature of tense sequencing and suggesting that some proposed examples of this phenomenon are actually similar to (51) above, Mitchell (1985, I, 360–2) discusses cases in which the occurrence of the Old English preterite in primary sequence is apparently due to its being perceived semantically as a perfect. The existence of two different types of behaviour in the sequence of tenses would then be suggestive of a corresponding semantic duality. 2.5.2.3 Translation Practice Although the practices of Old English translators in finding vernacular equivalents for Latin verb forms will be discussed in detail in the following two chapters, mention may be made at this point of the use of the Old English preterite to render the Latin present tense. A translation of this type can be seen in the following example: 69 (55) Mirum est valde quod wonderful-NEUT.NOM.SG be-3SG.PRES strongly what-ACC audio […]. hear-1SG.PRES (GD I.V.3, p. 100) ‘What I hear is very wonderful.’ Hit is swiþe wundorlic, þæt it-NOM be-3SG.PRES strongly wonderful that-NEUT.ACC.SG ic nu gehyrde æt þe. I-NOM now hear-1SG.PRET at thou-DAT (GD MS C I.XV.44.23) ‘That which I have now heard from you is very wonderful.’6 In the original text, the use of the present is intended to convey that the events of hearing that have already occurred form part of a continuing sequence of such events. Whether or not it is interpreted as a perfect, the use of the preterite in Old English represents a shift of emphasis to those events which have already occurred; however, if the preterite did not have some sense in which the present continued to function as a reference point, such a translation would seem to represent an inexplicably great deviation from the original text. 2.5.2.4 The Preterite and the Perfect Taking such translations together with the other forms of evidence discussed above, the hypothesis that the perfect existed as a distinct and coherent semantic domain represented not only by the periphrastic constructions but by certain preterites seems tenable. A well-known phenomenon is the possibility in Old English of using the preterite in the same sentence and with the same temporal reference as a periphrastic perfect or pluperfect (e.g. Mitchell 1985, I, 246). It has been suggested, based on an assumption of semantic unity for the 6 See also the discussion of verbal prefixes in Section 3.3.2.5. 70 preterite, that this interchangeability means that the periphrastic constructions themselves lacked the semantic properties of the perfect and were themselves preterites (e.g. Denison 1993, 352). If, however, the preterite has a greater degree of semantic complexity than such an analysis would assume, it is possible to explain the apparent equivalence of the two forms in certain contexts without assuming complete semantic identity. 2.6 Conclusion It is hoped that the foregoing discussion has made clear the semantic concepts that underlie the present course of research, which should provide a foundation for the methodology employed here. A semantic definition has been provided for the perfect, as a category that specifies not only the temporal relations of events but their pragmatic relevance to the present, and for the pluperfect, as a category that situates events prior to a pragmatically established reference point in the past; these semantic categories are represented by grammatical categories not only in Modern English but in other languages. The diagnostic criteria described above for interpreting the semantic nature of the perfect and for identifying grammatical forms as the expression of this semantic category should illustrate some of the issues that need to be taken into account in the methodology of the present study, such as the need to be sensitive to relatively subtle semantic distinctions. The next chapter will show the specific ways in which these methodological requirements are put into practice. 71 3. Methodology 3.1 Introduction In order to investigate the relationship between the periphrastic forms and the preterite as methods of expressing similar semantic content, it is necessary to ensure that the methodological decisions made not only reflect the semantic issues discussed in the previous chapter but also take into account other issues influencing the validity of the results. These issues include the selection of an appropriate and representative body of texts to be used as a data source, the formulation of criteria for identifying relevant tokens of the preterite and of the periphrastic constructions, and the identification of additional variables to be analysed, such as discourse context and translation usage, as well as the choice and definition of values for these variables. The decision-making process involved in the development of a particular methodology is dependent on an understanding of the relevant theoretical issues, including the semantic topics addressed in Chapter 2, for the conceptual foundations that any discussion of the relative merits of different sets of criteria must have. It is hoped that the following discussion will show the methodology adopted to be sufficiently robust to validate the conclusions drawn from research employing it; the results of this research will be discussed in the next chapter. 72 3.2 Data Sources 3.2.1 Introduction One of the primary considerations in the selection of texts for analysis is to provide a balanced and representative sample of the works in each language, a sample which is nevertheless of manageable size. In choosing texts for the present study, an effort was made to represent different periods and to provide as wide a stylistic range as possible from the data available, as discussed in the following section. Stylistic factors can interact with the tense-sequence phenomena discussed in Section 2.5.2.2 above; for example, historical narratives have a relatively large proportion of secondary tenses, while homiletic material abounds in primary tenses. Many texts are divided between passages of narrative, which is normally construed as reporting events, real or fictional, that have taken place before the time of writing (S in Reichenbachian terms), and dialogue, which provides direct-discourse reports of utterances made prior to the time of writing but nevertheless takes as a temporal reference point S the time of the original utterance rather than the time of writing (see e.g. Fleischman 1990, 52–63). This variation in the temporal relationships most likely to be expressed within different types of discourse gives rise to corresponding variation in the verb forms used to denote these relationships, which in turn is reflected in the distribution of the periphrastic tenses. The present perfect, being a primary tense, is more likely to occur within dialogue, to denote events prior to those referred to within the same dialogue by the present tense, and sharing with them the same Extended Now; conversely, the pluperfect is more likely to be used in secondary sequences within narrative, non-dialogue sections, referring to events prior to the past events 73 described in the narrative. It should be observed, however, that this correlation is not necessarily absolute. In more complicated passages of dialogue, the pluperfect may be used to make reference to multiple strata of past events, while in some forms of narrative the present perfect may be used, when reference is made not only to completed past events but to results of these events which persist at the time of writing. The introduction of such present perfects into narratives of past events may be especially felicitous in non-fictional texts, in which the events related may often have some connection to the real present of the author (see Fleischman 1990, 30–1); accordingly, the distribution of particular verb forms may exhibit some variation according to genre. Variation of this sort may be said to add difficulty to intertextual comparison; however, most texts contain at least some tokens from each group, and some texts provide numerous examples of both. The following sections will provide a brief overview of the texts analysed in the present study; various aspects of their history will be discussed in greater detail in Chapter 4 in connection with the data drawn from them. 3.2.2 Old Saxon In the case of Old Saxon, the selection process presents few complications, given that there are only two texts of any significant length; the unrepresentative nature of the sample, while regrettable, is unavoidable, and the question of which texts to exclude scarcely arises. The longest text in Old Saxon, the gospel retelling known as the Heliand, consists of 5983 lines divided into 71 fits, of which 36 have been analysed here; it was decided that this sample was sufficiently representative of the text as a whole that to broaden it would be unlikely to have any significant effect on the data. The text analysed was taken from the Behaghel 74 and Taeger edition (1996). The other principal Old Saxon text, the fragmentary poem Genesis, provides a valuable point of comparison between Old Saxon and Old English through the existence of its Old English translation, Genesis B, despite the limited overlap between their surviving portions; both poems are analysed in their entirety, giving samples of 337 and 851 lines respectively. For both Genesis poems, Doane’s edition (1991) was used. Both the principal Old Saxon texts are thought to have been written in the first half of the ninth century, and are known from manuscripts seldom more than half a century later (e.g. Derolez 1995). Apart from these poems, the only attestations of Old Saxon are texts such as very short inscriptions and charms and taxation lists (see e.g. Rauch 1992, 1; 251–2); such texts, both from their brevity and from their content, provide almost no data on the use of perfect tenses, and they have not been considered in the present study. 3.2.3 Old English For Old English a greater body of material is available, and therefore greater care in the selection of texts is required to achieve a balanced sample. With the exception of Genesis B, the Old English texts included in the present study are prose texts; the syntax of poetic texts is influenced by additional factors such as metre which are not at issue in prose, and the possibility of consistent register differences between poetry and prose texts adds an additional complication to the comparison of different texts. One text which provides an important resource for a study such as this is the Anglo-Saxon Chronicle, which provides a record of vernacular prose composition over an exceptionally broad time span, from the end of the ninth century to the middle of the twelfth. The two 75 longest manuscripts of the Chronicle are both included in the present study (MS A, the Parker manuscript, and MS E, the Peterborough manuscript); the data used in the present study are drawn from the excerpts in the Helsinki Corpus (Rissanen et al. 1996), which includes MS A in its entirety and MS E for the years 966– 1048, 1070–1087, and 1105–1121, excluding the verse sections from both manuscripts. As different sections of the Chronicle were written at different times, for the purposes of the present study the texts have been subdivided chronologically in accordance with the textual discussions in Plummer (1889, xxiv–xxxv), Whitelock (1965, xi–xviii), Bately (1986, xxi–xlvi), and Irvine (2004, xviii–xxiii), each subdivision being treated here as a separate text. For the purposes of analysis, MS A has been divided into four sections: the first ends with the annal for 891, and the second spans the years 892–923; the third section, for the years 924–956, includes no occurrences of relevant verb forms and is excluded from analysis; the fourth section is from 958–1001. While it might be possible to make finer chronological distinctions within these sections, this would have the undesirable effects of reducing the number of samples within each section and making meaningful comparisons more difficult. In the Helsinki Corpus, MS A is separated into two texts, ChronA1 and ChronA2, divided at the year 950, which is taken as the approximate midpoint of the Old English period; although this does not correspond to a natural textual division of the Chronicle, the absence of any data from the section split in this way obviates the need for any adjustment. MS E is treated as a single section from 966–1121. From the available evidence, the dates for the sections of the Chronicle may be most plausibly set not long after the final year that they describe, although such datings refer only to the final form of 76 the Chronicle that is known in manuscript; the possibility cannot be excluded that other, previously written texts may have been used in the compilation of the Chronicle, although nothing about such original sources is known that would allow this section of the Chronicle to be split into chronologically valid subsections for the present purposes (see further Bately 1978). Unlike the Chronicle, the other Old English texts analysed here each represent only a single period. One of the earlier texts analysed here is the Old English translation of Bede’s History. Although the earliest manuscript of this text, hereafter termed Bede in accordance with the abbreviations set out in the list of references, does not predate the tenth century, the translation itself is thought to have been composed in the late ninth century; unlike the other texts included in the present study, which are predominantly West Saxon, Bede shows a significant degree of Mercian influence (Miller 1890, xxxiii; Whitelock 1962, 57–9). The starting point for the analysis of Bede in the present study was the excerpts in the Helsinki Corpus. However, these excerpts were found to include no tokens of the periphrastic present perfect, which occurs elsewhere in the text; this deficit was remedied by the inclusion of two additional sections from the second book (9–10) beyond the end of the Helsinki sample. The sample was further enlarged to facilitate meaningful statistical analysis; the total sample comprises 1312 lines of text. Because of the need for additional sections, the edition of Miller (1890) was used as the primary source throughout for consistency. The translation of the Cura Pastoralis also dates from the end of the ninth century, as does the earliest known manuscript of this text (see Sweet 1871); as in the case of Bede, an enlarged sample including the Helsinki passages was used, to provide a total of 77 1876 lines of text. Another text from approximately the same period is Boethius; a 1381-line sample of text was drawn from the Godden and Irvine edition (2009, q.v. for a discussion of the text’s chronology), again including the Helsinki passages. The present study follows the Helsinki Corpus practice of omitting passages deriving from metrical sections in the Latin original. The Old English translation of the History of Orosius is another early text included in this study; although the text, formerly ascribed to King Alfred, was originally thought to date from the end of the ninth century (see e.g. Thorpe 1853, v–vii), it has been suggested that from the language of the earliest known manuscript a slightly later date is indicated (Bately 1980, lxxxvi–xciii). A sample of 1702 lines, including the Helsinki passages, was used. This was drawn from the edition of Sweet (1883), which is that used in the Helsinki Corpus, in order to allow the electronic form to be used where possible; however, use was made of Bately’s edition (1980) for its commentary and critical apparatus. A similar decision was made for the Chronicle, for which the Helsinki Corpus version based on the edition of Plummer and Earle (1892) was used. It is hoped that in a study such as this, which is affected less by editorial differences than research to which issues such as phonology are critical, these decisions will have little negative effect on the data. A special case was presented by the Dialogues of Gregory the Great, which exists in two different forms: the original Old English translation from the end of the ninth century, which exists most completely in MS C, and a partial revision made at least a century later, which has been observed to show considerable modernisation of the syntax, found in MS H (see Yerkes 1982; 78 1986). As will be seen below, a preliminary examination found that relevant forms were distributed sparsely within this work; accordingly, in order to obtain sufficient data a more extensive investigation was made of this text than of most others. The first book was analysed in its entirety, in order to provide meaningful statistics regarding the relative frequencies of the periphrastic constructions and semantically comparable preterites; all sections exhibiting periphrastic forms from the remaining books were also analysed, providing additional samples both of periphrastic constructions and of semantically comparable preterites. Although the latter data provide no meaningful information about the relative proportions of the two categories, it will be seen in Chapter 4 that they can provide valid information regarding such matters as translation practices. The primary source used was the print edition of Hecht (1900); however, for the purposes of searching, the electronic York-Toronto Corpus version (Taylor et al. 2003), based on the same text, was used. Texts from later periods of the language were also analysed. These include not only the later portions of the Chronicle but three texts by Ælfric. The Homilies were written toward the end of the tenth century, and from internal evidence probably date from the last decade of the century (see Godden 2000, xxix–xxxvi); the 548-line Helsinki Corpus excerpt was used as the basis for the present analysis. The Lives, which were written shortly after the Homilies (e.g. Skeat 1900, II, xlii), have similarly been analysed here using the excerpt in the Corpus of 900 lines. Ælfric’s translations of the Old Testament have also been included; these provide a valuable point of comparison with his original compositions on the one hand and with different Biblical translations, such as the 79 Gospels, on the other. The Helsinki Corpus excerpts were used, comprising Genesis 1. 1–3. 24, 6. 1–9. 29, 12. 1–14. 20, and 22. 1–22. 19; Numbers 13. 1–14. 45, 16. 1–17. 11, and 18. 1–21. 18; and Joshua 2. 1–7.26. These passages are drawn from those sections of the Heptateuch for which Ælfric’s authorship is most probable (see Clemoes 1966). Another late author whose works are included is Wulfstan; from his Homilies, numbers Ib, II, III, VIIIc, Xc, XII, XIII, XVIb, XVII, and XXe were analysed. As elsewhere, the sample is an expanded version of that in the Helsinki Corpus, although the text was drawn directly from the edition of Bethurum (1957); for purposes of comparison, as discussed in Section 3.4.2 below, consideration was given especially to those homilies with accompanying Latin material. The division made in the Helsinki Corpus between homilies known from earlier manuscripts and those known only from later manuscripts is not preserved here, as this division does not reflect their date of composition (see Bethurum 1957, 101–4). The West Saxon Gospels were also analysed; although the oldest manuscripts of this translation date from the late tenth or early eleventh centuries, and contain a number of late-period linguistic features unlikely to be found in texts from an earlier period, it has been suggested that this text is part of an ongoing tradition of Biblical translation and recension that began at a much earlier date, although we have little direct evidence of the immediate precursors of this text (see Skeat 1871; Grünberg 1967; Liuzza 2000). The Helsinki Corpus excerpt was used, consisting of the first eleven chapters of John. The Helsinki Corpus versions of the Heptateuch and the Gospels are based upon those of Crawford (1922) and Skeat (1871–1887) respectively; as in the case of Orosius, although 80 the older versions are the source for the data analysed in the present study, recourse has been made to the commentary in the more recent editions by Marsden (2008) and Liuzza (1994–2000). 3.2.4 Chronology For the purposes of chronological analysis in order to identify any diachronic trends in the use of these constructions, it was seen as desirable to group the texts chronologically, in order to facilitate comparisons between the earlier and later stages of the language. However, the foregoing discussion should make it clear that arranging the Old English texts analysed here in roughly chronological order is perhaps an easier task than attempting to provide valid estimates of the intervals of time elapsed between these ordinal points. Moreover, not all chronological distinctions that can be drawn are equally meaningful; for example, although it is known that the Lives of the Saints is more recent than the Homilies, since both works were composed by the same author within the same decade it is unlikely that differences between them will reflect diachronic trends operating in Old English generally. For the purposes of analysis, texts have been divided into four groups. The first group, representing approximately the late ninth to early tenth century, includes ChronA1, Bede, CP, Boece, Oros, and GD MS C; although, as previously mentioned, it may be possible to make finer chronological distinctions within this group, such smaller classifications would contain a smaller quantity of data and therefore make the task of statistical analysis more difficult. The second group, representing approximately the mid- tenth century, includes not only ChronA2 but texts which, if not originally composed at this period, at least owe their current form to this time: GD MS H, 81 WS, and GenB. The assignment of Genesis B to a particular period is difficult; although textual evidence suggests that the poem was originally translated into English around the beginning of the tenth century, there are also enough later features to show that the text was altered in the process of recopying, perhaps a century later (see Timmer 1948, 43–50; Doane 1991, 47–53). The third period, centring around the end of the tenth century, consists of ÆCHom, ÆLS, OT, and Wulf; the fourth period consists only of ChronE. The Old English texts chosen for analysis and the chronology assigned to them can be seen in the following table: Late 9th c.–early 10th c. ChronA1 I (–891) Bede CP Boece Oros GD MS C ChronA1 II (892–923) Mid 10th c. ChronA2 (958–1001) GD MS H I GenB WS Late 10th c.–early 11th c. ÆCHom ÆLS OT Wulf 11th c.–12th c. ChronE Table 2: Old English Texts In contrast to the chronological variety found among the Old English texts, the two Old Saxon texts, despite the attempts that have been made to determine the chronological priority of one over the other (see Doane 1991, 43–7), are treated here as essentially contemporary; accordingly, diachronic analysis has been attempted only in the case of Old English. 82 3.2.5 Editorial Practices At this point it may also be useful to provide some additional information regarding the citations of original texts that appear here. For the Chronicle, the Pastoral Care, and the Homilies, the line numbers given are those of the Helsinki Corpus, which are based on the editions of Plummer and Earle (1892–9), Sweet (1871), and Godden (1979) and Pope (1968) respectively; however, adjustments have often been made in the Corpus to accommodate the lineation of the printed texts to such considerations as sentence structure, and correspondence between the Corpus and the published editions may not always be exact. As a result of the incomplete correspondence in both numbering and line division between the Corpus and the printed editions, quotations from the original texts are cited here by the last explicit line number given above in the Corpus, due to the complexities involved in making any interpolated numbers meaningful. For Orosius and the Lives of the Saints, the lineation of the original editions of Sweet (1883) and Skeat (1881–1900) is used. It should also be noted that the section numbers in citations from the Dialogues do not reflect the numbers used in the text, which are taken over from the Latin original, but rather represent a sequential numbering of the sections into which the Old English text itself is divided. Although the analysis and glosses given here are my own, I am indebted to a number of previously published translations of the works studied here, which have been useful when questions of interpretation and ambiguities arose: namely Thorpe (1846; 1853) for the Homilies and Orosius, Skeat et al. (1881–1900) for the Lives of the Saints, Miller (1890) for Bede, Sweet (1883) for the Pastoral Care, Whitelock et al. (1965) for the Chronicle, Kennedy (1916) for Genesis B, Zimmerman (1959) for 83 the original Latin version of the Dialogues, Murphy (1992) for the Heliand, and Godden and Irvine (2009) for Boethius. I have of course differed with the interpretations of these authors at certain points, and take full responsibility for the readings upon which the present study is based. 3.3 Selection Criteria for Verb Tokens 3.3.1 Introduction It is necessary to obtain data on the use both of the periphrastic constructions and of semantically comparable preterites; as discussed in Section 1.4, this provides a meaningful standard of comparison by which the frequency of the periphrastic forms can be interpreted, and allows the different formal categories involved to be studied as alternative means of expression for a single semantic domain. It will be seen that the task of distinguishing the various domains of meaning corresponding to a single grammatical form complicates the identification of relevant occurrences of the preterite, those which are similar in meaning to the present perfect and pluperfect and for which therefore the presence of formal alternatives makes the use of the preterite a significant choice. The definitions of the perfect and pluperfect in semantic terms, established in Section 2.3, are essential to such a task. The gathering of quantitative data for the periphrastic constructions also presents its own complexities; it is necessary to distinguish the grammaticalized periphrastic perfect and pluperfect from the formally identical stative constructions from which they originally developed, in order to include only those periphrases with genuine temporal content, and the attempt must therefore be made to delineate a semantic boundary between the two 84 usages at a period when this formal identity led to a certain degree of semantic continuity. The process of analysing texts to identify relevant forms depends upon interpreting the original text to a sufficient degree to permit judgements both about what grammatical forms would be potentially usable for the representation of a given event and about which such forms are actually licensed within a particular context. This dependence upon interpretative procedures may be seen as introducing a certain element of subjectivity into the process of analysis; however, this interpretative quality is an inseparable part of any research into an area in which pragmatics plays a role, and in devising the methodology of the present study the goal has been to make the process of data collection as objective as possible. The following sections will describe some of the steps that have been taken to this end. 3.3.2 The Simple Preterite As discussed in Section 1.1.3 above, the simple preterite in Old English and Old Saxon originally had a breadth of meaning similar to that reconstructed for Proto-Germanic, and could be used in contexts in which specific perfect and pluperfect forms, after these arose, would later become usual. Prior to the development of the perfect and pluperfect as distinct formal categories, temporal and contextual information was not marked within the grammar to so great an extent as in Modern English; instead, such information was conveyed by implicit and explicit cues in the surrounding text. Accordingly, in order to identify those preterites which could potentially have been replaced by periphrastic constructions, attention is necessary not only to the verb forms themselves and the temporal properties of the events that they denote but to the contextual cues that 85 would have served to distinguish among several possible interpretations. Because of the differences between the perfect and the pluperfect in their relationship to the pragmatic context and the extent of their dependence upon it, as discussed in Chapter 2, it is necessary to make some adaptations in the methodological approach in identifying preterites corresponding to these two categories; the procedures used for the present perfect will be discussed first, followed by those used for the pluperfect. 3.3.2.1 Perfect-Like Preterites To recapitulate some of the discussion in Chapter 2, the representation proposed by Reichenbach (1947) of the present perfect as a semantic category is the formula ‘E — S, R’; in the context of the more complex model of the perfect proposed above, incorporating the perspective of more recent work on the perfect, this may be reinterpreted as referring to an event of which the starting point, and perhaps the endpoint, precedes a pragmatically salient reference point coincident with the present. In Modern English, the perfect therefore differs from the past tense in that the former explicitly marks the existence of such a reference point as separate from the event itself, whereas the past tense, as a result of its paradigmatic opposition to the perfect, may in some cases create an implicature that no such present reference is relevant (the existence of this paradigmatic opposition was already noted in such early works as Hoffmann 1934); Reichenbach’s representation of the past tense as ‘E, R — S’ reflects the absence of any time other than that of the event to which reference is made from the semantic content expressed by this form. It seems unlikely that prior to the evolution of this paradigmatic opposition between the two forms, there would 86 have existed any such implicature resulting from the use of the preterite in early Germanic languages. In the absence of such a marked opposition to the perfect, it was originally possible for the preterite to be used in contexts which effectively supplied such a present reference point by other means, so that the semantic content of the whole may be seen as equivalent to a perfect; with the development of a paradigmatic opposition between past and perfect and the concomitant growth in the explicit or implicit positioning of a reference point by the use of a particular verb form alone, the use of the preterite in such contexts would have come to seem more and more infelicitous. In order to track the growth of such oppositions and to identify those occurrences of the simple preterite which are equivalent to the present perfect, it is necessary to make use of an analytical procedure sensitive to pragmatic content of the sort conveyed by the perfect. Most proposals for formal representations of the perfect which address these pragmatic factors, such as Portner (2003), merely describe the pragmatic assertions created by the presence of perfect forms, without attempting to make any predictive statements such as an enumeration of the types of pragmatic assertion whose presence makes the perfect the only permissible, or only unmarked, form. Even discussions of such pragmatic assertions in terms of the variation observed in their use between different speakers or different dialects (see the discussion on dialectal differences in this section below) are generally restricted in their scope to the small number of contexts in which such variation is conspicuous. This absence of predictive ability is hardly surprising, however, given the extent to which such pragmatic phenomena can interact with various extralinguistic factors and can differ even 87 within a given context; to a considerable extent, speakers are free to choose whether or not to draw their interlocutors’ attention to a particular element of the context, such as a present result of a past event. Although the existence of such choice and variation is a considerable obstacle to any attempt at exhaustively listing necessary and sufficient pragmatic conditions for the use of the perfect, it is nevertheless possible to make certain generalizations about the contexts in which the perfect and its associated content would appropriately occur; otherwise, the perfect and the preterite would occur in completely free variation. That this is not the case is illustrated by examples such as the following: (56) Her forþferde Peada, & Wulfhere here forth.fare-3SG.PRET Peada-NOM and Wulfhere-NOM Pending feng to Miercna Pending-NOM seize-3SG.PRET to Mercian-NEUT-DAT-SG rice. kingdom-DAT (ChronA1 I 657.1, p. 28) ‘In this year Peada died, and Wulfhere, son of Penda, succeeded to the Mercian kingdom.’ (57) ?Her hæfð Peada forþfered, & here have-3SG.PRES Peada-NOM forth.fare-PA.PPL and Wulfhere Pending feng to Wulfhere-NOM Pending-NOM seize-3SG.PRET to Miercna rice. Mercian-NEUT-DAT-SG kingdom-DAT ?‘In this year Peada has died, and Wulfhere, son of Penda, succeeded to the Mercian kingdom.’ In (56), it is clear that the two events described follow one after the other in a single temporal sequence. In the corresponding hypothetical example (57), on the other hand, the temporal relationships among the events are more difficult to interpret, but it would seem at least less likely that the succession of the latter upon the former is intended in the same way; no such combinations of the perfect 88 and the preterite to express a single sequence of events have been identified in the present study. The differences between the perfect and the simple past can be seen reflected in the methodology of studies such as that by Dahl (1985); his study investigated the existence of the perfect as a grammatical category in different languages by means of a production test providing contextual cues likely to elicit particular uses of the perfect, such as experiential or resultative perfects. However, such a procedure is obviously inapplicable to the analysis of historical texts. Other methodology often used in semantic work, such as truth tests (see e.g. Pianesi and Varzi 2000, 12–27), is likewise inapplicable to distinctions such as those at issue here, which are primarily pragmatic in nature; it has been shown that the interchange of perfects and preterites, however infelicitous the effects that this may produce, often has no effect on the truth value of the sentences to the same extent that the interchange of other tenses and aspects might (e.g. Mittwoch 2008). This can be seen in examples such as these: (58) Mary has run a mile. (59) Mary ran a mile. If it is acceptable to say, as in (58), that Mary has run a mile, then it must be true that at some point in the past Mary ran a mile; conversely, if at some point in the past Mary ran a mile, then we may say in an experiential sense that Mary has run a mile, because running a mile is one of the things that she has done. Even in cases such as the Gutenberg example in (28), in which the perfect is pragmatically inappropriate, its use does not make the proposition false; we cannot say that it is untrue that Gutenberg has invented the printing press. Because the difference 89 between the two formal categories in question is not truth-functional, a methodological approach is needed which makes use of criteria other than truth conditions. The procedure adopted here for the identification of relevant occurrences of the simple preterite, despite certain limitations, is considered to be the best available. It is essentially a straightforward translation test; if a Modern English present perfect is the only acceptable idiomatic equivalent of an Old English or Old Saxon preterite, the latter is counted as falling within the perfect domain in its meaning. As discussed above in Chapter 2, there is evidence for the existence of an abstract semantic/pragmatic perfect category with cross-linguistic validity, represented by grammatical forms of which the Modern English perfect is a characteristic example; if this is the case, then the possibility of using a Modern English perfect in a given context to refer to an event denoted within the same context by the preterite can be taken as indirect evidence that these grammatical forms are in this case similar enough semantically to be considered as belonging to a single domain. Although this approach avoids the difficulties inherent in any attempt at providing a formal definition, sufficiently exhaustive to have predictive force, of the pragmatic constraints upon this category, it involves other issues that must themselves be addressed. In languages such as Modern English in which the perfect exists as a discrete grammatical category, the use of this form can in itself indicate the presence of the pragmatic component of the meaning of this category, without the need for any additional contextual cues; it is therefore often possible to translate an Old English or Old Saxon preterite with a Modern English perfect, essentially by interpolating pragmatic content absent from, but compatible 90 with, the original text. It is for this reason that where any question exists as to which of the Modern English tenses should be used as a translation for an original preterite, there is considered to be insufficient evidence for the latter to be counted as equivalent to a prefect; only those cases for which no such ambiguity exists will be counted. It has often been noted that the contexts in which a perfect is necessary vary dialectally; for example, in American English the use of the simple past to describe relevant recent events (e.g. I already ate) is more acceptable than in British English (see e.g. Michaelis 1994, 124–5; for further bibliography Hundt and Smith 2009). The semantic nature of such differences has been subject to varying interpretations; for example, Lindstedt (2000, 370–1) sees this phenomenon as evidence that the American perfect is further removed from prototypical perfect semantics, while McCoard (1978, 241–6) views it as a difference not in the perfect but in the past tenses of these dialects. In the absence of any widely accepted theory of the long-term diachronic development of this variation (see Hundt and Smith 2009, 45–7), the present study follows Dahl (1985) and Dahl and Hedin (2000) in considering the use of the perfect in such contexts as a genuine reflection of the semantics of this cross-linguistic category; if Old English should resemble present-day American English in avoiding the periphrastic forms in such contexts, this would be in itself a meaningful datum that could not be studied if these contexts were excluded from the present analysis, one which might shed further light on the semantic content of perfect- like preterites. However, from a practical point of view, the contexts in which the variation observed in present-day usage could occur are relatively rare in the data; such recent-past contexts make up only a small proportion of the written texts 91 studied, and those that do occur often contain elements, such as the present-time adverbial in (55), that preclude the past tense in all standard forms of Modern English. Another issue is that although the perfect, as a semantic/pragmatic category, shows a significant degree of cross-linguistic stability, there is also a certain degree of cross-linguistic variation; for example, it was observed in Section 2.4.1 above that the English incompatibility of the perfect with definite past time adverbials is not paralleled in a number of languages with otherwise comparable grammatical categories. Rather than using Modern English as the sole basis for defining the fine-grained pragmatic conventions of the languages being studied, a risk which a translation-based method of analysis might seem to carry, the distributional properties of the periphrastic constructions in the languages themselves may be used as an indicator of the constraints then prevalent, and any significant discrepancy from Modern English that might be found would be considered grounds for overriding the criterion of translation acceptability. It may be seen that the use of a method of this nature for data analysis places considerable weight on the ability of the reader to form judgements regarding the content and pragmatic import of historical texts. However, similar methodological assumptions may be seen as underlying previous semantic comparisons of the preterite and the perfect (e.g. Mitchell 1985); moreover, in the absence of any well-established alternative diagnostic procedures, it is felt that the inevitable level of error introduced by this methodology is less than would be involved in any attempt to create a more 92 formalized system for analysis, especially in an area where so little consensus exists on many essential points. The application of the methodological approach described above to the analysis of data can best be illustrated by means of examples. (60) Nu ic þyses Alexandres her gemyndgade, […] now I-NOM this-MASC.GEN Alexander-GEN here recall-1SG.PRET nu ic wille eac þæs now I-NOM will-1SG.PRES also the-MASC.GEN.SG maran Alexandres gemunende beon, […] more.MASC.GEN.SG Alexander-GEN mention-PR.PPL be-INF þeh ic ymbe Romana gewin on though I-NOM about Roman-GEN.PL struggle-ACC.SG on þæm gearrime forð ofer þæt the-NEUT.DAT.SG year.count-DAT forth over that-NEUT.ACC geteled hæbbe. tell-PA.PPL have-1SG.PRES (Oros III.7.110.12) ‘Now I have recalled (??recalled) this Alexander here, I will now also mention Alexander the Great, though I have already told above of the Romans’ struggle in the following years.’ This example presents one of the more straightforward cases of identification. The gloss above illustrates the obligatory nature of the Modern English perfect as a translation of the preterite gemyndgade, and the relative unacceptability of the Modern English past tense as an equivalent. The results of the translation test are corroborated in this instance by other features of the original text which combine to provide further evidence for the perfect-like semantics of the preterite in this example. The verb in question is used to refer to a past action with results of present relevance, in a sequence of primary tenses and in conjunction with the present-time adverb nu ‘now’; within the same sentence is also a periphrastic perfect referring to an event of similar temporal position and pragmatic relevance, 93 a pattern of co-occurrence whose frequency has previously been remarked (see Section 2.5.2 above). (61) Ic heold nu nigon gear wið ealle I-NOM hold-1SG.PRET now nine year-ACC with all-FEM.ACC.SG hynða þines fæder gestreon loss-ACC thy-MASC.GEN.SG father.GEN property.ACC (ÆLS I.212.42) ‘I have held (*held) your father’s property nine years now against all loss’ The interpretation of this example too is uncontroversial. In addition to factors similar to those discussed above, the context of this example makes it clear that the preterite heold refers to an event still going on at the time of utterance; accordingly, using the Modern English past tense rather than the perfect as a translation would give an entirely different sense from that of the original. Like (60), this sentence includes the temporal adverb nu ‘now’; however, the continuing nature of the event would be made sufficiently clear by the context even in the absence of such explicit marking. In contrast to the preceding examples, some occurrences of the preterite are less easily categorized. In the following example, the preterite form may be translated acceptably by either the Modern English preterite or the perfect, and although it might be possible to make a case for either choice, the original context does not provide enough cues to permit the conclusion that the intended sense was definitely that of a perfect; accordingly, this example was not included in the count of perfect-like preterites. (62) Þu þe þyrstende wære monnes blodes thou-NOM REL thirst-PR.PPL be-2SG.PRET man-GEN blood-GEN XXX wintra, drync nu þine fylle thirty winter-ACC.PL drink-2SG.IMP now thy-FEM.ACC.SG fill-ACC (Oros II.4.76.33) 94 ‘You who have been thirsting / were thirsting for man’s blood for thirty years, drink now your fill.’ It may be seen that a wide range of factors can be taken into account through the procedures employed here to quantify the presence in texts of preterites of a significant degree of similarity in content to the contemporary periphrastic constructions, and that what other criteria are available tend to corroborate the results obtained in this way. 3.3.2.2 Pluperfect-Like Preterites The pluperfect is distinct not only from the simple preterite and present but from the present perfect in that it makes reference to three separate points in time, occurring in a specific order; this salient property can be seen in the previously discussed formulaic representation of the pluperfect by Reichenbach (1947), as ’E — R — S’. As a result, the methodological processes involved in identifying preterite forms falling within the semantic–pragmatic domain of the pluperfect are different in some respects from those used in identifying preterites falling within the domain of the present perfect. The temporal and aspectual content of the perfect makes reference to only two points in time, the time of the event and the time of utterance (or Extended Now), and therefore the principal task in determining whether a particular preterite falls within the domain of the present perfect is to decide whether the present functions as a reference point R in addition to being the time of utterance S. In the case of the pluperfect a preliminary step to pragmatic evaluation of the sort performed for the perfect is the simple task of determining whether there is any reference in the text to a point in time intervening between the event and the moment of utterance such that it 95 could potentially be a reference point R; only if a point of this sort is present is it necessary to decide whether there is justification for viewing that point as pragmatically salient enough to be integrated into the content of the verb in a pluperfect-type schema. Although this preliminary elimination allows preterites that are candidates for pluperfect-like status to be identified more easily, pragmatic criteria are still necessary to determine which of these candidates are actually comparable to the pluperfect; this task is complicated by the fact that in Modern English variation exists between the simple past and the pluperfect to a degree not found with the perfect. This can be seen in the following examples, repeated here from Chapter 2; the pairs in question differ little or not at all both in truth value and in pragmatic effect. (63) John was surprised that he broke the window. (64) John was surprised that he had broken the window. (65) John said that Mary ran a mile. (66) John said that Mary had run a mile. From the perspective of the present study, such variation is most plausibly interpreted as evidence that in Modern English it is still possible to use preterite forms as expressions of pluperfect meaning in the presence of appropriate contextual cues; in other words, that the association of a single semantic domain with multiple grammatical categories, which was described above as occurring in the older languages, persists to a certain extent in Modern English. Nevertheless, the degree of permissible variation is less in Modern English than in Old English, as can be seen from examples such as the following: 1 1 From the context it is clear that an imperfective interpretation ‘was travelling’ was not meant. 96 (67) Þa se gehyrde þæt se when that-MASC.NOM.SG hear-3SG.PRET that the-MASC.NOM.SG Hælend for […] to Galilea he com healer-NOM fare-3SG.PRET to Galilee he-NOM come-3SG.PRET to him to he-DAT (WS John 4. 47) ‘When that man heard that the Saviour had travelled (*travelled) to Galilee, he came to Him.’ The persistence of variation between the simple past and the pluperfect raises certain issues for translation tests of the sort previously described. There will be some Old English and Old Saxon preterites which, despite being pluperfect-like in the same way in which other preterites in these languages are perfect-like, may be translated in Modern English by a preterite. Moreover, as with the perfect there are also original preterites which could be replaced in translation by a Modern English preterite referring to the same event, but only through an interpolation of pragmatic information not present in the original text, giving greater prominence to the reference point at issue than was the original intention. However, even within Modern English there are criteria by which pluperfect-like preterites can be identified as a separate class; for example, as discussed in Section 2.5.2.2, such forms have a different temporal reference in subordinate clauses from other past- tense forms. When sufficient contextual clues exist to determine that temporal anteriority is denoted by the verb in this way, the identification of a preterite as pluperfect-like is generally unproblematic. Where no such context exists, the question is unlikely to arise; Mitchell (1975, 159–66) argues that the ascription of pluperfect semantic content to Old English preterites by native speakers was probably dependent on the presence of such contextual cues, and provides a discussion of the environments in which a pluperfect interpretation is feasible. 97 Although little attention has been given explicitly to this issue in Old Saxon, it might seem reasonable to hypothesize a similar situation for both languages. As with the perfect, the effects of context in the case of the pluperfect are crucial criteria to which attention must be paid in order to determine whether a particular interpretation is justified. As in the case of the perfect, the practical application of the criteria above can be seen most clearly with reference to examples drawn from the texts under analysis. (68) Hæfdon hi hiora onfangen ær have-3PL.PRET they-NOM they-GEN on.take-PA.PPL ere Hæsten to Beamfleote come, & Hæsten-NOM to Benfleet-DAT come-3SG.PRET-SUBJ and he him hæfde geseald gislas he-NOM they-DAT have-3SG.PRET sell-PA.PPL hostage-ACC.PL & aðas & se cyng him and oath-ACC.PL and the-MASC.NOM.SG king-NOM they-DAT eac wel feoh sealde & eac swa also well money-ACC sell-3SG.PRET and also so þa he þone cniht agef when he-NOM the-MASC.ACC.SG knight-ACC out.give-3SG.PRET & þæt wif and the-NEUT.ACC.SG wife-ACC (ChronA1 II 894.55, p. 86) ‘They had sponsored them before Hæsten came to Benfleet, and he had given them hostages and oaths, and the king had also given him money well, as he did when he gave back the boy and the woman.’ There are some similarities between this example and the perfect in (60) above; the preterite sealde is found in combination with a periphrastic pluperfect, hæfde geseald, in a context in which there is no apparent difference in the temporal relations of the events denoted by the two verbs. The use of two different forms for an apparently similar purpose simplifies the identification of preterites 98 comparable in meaning to the pluperfect, even in an environment such as this, a main clause with no temporal adverbials (cf. Mitchell 1975, 162). (69) Thô fôrun eft thie man then fare-3PL.PRET back the-MASC.NOM.PL man-NOM.PL thanan[…] al sô im the engil thence all so they-DAT the-MASC.NOM.SG angel-NOM godes […] giuuîsde God-GEN advise-3SG.PRET (Heliand VIII.31.693) ‘Then the men travelled back from that place as the angel of God (had) advised them.’ This example illustrates some of the variability previously discussed. The sentence in (69) makes reference to two events, of which one is clearly anterior in time to the other; the later event thus has at least the potential to be used as the reference point of a pluperfect; however, the translation shows that in Modern English either a preterite or a pluperfect can be used to refer to the anterior event. The position taken here is that this variability in form is not reflected in any significant difference in content, whatever other differences there may be in areas such as style, and that a context such as this is sufficiently rich that the additional marking provided by the pluperfect is to a certain extent redundant; it is precisely because the pragmatic content of the Modern English periphrasis is already conveyed by other means within the original text that its use in translation is not considered an unjustified departure, and preterites such as this are counted as falling within the pluperfect domain. However, not all chronological information within the text provides a potential reference point for the pluperfect, as the following example shows: 99 (70) On þys gere foran to on this-NEUT.INSTR.SG year-INSTR before to middum sumera for Eadweard mid-MASC.DAT.SG summer-NOM fare-3SG.PRET Edward-NOM cyning to Mældune[…]. king-NOM to Maldon-DAT (ChronA1 II 920.1, p. 100) ‘In this year before midsummer King Edward travelled to Maldon.’ Sentences such as that in (70) above make explicit reference to a point in time (in this case midsummer) which intervenes between the time of the event (travelling) and the time of utterance and which therefore may be considered as a potential reference point R in accordance with the definition of the pluperfect discussed above. However, if the preterite for were translated by a Modern English pluperfect, the import of this translation would be quite different from that produced in the original by the preterite; the emphasis would be on a state of affairs, existing at midsummer, in which the event of travelling was already completed, and there would be little justification for saying that a similar pragmatic effect is created in the original text simply by the mention of a point in time subsequent to the event. Accordingly, preterites such as that in (70) are not considered as comparable to pluperfects. 3.3.2.3 The Preterite and Mood The process of determining whether an occurrence of the preterite is relevant to the present study not only makes reference to the criteria described above but takes other factors into account; one important issue that has not been previously discussed is that of mood. Up to this point, semantic discussions of the preterite and the perfect system have assumed that the events to which verbs refer are real, or at least treated as real grammatically. When moods other than the 100 indicative are introduced, the semantic factors to be taken into account become more complicated, and other factors have the potential to interact with the phenomena under analysis; for example, in the case of the subjunctive the interaction of this category with modal verbs and their diachronic development would become relevant. It might seem that a reasonable precaution to reduce the number of variables that could potentially be conflated would be to focus entirely on the indicative, as a grammatical category that could legitimately be studied in its own right, and that no distortion would be introduced into the data by such a step. However, one obstacle to this course of action is the considerable syncretism existing in the Old English preterite between the indicative and the subjunctive; any attempt to exclude all subjunctive verbs would necessitate making a large number of possibly suspect judgements regarding ambiguous forms, including the numerous forms whose interpretation has been the subject of past controversy (see e.g. Mitchell 1985, I, 231–2). Syncretism between the subjunctive and the indicative does not exist to the same extent in Old Saxon, and in general it would be feasible to identify and exclude the majority of Old Saxon subjunctives based on formal grounds alone; however, cross-linguistic comparison of the data would be difficult without the use of a single set of criteria for both languages. The solution adopted here is to use semantic rather than morphological grounds to identify those subjunctive forms that would introduce such extra semantic variables. There are many constructions in which subjunctive morphology has little semantic content that would differentiate it from the indicative, and in which its use is necessitated by a particular syntactic context, such as subordination of the sort found in indirect discourse; it is precisely in such 101 semantically neutral contexts that there is the greatest difficulty in distinguishing the subjunctive from the indicative (see Mitchell 1985, I, 369–70). Subjunctives of this sort, even those which are morphologically marked, are treated in the same way as indicatives: (71) Atheniense bædan Philippus, þæt he heora Athenian-NOM.PL bid-3PL.PRET Philip that he-NOM they-GEN ladteow wære[…], þeh hie ær hiera leader-NOM be-3SG.PRET.SUBJ though they-NOM ere they-GEN clusan him ongean belucen. bolt-ACC.PL he-DAT against lock-3PL.PRET.SUBJ (Oros VII.114.21) ‘The Athenians asked Philip to be their leader, although they had previously barred their gates against him.’ In this example, the subjunctive morphology of belucen does not assert the unreality of the event described, and the preterite has the same temporal significance as it would in an indicative sentence such as Hie ær belucon hiera clusan; accordingly, the subjunctive construction in (71) is included in the count of pluperfect-like preterites. However, not all subjunctive constructions fall into this category. In some cases the use of the subjunctive is related to more pronounced semantic differences; in constructions such as conditionals, a genuine irrealis force is present: (72) Gif Abraham ne ongæte Lazarum, if Abraham-NOM NEG recognize-3SG.PRET.SUBJ Lazarus-ACC ne spræce he nænigra þinga NEG speak-3SG.PRET.SUBJ he-NOM NEG.any-GEN.PL thing-GEN.PL swa to þam weligan men[…]. so to the-MASC.DAT.SG wealthy-DAT.SG man-DAT (GD MS C IV.XXXIV.310.24) ‘If Abraham had not known Lazarus, he would not have spoken of anything thus to the wealthy man.’ 102 Even in the Modern English equivalents of sentences such as (72), the pluperfect (subjunctive) is not used with the sense found in the indicative, expressed in Reichenbachian terms as indicating the anteriority of a past event with respect to another point in past time, but rather to indicate the counterfactual nature of a past condition; similarly, the use of the preterite subjunctive in Old English indicates the unreality of the event denoted by the verb. The semantic differences between these two types of subjunctive construction are reflected in their separate treatment within the present study. The position adopted here is that the semantic differences between subjunctives of the sort seen in (72) and indicatives are great enough that their development cannot safely be assumed to have followed parallel courses. The later diachronic development of these two types of subjunctive provides a certain degree of support for the decision to treat them differently; subjunctives such as that seen in (71) have simply been replaced in Modern English by the corresponding unmarked indicative forms, while the semantic content expressed by a subjunctive in conditional sentences such as (72) receives a greater degree of formal marking. Even when there is no morphologically distinct subjunctive form the use of secondary tenses conveys a distinct irrealis force, and although morphological marking of this modal content has been reduced, the use of periphrases with modal verbs has provided some degree of compensation, a process whose effects can be seen in the use of would in the apodosis of the Modern English translation of (72). Accordingly, it is only subjunctives of this latter type that are excluded from the present study, while those of the former type are included. 103 3.3.2.4 The Preterite and Other Aspects As discussed previously, the perfect as a semantic category interacts with other aspectual distinctions; for example, in Section 2.3.1 it was mentioned that the combination of the perfect with progressive aspect normally results in a continuative interpretation. The numerous morphologically distinct grammatical forms available in Modern English to express nuances of tense and aspect make it possible to draw distinctions between continuing events with a salient relationship both to the past and to the Extended Now (present perfect progressives) and continuing events in progress at a reference point in the past (pluperfect progressives) from other continuing events, about which nothing is asserted except their occurrence in the present or the past. In Old English and Old Saxon, there are some preterites that could be interpreted as perfects only if it were assumed that they corresponded semantically to Modern English perfect progressives in the same way in which some present tenses in these languages correspond to Modern English present progressives. However, in such cases the preterite can also be interpreted as having pure past-tense semantics, without any of the properties identified above as criteria for its identification as perfect-like or pluperfect-like. As a result the formal distinction among the semantic domains of the past, perfect, and pluperfect is entirely neutralized where these domains intersect with that of the progressive: 104 (73) Se here […] forsawon ælc the-MASC.NOM.SG host-NOM despise-3PL.PRET each-MASC.ACC.SG frið þe Eadweard cyng & his peace-ACC REL Edward-NOM king-NOM and he-GEN witan him budon. councillor-NOM.PL they-DAT offer-3PL.PAST (ChronA1 II 911.1, p. 96) ‘The army despised every peace that King Edward and his councillors offered / were offering / had been offering them.’ This example shows the extent to which ambiguity exists regarding the most appropriate Modern English equivalent of the Old English preterite. According to one possible interpretation of the effects that a given choice would have on the Modern English translation, the pluperfect progressive would explicitly mark the time of despising as a salient reference point with respect to the event of offering and create an implicature that this attitude, in the form of some overt expression, put an end to the offering, while the simple preterite would explicitly indicate that the event of offering was terminated but would provide no overt marking of the temporal relationships between the two events, and the preterite progressive would make the least commitment as to the temporal relationships between these events and the points in time at which they ended. While it might be possible to propose an analysis in which these three shades of meaning could have existed in the mind of an Old English speaker who would then have chosen to represent any of them by the single form in (73), it would seem far simpler and more parsimonious to suppose that in such cases only a single, less specific past-tense meaning was intended; unlike the perfect-like and pluperfect-like preterites discussed above, there are no firm criteria for distinguishing preterites such as this either in form or in meaning, and they are therefore not included in the present analysis. 105 3.3.2.5 The Preterite and Verbal Prefixes The use of verbal prefixes can in general affect the semantics of verbs and the utterances in which they occur. One prefix that is particularly at issue in a discussion of the semantic domains considered in the present study is the Germanic morpheme represented by Gothic ga-, Old Saxon gi-, and Old English ge-, referred to in this work simply as ge. This prefix can be attached to nouns and verbs, and in Old English and Old Saxon it generally occurs with the past participles even of verbs that lack the prefix in their other forms. The semantic content and morphosyntactic function of ge have been the object of considerable discussion and controversy; among the proposals that have been made are the theories that this prefix expresses perfect or perfective meaning (for bibliographic discussion see Lindemann 1970, 2–10). However, there is ample evidence that ge did not contribute any such semantic content as part of a regular and obligatory grammatical system of aspect of the sort found in the Slavic languages. For Old English, Scherer (1958) and Lindemann (1970) have shown that the distribution of forms with ge is not correlated with that of any semantic domains such as the perfective or the perfect; for example, a preterite such as gehyrdon ‘heard’ can be used to translate a Latin imperfect tense denoting ongoing, incomplete action (Scherer 1958, 247). If ge is not a grammatical morpheme expressing aspect, it may be more plausible to see its use as a process of lexical derivation that affects the Aktionsart of a verb; the Aktionsart properties produced in this way can then interact with pragmatic factors and grammatical factors such as tense to produce a wide variety of completeness/incompleteness readings for any given verb form (e.g. Mitchell 1985, I, 365–6). The question remains of what the semantic 106 contribution of ge is; it has been suggested (e.g. Lindemann 1970, 28–38) that from an original, more concrete meaning, possibly directional in nature, the compounds in which it occurred came to form a semantic spectrum ranging from those that preserve this spatial sense to those in which a more perfective-like meaning has arisen through the notion of attainment of a goal. One of the ways in which this distinction between forms with and without ge was manifested can be seen in sentences such as the following: (74) We gehierdon betueoxn eow unryhthæmed, we-NOM hear-1PL.PRET betwixt you-DAT unright.intercourse-ACC ge sua unryht sua we furðum betwuxn hæðnum and so unright so we-NOM even betwixt heathen-DAT.PL monnum ne hierdon[…]. man-DAT.PL NEG hear-1PL.PRET (CP XXXII.211.7) ‘We have heard of unright intercourse among you, and so unright as we have not even heard of among the heathens.’ Such a use of ge in a positive form but not in a coordinated negative form is a recurring pattern in Old English (Lindemann 1970, 23). This pattern can be related to the semantic concept of completeness; the omission of the prefix from the negative form would thus have conveyed the idea that not only was there no complete event of hearing, but there was not even a partial or incomplete event. 2 However, it is important to note that this notion of completeness is not related semantically to the perfect, as both preterites are equally perfect-like according to the criteria used here. Nor is it connected to perfective aspect as this category is generally understood; the two verbs do not differ in the extent to which they present hearing as a unitary occurrence as opposed to a continuing process (cf. 2 Here, as elsewhere in the present work, the use of identical glosses for simple and compound forms of a verb does not mean that there is no semantic difference between the two, but rather that any distinction would be neutralized at the lexical level in Modern English. 107 Comrie 1976, 16–21). Although there is less evidence for the situation in Old Saxon, in that language ge similarly lacks the distributional patterns associated with a marker of grammatical aspect. Rauch (1992, 185–204) considers it to function as a means of lexical derivation except for its use with past participles; Watts (2001, 133) suggests that the use of ge with Old Saxon participles may be the product of an earlier system in which oppositions between forms with and without the prefix occurred in all forms of the verb. As Lindemann (1970, 28–35) observes, ge is not the only prefix that can have an effect upon the Aktionsart of a verb; in the present study the effects of ge and of other verbal prefixes do not receive separate treatment of a sort not given to other Aktionsart phenomena. 3.3.3 Periphrastic Forms The methodological techniques used for analysing texts to identify relevant periphrastic constructions differ in a number of ways from those used for comparable occurrences of the preterite. While the preterite forms relevant to the present study are merely a small subset of a large formal class, all of which must be examined in order to identify the relevant cases, the periphrastic constructions constitute an easily identifiable formal class, of which far fewer instances must be rejected as not relevant to the present study. The semantic and pragmatic differences between the present perfect and the pluperfect have substantial methodological implications for the analysis of the preterite data; however, it will be seen that these differences have little to do with the semantic and pragmatic factors most relevant to the analysis of the periphrastic forms, and that the different tenses of the periphrasis can therefore be treated in essentially the same way. Although in this respect the quantitative analysis of the periphrastic perfect 108 forms is less complicated than that of the simple preterite, both forms of analysis require well-defined criteria for the identification of relevant forms and the exclusion of those which are irrelevant; one important issue is that of distinguishing the use of periphrastic constructions as a grammaticalized marker of the pluperfect from the stative use that derives from an earlier stage in the process of grammaticalization. 3.3.3.1 Auxiliaries As mentioned in Section 1.2.3.1 above, although the present study concentrates upon perfects formed with auxiliaries originally meaning ‘have’, Old English and Old Saxon had other auxiliaries available for the formation of perfects. In addition to the verbs habban/hebbian ‘have’, wesan/uuesan ‘be’ was also used; these auxiliaries showed semantic differentiation as described in Chapter 1. Another verb, weorþan/werðan ‘become’, could be used like wesan to express passive meaning and may also have been used as an auxiliary of the perfect with a similar semantic distribution to that of wesan, although it may be questioned whether it was ever grammaticalized to the same extent as the latter (see e.g. Mitchell 1985, I, 298–301; Rauch 1992; 163); to the extent that it was used in this way, the following remarks on wesan may also be taken as applicable to it. Some of the reasons for concentrating on have-perfects are methodological. Auxiliaries originally meaning ‘be’ were used both for the perfect and for the passive, and the lack of formal distinction between the two types of construction produces certain methodological complications not existing in the case of the have-perfects; not only could be-perfects have either the original 109 stative sense or the later eventive sense, but for the large number of verbs that could be construed either as transitives or as intransitives these constructions could also have a passive sense. For example, an Old English phrase such as wesan gewanod could, depending on the context, be interpreted as ‘to be less’, ‘to be lessened’, ‘to have lessened’, or even ‘to have been lessened’ (see further Mitchell 1985, I, 315–19). The great frequency of such ambiguities, whose identification as perfects or passives would often necessarily be arbitrary, has contributed to the decision to focus in the present study upon the have- periphrases. In addition to these methodological considerations, however, there are more pressing semantic reasons for treating have-perfects and be-perfects separately. As discussed in Section 1.2.3, the two types of auxiliary had different origins, were grammaticalized in different ways, and followed different paths of development; it might be asked whether a high degree of semantic similarity existed between the types of construction in which they were used. At least in the case of Old English, in which have-perfects could occur with unaccusative verbs, some controversy exists as to whether the periphrastic constructions with wesan were actually comparable in their semantics to the habban-constructions (e.g. Mitchell 1985, I, 303–4; Rydén and Brorström 1987; McFadden and Alexiadou 2010); even if the two constructions could in fact be used synonymously in some instances, the assumption that this was always the case is perhaps not a position that can safely be used a priori as a premise on which to base new findings. Even in Old Saxon, in which the distribution of the two auxiliaries is more prototypical and shows signs of complementarity to a greater degree than is found in Old English, there are many ways in which the choice of auxiliary could be altered 110 from the unmarked pattern on semantic or pragmatic grounds (Arnett 1997). It might be desirable to treat be-perfects and have-perfects as separate categories that may each be studied in their own right, regardless of the degree of similarity that may be found to exist between them; the concentration of the present study on the latter of these categories is therefore not necessarily an exclusion of data crucial to the questions under examination. It is nevertheless important to be aware of the potential effects of this decision upon the data. One possible consequence of this decision upon the data is that the figures for the preterite will include not only cases corresponding to have-perfects but those corresponding to be-perfects; this might tend to weight the ratio of simple preterites to periphrastic constructions on the side of the former. For Old English this may be seen as a methodological inevitability; as discussed in Section 1.3.3, the periphrasis with habban was used in Old English for verbs of all types even at a relatively early date, including unaccusatives of the sort that also used wesan as an auxiliary of the perfect; there are therefore few lexemes for which the possibility of their taking habban as an auxiliary of the perfect can be discounted a priori (see Mitchell 1985, I, 302). In Old Saxon the situation is somewhat different; Arnett finds that many verbs forming be-perfects are never found to form have-perfects (1997, 35). However, the adoption of the same procedure for Old Saxon as for Old English might be more desirable in that the use of parallel criteria would facilitate cross-linguistic comparison. Although the inclusion of all semantically suitable preterites, regardless of auxiliary selection, may introduce a certain bias into the data for any individual text, it is hoped that 111 this bias will be sufficiently consistent across different texts and different time periods not to invalidate the comparisons made. The data from Old Saxon provide an illustration of the specific effects of this choice. In the Heliand, Arnett identifies 39 be-perfects and pluperfects (1997, 52–5), of which 23 fall within the sample analysed here. The same sample of the Heliand includes 3 perfect-like and 8 pluperfect-like preterites of the same lexemes, out of a total of 19 and 35 respectively; none of these lexemes was found in a relevant form in Genesis. If auxiliary selection in Old Saxon were treated as strictly complementary, these forms would all be excluded from the analysis. Analysis of the Old Saxon data has been performed both with and without these forms; it will be seen in Chapter 4 that their inclusion or exclusion has no statistically significant effect on the results. 3.3.3.2 Stages of Grammaticalization The focus of this study is upon periods during which the position of the periphrastic perfect constructions within the grammars of Old English and Old Saxon were in a certain degree of flux; not only was the relationship of these periphrases to the verbal system changing, but they were still in the process of becoming differentiated semantically from the original stative constructions, which still continued to be used beside the grammaticalized constructions with little or no formal differentiation between the two categories. As discussed in Section 1.2.2, the persistence within the language of constructions representing both earlier and later stages of grammaticalization is a widespread and characteristic phenomenon (see e.g. Hopper and Traugott 1993, 3), which makes it necessary to evaluate carefully the stage of grammaticalization represented by a 112 given construction. In this case care must be taken to determine whether a particular periphrasis with habban/hebbian and a past participle is to be taken in the stative or the perfect sense, an issue which has been central to many previous studies (e.g. Hoffmann 1934; Wischer 2002; for further discussion see Mitchell 1985, I, 292–8). Potential tests for discriminating between the earlier and later stages of grammaticalization may be broadly divided into two groups, morphosyntactic and semantic; of the potential morphosyntactic criteria for determining grammaticalization, two possible indicators have commonly been discussed: participle agreement and word order (see e.g. Wischer 2002, 244–5). Regarding participle agreement, however, there seem to be few generalizations that can be made about the situation in Old English. Uninflected participles are found in constructions in which they clearly have a purely adjectival function, and which are therefore statives rather than perfects, while inflected participles are found in constructions which must be interpreted as having a perfect meaning, as the following examples show (from Wischer 2002, 246): (75) Gyf he ænigne gylt ungebet if he-NOM any-MASC.ACC.SG guilt-ACC un.remedy-PA.PPL hæfð[…] have-3SG.PRES ‘If he has any sin unremedied…’ (76) Loca nu; þin agen look-2SG.IMP now thy- MASC.NOM.SG own-MASC.NOM.SG geleafa þe hæfð gehæledne. belief-NOM thee-ACC have-3SG.PRES heal-PA.PPL-MASC.ACC.SG ‘Look now: your own belief has healed you.’ Inflected participles of the sort seen in (76) are generally less common than uninflected forms, with which they are often found coordinated; inflected 113 participles do not occur in cases other than accusative, and may fail to agree with the noun in gender (Mitchell 1985, I, 282–98). Although generalizations may be made regarding the constructions in which inflected participles are most likely to occur, and the changes in their distribution over time, the extreme variability found in participle inflection, even among examples from the same author and text, has led some to suggest that even in the earliest Old English texts for which we have evidence participial agreement may no longer have been fully productive (e.g. Wischer 2002, 244–5); even if a single productive grammatical system were responsible for producing the range of recorded forms, it is certainly too poorly understood for participle inflection to be used as a reliable test of grammaticalization status. In Old Saxon, the situation is not dissimilar to that in Old English. Although Rauch (1992, 162–4) follows Lussky (1921) in adopting as a rule of thumb the association of inflected participles with the stative construction, she notes that the occurrence of zero inflection in Old Saxon and the restricted number of cases syntactically permissible in such constructions result in a substantial number of indeterminate forms, and that there remain a number of constructions with patterns of participle inflection that appear exceptional or semantically ambiguous. 3 Even apart from this ambiguity in the data, the value of participle agreement as an indicator of syntactic structures characteristic of the primitive state may be questioned. It might be observed that in grammatical categories such as the Modern French passé composé, which are undoubtedly the product of a relatively advanced stage of grammaticalization, participle agreement persists with far greater regularity than in the perfects of Old English and Old 3 See (43) above for an example of an Old Saxon perfect with inflected participle. 114 Saxon; it would certainly be unwise to infer from the presence of agreement in such a case that the auxiliary in these periphrastic tenses still has the syntax of an ordinary transitive verb. Word order is the other morphosyntactic criterion commonly suggested as a test for identifying stages of grammaticalization. In Old Saxon, the stative constructions and the periphrastic perfect constructions have been said to exhibit in general the same word order (Rauch 1992, 164–9); because of this formal identity between the two categories that the present study attempts to distinguish, word order is of little use as a test for this language. In Old English, to a greater extent than in Old Saxon, the order of constituents in such constructions could vary considerably based on the interaction of a number of factors (see Mitchell 1985, I, 282–3); however, it has sometimes been suggested that the order have + participle + object, similar to that found in Modern English, would be more compatible with the underlying syntax of the grammaticalized state than the other word orders found (e.g. Wischer 2002, 244). It should nevertheless be said that word order, as a morphosyntactic criterion for determining the stage of grammaticalization shown by a construction, depends for its own validity upon the outcome of semantic tests; to take a hypothetical example, if periphrases with perfect meaning showed a negative correlation with the modern word order and a positive correlation with other word orders, this would seem to show that the appearance of the modern word order in participial constructions with habban was not diagnostic of the grammaticalization of these constructions at the period in question. If the syntax of the language is well enough understood that possibilities of this sort can generally be ruled out a priori, the evaluation of new 115 cases is not dependent on semantics to such a degree. However, the syntax of Old English in particular exhibits a number of complexities, and the interpretation of the syntactic structures most relevant to the questions investigated here is bound up with a number of other issues, such as the shift from OV to VO order, about many aspects of which a consensus has not yet been reached (see e.g. Mitchell 1985, 1, 282–91; Fuss and Trips 2002). In the absence of more conclusive syntactic evidence for Old English and the interrelationship of the existing evidence with semantic factors, it may seem more fruitful to give more emphasis to semantic criteria in determining the degree of grammaticalization attained by a given form; as will be seen, semantic criteria have been favoured in many previous studies. Semantic tests for determining the degree of grammaticalization represented by a particular construction are of essentially two types. If the construction can be shown to be semantically incompatible with the sense of the earlier stage, it can be taken to represent the later stage; conversely, if the construction can be shown to be semantically incompatible with the sense of the later stage, it can be taken to represent the earlier stage. For tests of the latter sort there are numerous samples to serve as illustrations, involving stative constructions for which the temporal reference is clearly different from that of the corresponding present perfect or pluperfect: (77) He þæt […] weorð hæfde þa gyt he-NOM the-NEUT.ACC.SG worth-ACC have-3SG.PRET then yet on his cyste gehealden. on he-GEN chest-DAT hold-PA.PPL (GD MS C I.XXIV.64.5) ‘He still had the purchase-money then, kept in his chest.’ 116 It is clear in this example that the temporal value of gyt, which indicates that the event denoted by the participle still continued, would be semantically incompatible with the notion of anteriority that would be present if the construction with hæfde were understood as a pluperfect, and that it must therefore be understood in a stative sense. The existence of stative constructions such as this, although an important factor that must be borne in mind during any attempt to quantify the distribution of the periphrastic perfect, nevertheless has relatively little significance for the determination of the stage of grammaticalization attained by the periphrastic perfects in a given text, due to the aforementioned coexistence of the stative and perfect periphrases at the same periods. More meaningful information is provided by tests that look for constructions that are semantically incompatible with earlier stages of grammaticalization, and thereby establish the existence in the grammar of innovative forms that did not exist at earlier periods. For the present purposes, such a test would consist in determining whether a given occurrence of the verb habban is used in a way incompatible with the meaning of the original lexical verb. In some cases, as in the following example, syntactic cues simplify the task of identification: (78) Þa hie […] þær to gewicod hæfdon. þa when they-NOM there to encamp-PA.PPL have-3PL.PRET then onget se here þæt hie ne realize-3SG.PRET the-MASC.NOM.SG host-NOM that they-NOM NEG mehton þa scypu ut brengan […]. may-3PL.PAST the- NOM.PL ship-NOM.PL out bring.INF (ChronA1 II 896.12, p. 89) ‘When they had encamped for this, then the army perceived that they could not bring the ships out.’ 117 In (78), the pluperfect is formed from an intransitive verb which therefore lacks an object that could be construed as dependent on lexical habban, and the task of evaluation is not particularly difficult. However, in the absence of such syntactic evidence such evaluations are generally dependent on semantic judgements of a more involved nature. One issue complicating the question of whether a given meaning may be seen as semantically compatible with the lexical verbs habban/hebbian is the fact that verbs such as these develop easily into auxiliaries precisely because of their broad range of often abstract meanings (see Heine 1993, 30–2). A certain amount of variation may be seen in the approaches taken to this issue in previous studies; some authors, such as Wischer (2002), consider have to be a lexical verb only when it can be taken as referring literally to physical possession, while others, including Carey (1994) take the semantic range of have as a lexical verb to include more abstract senses, such as that of holding something in a specified state. It can be seen that the definition adopted may have a significant effect on the findings of a study; too strict a definition for the lexical verb have will result in false positive identifications of the grammaticalized construction, while too loose a definition will result in false identifications of the original stative construction. The position of the present study is to adopt a relatively broad definition for have, more similar to Carey’s than to Wischer’s; this position would seem to be supported by lexicographical evidence, such as that provided by the OED and Sehrt (1966), regarding the usages observed in the relevant languages. Some implications of this decision may be seen with the aid of the following examples: 118 (79) Ða Swiðberht hefde bisscophade onfongen, when Swithberht have-3SG.PRET bishop.hood-ACC receive-PA.PPL þa ferde he eft of Breotene […]. then fare-3SG.PAST he-NOM back of Britain-DAT (Bede XII.420.15) ‘When Swithberht had received the episcopate, then he travelled back from Britain.’ (80) Hie alle on þone Cyning they-NOM all-NOM.PL on the-MASC.ACC.SG king-ACC wærun feohtende oþ þæt hie hine be-3PL.PRET fight-PRES.PPL until that they-NOM he-ACC ofslægenne hæfdon[…]. slay-PA.PPL-MASC.ACC.SG have-3PL.PRET (ChronA1 I 755.16, p. 46) ‘They were all upon the King, fighting, until they had slain him.’ The translations given above represent the periphrastic constructions in these sentences as equivalent in meaning to the Modern English pluperfect. However, in both these cases it may be asked whether the periphrasis must be interpreted as a grammaticalized pluperfect or whether it could have another sense. These sentences might alternatively be interpreted in such a way as to give translations such as ‘had the episcopate received’ or ‘had the received episcopate’ for ‎(79) and ‘had him slain’ for ‎(80). Interpretations such as these depend on the range of meanings that can permissibly be assigned to have if it is not to be interpreted as an auxiliary; if have is considered to refer only to literal possession, then only ‎(79), and not ‎(80), would have even the possibility of being interpreted as anything other than a pluperfect. As stated above, the position taken in the present work is that both the alternative translations given above fall within the observed semantic range of have. Once it is established that for a given construction either the stative or the perfect interpretation is tenable, there remains to be addressed the question of 119 which interpretation to adopt. The position adopted in the present study is that at any stage when the grammaticalized perfect is known to exist in the language, indeterminate forms cannot be excluded from consideration and should be counted. If, as Wischer (2002) suggests, the development of these constructions in Old English had reached the ‘isolating stage’ (see Section 1.2.2 above), this indeterminacy, rather than being merely a limitation imposed by the available data, would have existed for the speakers themselves and been a major factor in the evolution of these constructions; similarly, in the case of Old Saxon Rauch (1992, 162) does not exclude the possibility of a certain degree of ambiguity within the language itself. Although the formal ambiguity in English eventually decreased, owing to the development of visible syntactic differences between the stative and perfect constructions, attempting to determine precisely the extent to which ambiguity was presented at this period by the periphrastic constructions is often difficult. Statives and perfects often have similar effects upon the truth conditions of a sentence; even in Modern English it is frequently the case that if one has done something, one has it done. There are of course a number of conditions, such as those discussed above, under which the two types of periphrastic construction can be distinguished semantically, but such conditions do not always occur with great frequency in texts, and their occurrence may vary for reasons unconnected to the developments of the constructions themselves. Under these circumstances, it is felt that to exclude large numbers of constructions because of their potential ambiguity would be inadvisable. The effects of this decision can be illustrated best through the use of relevant examples such as the following: 120 (81) Þær wæs micel ungeþuærnes […] & there be-3SG.PRET much-NOM unrest-NOM and hie hæfdun heora cyning they-NOM have-3PL.PRET they-GEN king-ACC aworpenne Osbryht, & off.throw-PA.PPL-MASC.ACC.SG Osbert-ACC and ungecyndne cyning underfengon ællan[…]. unkind-MASC.ACC.SG king-ACC receive-3PL.PRET Ælla-ACC (ChronA1 I 867.1) ‘There was great unrest, and they had cast out their king, Osbert, and accepted an alien king, Ælla.’ By some authors (e.g. Mitchell 1985, I, 295) this passage has been interpreted in a stative sense. However, using the approach underlying the present study it is necessary to question whether there are any grounds for such an interpretation. If, as discussed above, the presence of participial agreement and of a particular word order is not sufficient evidence to establish the construction as stative, and if periphrastic constructions of this form were available as expressions of pluperfect meaning, a pluperfect interpretation of this construction cannot be summarily rejected. Moreover, from a pragmatic perspective it might seem more likely that the author would have wished to talk about the act of casting out the king, which would be denoted by a pluperfect, rather than the subsequent keeping of him in an exiled state, which would be denoted by a stative construction. Accordingly, constructions such as (81) and (80) above have not been excluded from the count of periphrastic tense forms. 3.3.3.3 Unavailability of the Periphrastic Form The effect of semantic factors upon the availability of the periphrastic perfect has been addressed above; however, the distribution of these constructions was also influenced by morphological factors. In both Old English and Old Saxon 121 there were certain defective verbs which lacked a past participle and were therefore unable to form periphrastic perfects. The most prominent such verb is wesan/uuesan ‘be’; in Old Saxon the participle for this verb is unrecorded (see Sehrt 1966; Rauch 1992, xxxiii), while in Old English the corresponding participle is of very late emergence and seldom attested (see Mitchell 1985, I, 468). As wesan was not only a lexical verb in its own right but an auxiliary of the passive, this means that the periphrastic perfect was also unavailable for passives formed with wesan from any verb. The unavailability of the periphrastic perfect, due to this absence of a participle that could be used in such constructions, makes the use of the simple preterite in contexts in which a periphrastic form might otherwise be employed less significant than in the case of verb for which there was no morphological barrier to the use of either form. Accordingly, all occurrences of the preterite of wesan, whether as a main verb or an auxiliary, have been excluded from the data. Other defective verbs existed, including most Old English and Old Saxon modal verbs (e.g. Mitchell 1985, I, 416; Rauch 1992, 204– 5); however, the frequency of the other defective verbs in the data is so much less than that of wesan that the inclusion of preterites from these verbs was felt to be unlikely to introduce significant bias into the data. In addition to verbs of this sort, whose defective status is relatively secure, there are verbs for which no past participle is recorded, but for which there is no reason to suppose that this absence is the result of the verbs’ morphologically defective nature rather than simply because of the limitations of the textual record; as the onus would normally be to demonstrate the defective nature of the verb, no special treatment of these verbs has been made either. However, for a verb such as wesan, which is not only 122 known to be defective but is extremely frequent in occurrence, it was felt that some methodological recognition of these circumstances was warranted. 3.4 Additional Variables 3.4.1 Discourse Context It was said in Section 3.2 above that genre differences may produce variation in the extent to which primary and secondary tenses are found in different types of text, resulting in corresponding asymmetries in the distribution of the present perfect and the pluperfect. Accordingly, a study that concentrates specifically on only one of these categories to provide evidence for determining the stage of development attained in particular texts by the periphrastic constructions may risk conflating the distributional patterns which reflect differences in the process of grammaticalization with those which are merely the effect of variation in discourse style. Any observed association of a particular form with a particular discourse style, such as dialogue, may be due in part to factors such as to the sequence of tenses, which would operate differently on the present perfect and on the pluperfect, and in part to factors such as register, which may apply equally to both tenses (see Mitchell 1985, I, 281). Discrimination between these two types of factors is especially important in studying languages such as Old English and Old Saxon, where there is a contrast between the perfect tenses and the simple preterite not only as expressions of distinct semantic content but as innovative and conservative forms for the expression of the same meaning, and where the two formal categories may therefore be distributed differently in different registers. Furthermore, in comparing different languages, even those 123 whose verbal systems are similarly structured and at similar stages of development, individual grammatical forms may differ in their association with the various styles of discourse, and such cross-linguistic variation may be the result of semantic differences between the tenses themselves or alternatively reflect pragmatic differences between the two languages; such cross-linguistic differences may relate either to the pragmatic content of individual grammatical forms, such as was discussed above with reference to the Modern English perfect, or to the broader pragmatic practices of the languages as a whole. In order to avoid conflating factors differing in their ultimate origins, it is necessary to identify the discourse contexts in which verb forms occur. The importance of identifying and categorizing such contextual factors has been recognized in much previous work (see e.g. Fleischman 1990, 52–63 on French; Zeman 2010, 16–40 on Middle High German); the system of categorization used here has been devised specifically for the present study, with reference to the factors most likely to influence context-related variation. Each token of a relevant preterite or periphrastic form has been coded for discourse type using these categories, which are described below. 3.4.1.1 Direct Speech The first of the five discourse types distinguished within this system of categorization is direct speech. This category is used for contexts in which the utterance of another speaker is reported directly, with its original temporal reference and other deictic properties preserved intact rather than modified to fit the perspective of the surrounding text, as the following example shows: 124 (82) Siu […] quað that siu uuâri mid shei-NOM say-3SG.PRET that shej-NOM be-3SG.PRET.SUBJ with suhtiun bifangen: ‘bedrogan habbiad sickness-DAT.PL seize-PA.PPL suffer-PA.PPL have-3SG.PRES sie dernea uuihti. […]’ shej-NOM evil-ACC.PL wight-ACC.PL (Heliand XXXVI.109.2986) ‘She said that she was seized by sickness: “She has suffered from evil beings.”’ In this example, the phrase bedrogan habbiad occurs in a passage of direct speech, which stands as a separate quotation clearly set off from the main narrative and identified by a speech verb. The category of direct speech is in general relatively sharply delineated and easy to identify; however, a translation device used in Old English introduces a potential complication. This device, found in texts such as the works of St. Gregory and of Orosius, involves the insertion of phrases such as cwæð Orosius into translations of texts that in the original Latin were written from the first-person perspective of the Latin author, in order to make explicit the secondary nature of the translated text (see further Godden 2004, 7–8), and has the effect of essentially turning entire works into passages of direct discourse. Rather than use the direct discourse category for all verb tokens throughout such works, the decision has been made here to consider insertions such as these to be parenthetical in nature; adopting this position allows the textual variations present in the original document to be analysed and to determine the extent to which they are reflected in the Old English translation. Conversely, in the case of Boethius the dramatized dialogue format of the text was felt to be an integral part of the work; accordingly, verb forms from this dialogue have been characterized here as direct speech. 125 3.4.1.2 Personal Discourse The second discourse category distinguished within the present study is illustrated by example (83) below. This category comprises those contexts in which the author directly addresses another party within the main text, rather than in an isolated quotation as in the case of direct discourse; it includes dialogue of the sort found in works such as the Dialogues, as well as apostrophe, as in the following example: (83) Þu, fæder Agustinus, hie hæfst thou-NOM father-NOM Augustine-NOM they-ACC have-2SG.PRES on þinum bocum sweotole gesæd […]. on thine-DAT.PL book-DAT.PL clearly say-PA.PPL (Oros III.3.102.22) ‘You, Father Augustine, have clearly said those things in your books.’ Examples of this type resemble those of the preceding category in their time reference, tending to make more frequent use of primary tenses, for reasons that will be made clear in the discussion of the remaining categories; as a result, the two categories might be expected to be affected similarly by any factors related to the use of different sequences of tenses. However, an important difference between the two categories is that contexts falling into this category generally show greater stylistic unity with the surrounding text than is necessarily the case for direct discourse; accordingly, the two categories have the potential to be affected differently by any stylistic factors influencing the phenomena being studied. 3.4.1.3 Exposition The third discourse category identified here comprises utterances addressed by the author to his audience in general, rather than to a specific 126 interlocutor as in the previous example. This is a broad category which includes sentences such as the following: (84) Nan cræft nis to læranne no-MASC.NOM.SG craft-NOM NEG.be-3SG.PRES to teach-INF.DAT ðæm ðe hine ær geornlice ne leornode […]. that.MASC.DAT.SG REL he-ACC ere willingly NEG learn.3SG.PRET (CP I.25.15) ‘There is no power to teach him who before now has never willingly learned.’ Much homiletic and expository material of a diverse nature falls within this category. As might be expected, sentences from this category are similar in their temporal reference to those from the preceding category, given that the primary criterion distinguishing the two is in the audience to which they are addressed rather than in their content. However, while the utterances from this previous category can be clearly identified as being within a dialogue context, it is often difficult to make such assessments about utterances from this category; some utterances from within this category may have been conceived as purely impersonal exposition, whereas in other contexts, such as that of a homily, the author may have envisioned the text as something that might be spoken to a congregation in a less impersonal manner. The morphosyntactic consequences of any stylistic effects of this difference between these two categories would be difficult to predict a priori; making a distinction between the two categories should allow any such consequences that might exist to be discerned in the statistical analysis. 3.4.1.4 Narrative and Indirect Speech The final two categories distinguished in this study are those of historical narrative and indirect speech. Strictly speaking these constitute subsets of the 127 preceding category, being addressed to the author’s audience in general, but the distinctive properties of these discourse types merit their separate treatment. The category of indirect speech can be seen in the following example: (85) Thea uuîson man […] quâdun that the-MASC.NOM.PL wise-NOM.PL man-NOM.PL say-3PL.PRET that sea ti im habdin giuuendit hugi[…]. they-NOM to he-DAT have-3PL.PRET.SUBJ turn-PA.PPL thought-ACC (Heliand VIII.31.687) ‘The wise men said that they had turned their thoughts towards Him.’ This category comprises utterances which, rather than preserving their original temporal and deictic reference, have been shifted to match those of the surrounding narrative. Indirect speech is not placed within the category of historical narrative, which is used for utterances that not only refer to events in the past, but form part of an extended passage relating a sequence of past events sharing the same temporal framework. Many of the examples given previously fall within the category of historical narrative, including (68), (69), and (71) above. Although these two categories may at first appear sharply distinct, there are a number of situations in which the boundary between the two is less clearly defined. First, there is the question of which forms of subordination are to be considered as falling with the category of indirect speech. Sentences with verbs of speaking, such as (85) above, are straightforward examples of this type, but in Old English and Old Saxon such sentences bear syntactic similarities to others with verbs referring to mental states and verbs of perception. In the present study, these constructions have been classed together; their similar behaviour with reference to temporal shifting and to the sequences of tenses used in them is seen 128 as justifying parallel treatment. Second, ambiguities of scope are found in some sentences, as the following example shows: (86) Þa Langbeardiscan men […] bædon, þæt the-NOM.PL Lombard-NOM.PL man-NOM.PL bid-3PL.PRET that he heom þone agæfe, þe he-NOM they-DAT that.MASC.ACC.SG give-3SG.PRES.SUBJ REL hi him ær befæston[…]. they-NOM he.DAT ere entrust-3PL.PRET (GD MS C III.XXXVII.253.27) ‘The Lombard men told him to give them back him whom they had previously entrusted to him.’ Although Modern English translations for such sentences must often commit to a specific interpretation, in the original text a greater degree of ambiguity exists; a sentence such as this can be interpreted as it is here, with the relative clause forming part of the indirect speech, but it can alternatively be construed as a report of an original utterance saying only, ‘Give him back to us,’ with subsequent explication within the main narrative of the identity of him. The resolution of such ambiguities is often a question of interpretation which must be carried out with reference to the context of the construction rather than to a specific list of criteria. 4 Despite the existence of such ambiguities between contexts of indirect speech and historical narrative, attempting to distinguish them as separate categories allows the possibility that stylistic differences may exist between the two categories to be investigated. If differences between the spoken and the written language have any effect on the distribution of the grammatical categories that form the subject of the present study, it is possible that the forms falling within the category of indirect speech will reflect the spoken norms to a greater 4 For a fuller discussion of the interpretative issues connected with such constructions see Mitchell (1985, II, 86–90; 112–20). 129 extent than historical narrative in general; although there is no necessity for indirect speech to bear a stylistic resemblance to the spoken language, the existence of such a resemblance is at least more likely than in a category with no connection to the spoken language of any sort. Although the discourse contexts distinguished within this study are not sharply bounded categories among which the position of a given utterance is always indisputable, they can nevertheless provide a means of measuring some of the pragmatic factors that may potentially influence the distribution of the grammatical categories being studied here. 3.4.2 Translation Practices In contrast to the Old Saxon texts analysed here, which retell Biblical narratives in a very loose form without attempting to provide a translation as such of a single text, many of the Old English works analysed here are translations of specific Latin works. The high proportion of translations among the surviving Old English prose texts has raised questions regarding the possible influence of Latin upon Old English syntax; this can be seen in matters such as the discussion of absolute participial constructions in Section 2.3.3. Although it is unlikely that the periphrastic perfect was borrowed wholesale from Latin (see Section 1.3 above), the possibility remains that the form of original Latin texts may have influenced the relative distribution of the periphrastic perfect and the preterite in one direction or another. In order to identify such potential influences, the present study records for each relevant Old English verb in a translated text the form of the Latin original to which it corresponds. Before describing the categorization system employed here, it may be useful to provide some additional background on 130 translation practices in Old English in general and in the texts analysed here in particular. A starting point for any consideration of Old English translation practices is provided by the words of St. Jerome, who wrote of translating the Scriptures ‘non verbum e verbo sed sensum […] de sensu’ (‘not word for word but sense for sense’) (Labourt 1953, 59; see further Liuzza 2000, 50). This view of translation as a practice was known to and echoed by many Old English translators; in King Alfred’s preface to the Pastoral Care, he speaks of turning the original into English ‘hwilum word be worde, hwilum andgit of andgite’ (‘sometimes word by word, sometimes sense for sense’) (CP 7.19), while similar views of the translation process are expressed in the works of Ælfric (see e.g. Minkoff 1976). Although St. Jerome’s words on translation were presumably familiar and respected among writers of Old English, even among modern authors there exists a certain amount of dispute as to how they are to be interpreted (e.g. Minkoff 1976; Liuzza 2000), and it will be seen that a wide variety of translation practices obtained in the Anglo-Saxon context in which these allusions were made. Old English translations differ widely in their fidelity to the original Latin text. Among the most literal translations is that of the Gospels; although the translation is seldom so literal as to be unidiomatic, the Old English text adheres closely to the Latin both in content and in form, resulting in ‘a phrase-by-phrase rendering of the Latin gospels without summary, explication, or ornament’ (Liuzza 2000, 50–51). Where departures from the Latin original occur, they most commonly take the form of minor narrative additions, such as clarifications of pronoun reference (see Liuzza 2000, 51–82). It is perhaps only to be expected 131 that a Biblical translation would be more literal than translations of other texts of a less sacred nature, for which preserving the exact form of the original might seem less important than merely conveying similar content. However, the literalism of the Gospels is greater than that of Ælfric’s Old Testament translations, which in some passages condense and reorder the Biblical text to a considerable degree (see e.g. Marsden 2008); it has been suggested that Ælfric was more concerned with rendering the ideas of the original text in a manner designed to avoid misconceptions than with the literal rendering of individual words and phrases (Clemoes 1966). Another translation that is comparatively faithful to the original text is the Pastoral Care, a work which, however, shows greater fidelity in content than in form; this text makes much more frequent use of paraphrase than the Gospels, often differing in its syntax from the original text even where there is no apparent reason for avoiding a literal translation, and expanding the original text for the sake of clarity (see Sweet 1871, xli). Varying interpretations have been placed on the greater preference shown in this text for literal translation than in other contemporary works; this has been seen as a sign of inexperience on the part of the translator (Godden 2007, 13), of a desire to provide the plainest and least ambiguous rendering possible for an uneducated audience (Sweet 1871, xli), or simply of an original text that was in less need of correction and elucidation than other Latin texts being translated (see Bately 2009, 191). Whatever the reason, it will be seen in Chapter 4 that the translations of the Gospels and the Pastoral Care were far more faithful to the original text than other Old English translations. 132 Many Old English translations deviate from the original text to a substantial degree. A relatively minor example is the case of Bede, in which these alterations mainly take the form of omissions; particular sections and topics are consistently omitted, producing a translated text which, despite some resulting minor inconsistencies, generally presents a seamless whole (see Miller 1890, lvii– lix; Whitelock 1962, 61–2). For some of the omissions, political motives have been adduced (Miller 1890, lvii–lix); however, a substantial proportion of the omitted material deals with general, uncontroversial topics such as geography, and the abridgement of the Latin original has been interpreted simply as a sign of narrower interests on the part of the translator (Whitelock 1962, 64). Where no such omissions have taken place, however, the Old English text of Bede closely reflects its Latin original; the translation is so literal as to have been seen as almost stilted at times, although some passages are more flexible and idiomatic in style and take a commensurately freer approach to translation (see Whitelock 1962, 75–7). In comparison with Bede, the text of Orosius stands further from its Latin original. Not only does the Old English version of this text abridge the Latin original freely, but it adds a considerable amount of material from additional sources, both within the body of the text and in the geographical preface peculiar to the Old English translation; these omissions and interpolations have a noticeable effect on the overall tone and emphasis of the work, broadening its scope from a comparison of the Roman Empire before and after the advent of Christianity to a more general historical work (see Bately 1980, xciii–c). The Old English Boethius is at least as far removed from its Latin original. In many places the philosophical discussion of the original text is considerably simplified, while 133 in others there are substantial changes in argument, and as with Orosius, the changes made in the process of translation affect the emphasis of the work as a whole; one of the most noticeable differences between the Old English text and its Latin counterpart is that in the original De Consolatione Philosophiae Boethius, who had also written on topics of Christian theology, seems to have intended to write a more secular work, whose speculations regarding religious matters avoid committing themselves to a particular system of belief and can be interpreted in terms of Neo-Platonism as easily as of Christianity, while its Old English translation situates the discussion within a more explicitly Christian framework (see e.g. Godden and Irvine 2009, 61–4). The differences between the two texts have variously been viewed as the result of imperfect understanding of the Latin original on the part of the translator and as the result of great familiarity with the original and its subject matter coupled with a differing perspective (see Sedgefield 1899, xxv–xxxv; Godden and Irvine 2009, I, 50–61). Even where Old English translations differ from the original to such an extent, there is seldom any distinction drawn within the text between original material and interpolations. In fact, interpolated material is often explicitly attributed to the author of the original work; for example, in Orosius interpolated passages and translated material alike are often introduced with the phrase cwæð Orosius (‘said Orosius’) (see Godden 2004, 7–8). A full analysis of the range of translation practices shown in Old English texts and their relationship to other aspects of the Anglo-Saxon cultural milieu would require a lengthy historical study; from a linguistic perspective, the range of variation shown among Old English translations means that from the fact alone that an Old English text was translated from Latin it is not possible to 134 predict the degree to which the syntax and semantics of the Old English text reflect the Latin original, let alone the extent to which the latter may have produced some departure from the native idiom. The range of approaches available to Old English translators is further illustrated by texts of which the original translation has undergone further revision. One such text is the Dialogues of Gregory the Great, which exists both in the original ninth-century translation and in a revised form from approximately a century later; while the original text was an extremely literal rendering of the Latin, the revision modernized the language, making it more idiomatic, and also made use of the Latin original to correct errors, providing in this way a more faithful representation of the original content (see Yerkes 1982). Another translated text that underwent revision is Boethius. The original Latin text was divided into alternating sections of prose and verse, which in the earliest Old English version were rendered as prose throughout; at a somewhat later date, a version was produced in which most of the sections corresponding to verse in the Latin version were turned into Old English verse, primarily on the basis of the Old English prose translation rather than the Latin original (see Godden and Irvine 2009, 44–9). The evidence for whether both versions were produced by the same writer is not entirely conclusive (Bately 2007; Godden and Irvine 2009). In contrast to the Dialogues, most of the differences between the two versions are related to the stylistic markers associated with poetry as a genre; in the metrical version of Boethius, elements of poetic diction are more frequent in passages translating Latin verse than in the corresponding passages of the prose version (Godden and Irvine 2009, 44). Such alterations further reveal the wide variation 135 among Old English texts translated from Latin; in any comparison of original texts and translations, it is important to note that status as a translation does not denote a single, consistent relationship among texts. 3.4.2.1 Introduction In order to analyse the relationship between Latin tenses and the Old English grammatical forms used to translate them, for each Old English preterite or periphrasis in a translated text the Latin grammatical category to which it corresponded was noted. The following categories, whose signification will be described below in greater detail, were used to classify the original Latin forms: ‘perfect’, ‘pluperfect’, ‘perfect participle’, ‘imperfect’, ‘present’, and ‘other’. As one of the questions being examined in the present study is whether the choice of form in Old English is related to a desire for literal closeness to a Latin original, the above categories are used only for cases of relatively literal translation, and in other contexts four further categories are used: ‘interpolation’, used for new material added in the Old English text with no equivalent in the original Latin; ‘recast’, used when changes have been made in the Old English text which affect the temporal content while leaving other components of the meaning relatively unaltered; ‘interchange’, used in cases of more substantial alterations; and ‘expansion’, used when a single Latin verb is rendered into Old English by two nearly synonymous verbs. It should also be noted that the exceptionally large and diverse manuscript tradition through which Biblical texts were transmitted gives especial prominence in the case of Biblical translations to the possibility that differences between the Old English text and modern editions of the Vulgate may result from differences in the Latin original used by the translator; however, none 136 of the verb forms analysed in the present study are included in the discussions by Marsden (1995, 395–419) and Liuzza (2000, 26–49) of passages where variation in the Latin texts may be at issue. The exact definitions of the categories used in this work to categorize translated forms may best be understood with the aid of the following examples and discussion. 3.4.2.2 Perfect The definition of the ‘perfect’ category is simple; it indicates that an Old English form translates a Latin perfect tense. As discussed in Section 2.5.2.2, the Latin perfect was polysemous, having both a perfect and a perfective (or aorist) meaning (e.g. Gildersleeve and Lodge 1885, 159–60), and therefore instances in which the Latin perfect was translated by an Old English preterite provide little information about the precise signification of the Old English verb; because the semantic range of the Old English preterite encompassed both the senses of the Latin perfect, such a translation provides no additional information as to the exact sense in which the Latin verb was understood by the translator and therefore no additional support to the interpretation in the present study of the preterite in question as perfect-like in its semantics. In cases in which a Latin perfect is translated by an Old English periphrastic perfect, as in the following sample, this provides greater evidence that the Latin text was understood by the Old English translator to be semantically a perfect: (87) In primo autem parente didicimus […]. in first-MASC.ABL.SG moreover parent-ABL.SG learn-1PL.PF (CP III.29, II, p. 474) ‘Moreover, by the example of our first parent we have learned […].’ 137 We habbað geascod from we-NOM have.3PL.PRES discover-PA.PPL from urum ærestan mæge Adame […]. our-MASC.DAT.SG erst-DAT kinsman-DAT Adam-DAT (CP III.417.19) ‘We have found out from our first kinsman Adam […].’ Alternatively, a Latin perfect may be translated either by an Old English periphrastic pluperfect or by a preterite considered to be pluperfect-like in its semantics, a phenomenon for which the explanation may vary from one case to another. In Latin, as in English, there were some contexts in which it was not necessary to mark temporal anteriority explicitly through the use of a pluperfect: (88) Quod postquam indicavit, adiunxit […] REL-NEUT.ACC.SG after indicate-3SG.PF adjoin.3SG.PF (GD IV.XX.2, p. 366) ‘After he said which, he added […].’ Þa æfter þam þe he hit him then after that-NEUT.DAT.SG REL he-NOM it-NOM he-DAT gesæd hæfde, he cwæð say-PA.PPL have-3SG.PRET he-NOM speak-3SG.PRET (GD MS C IV.20.291.1) ‘Then after he had said it to him, he said […].’ In examples such as this, which make use of a periphrastic construction, the relevant temporal relationship is marked more explicitly. 3.4.2.3 Pluperfect Another straightforward category is ‘pluperfect’, which is used for Latin pluperfect forms; as might be expected, there are no instances of this Latin tense being translated by an Old English form classed as perfect. The Latin pluperfect, having a narrower semantic range than the Latin perfect, provides more information about the temporal relationships intended to be expressed by an Old 138 English form translating it. As the figures below show, the Latin pluperfect is frequently rendered by the Old English preterite, as in the following example: (89) Sucessit Augustino in episcopatum Laurentius, succeed-3SG.PF Augustine-DAT in episcopate-ACC Laurentius-NOM quem ipse idcirco adhuc REL-MASC.ACC.SG self-MASC.NOM.SG thereto still vivens ordinaverat living-NOM.SG ordain-3SG.PLP (Bede II.IV, I, p.218) ‘Laurentius, whom Augustine had consecrated for the purpose while still living, succeeded him in the episcopate.’ Æfter Agustini fylgide in biscophade after Augustine-DAT follow-3SG.PRET in bishop.hood-DAT Laurentius, þone he forðon Laurentius-NOM REL-MASC.ACC.SG he-NOM for.that bi him lifigende gehalgode by him-DAT live-PRES.PPL-MASC.DAT.SG hallow-3SG.PRET (Bede II.IV.106.18) ‘Laurentius, whom Augustine consecrated for the purpose while living, succeeded him in the episcopate.’ However, there are other instances in which a Latin pluperfect is translated by an Old English periphrastic pluperfect, as in the following example: (90) Nec idolis ultra servivit, ex quo not.and idol-DAT.PL further serve-3SG.PF out rel-NEUT.ABL.SG se Christo servitur esse self-ACC Christ-DAT serve-FUT.PPL-MASC.ACC.SG be-INF promiserat promise-3SG.PLP (Bede, II.IX, I, p. 252) ‘And he served idols no longer, after he had promised to serve Christ.’ Ofer þæt deofolgeldum ne þeowode, seoðþan over that-ACC devil.yield-DAT.PL not serve-3SG.PRET since he hine to Cristes þeowdome he-NOM him-ACC to Christ-GEN service-DAT 139 gehatenne hæfde promise-PA.PPL-MASC.ACC.SG have-3SG.PRET (Bede II.VIII.124.16) ‘After that he never practised devil-worship, since he had dedicated himself to Christ’s service.’ As will be seen in Chapter 4, there is variation among individual texts in their practices for the translation of such forms. One issue that should be mentioned arising in the assignment of Latin forms to the perfect and pluperfect categories concerns the passive of the perfect and pluperfect. In Classical Latin, the passive of these tenses was formed with the perfect participle and the present and imperfect respectively of sum, the participle itself being sufficient to mark the construction as perfect; as a result, passive constructions had forms such as amatus est ‘he has been loved’ and amatus erat ‘he had been loved’. However, with the decline of the inflected passive in Late Latin, the tendency arose to use these periphrastic forms as passives of the present and imperfect, meaning ‘he is loved’ and ‘he was loved’ respectively, and to express the old meanings using new formations with the auxiliary in the perfect and pluperfect (such as amatus fuit and amatus fuerat); such forms are also found in Medieval Latin (see e.g. Stotz 1998, 329). Accordingly, there may sometimes be ambiguity as to whether a construction such as amatus erat in a Medieval Latin text should be interpreted in the Classical sense, as a pluperfect, or in the later sense, as an imperfect; although genuinely ambiguous cases occur infrequently in the data for the present study, the Classical interpretation has here been adopted as the default. 140 3.4.2.4 Imperfect The category ‘imperfect’ is similarly easy to define, being used for instances in which the relevant Old English form translates a Latin imperfect; however, the semantic differences between the imperfect and the perfect and pluperfect may raise the question of why such translations should be found by the present study. Again, explanations may vary from one case to another for the correspondence of a Latin imperfect to an Old English verb form considered to be semantically perfect-like or pluperfect-like. In some cases, this reflects a Latin rule regarding the sequence of tenses in which an imperfect subjunctive may be used to refer to an anterior event: (91) Hic cum audisset quia Iesus this-MASC.NOM.SG when hear-3SG.PLP.SUBJ because Jesus-NOM adveniret […] in Galilaeam abiit ad eum to.come-3SG.IMPF.SUBJ in Galilee-ACC off.go-3SG.PF to he-ACC (Vulg John 4. 47) ‘When this man had heard that Jesus had come (*came) to Galilee, he went to Him.’ Þa se gehyrde þæt se when that-MASC.NOM.SG hear-3SG.PRET that the-MASC.NOM.SG Hælend for […] to Galilea he com to healer-NOM fare-3SG.PRET to Galilee he-NOM come-3SG.PRET to him he-DAT (WS John 4. 47) ‘When that man heard that the Saviour had travelled (*travelled) to Galilee, he came to Him.’ In the above example the event denoted by Old English for clearly meets the semantic criteria for the pluperfect domain, despite the correspondence of the Old English verb to a Latin imperfect subjunctive. In other cases, the inclusion by these criteria of other Old English verb forms used to translate Latin imperfects 141 may result from the possibility of conceiving an event either as a terminated action or as an ongoing state. (92) Aufugerunt omnes qui me off.flee-3SG.PF all-MASC.NOM.PL REL-MASC.NOM.PL I-ACC forcipibus rapere quaerebant spiritus forceps-ABL.PL seize-INF seek-3PL.IMPF spirit-NOM.PL (Bede V.XII, II, p. 260) ‘All the spirits that sought to seize me with tongs fled away.’ Onweg flugon ealle ða […] away flee-3PL.PRET all-NOM.PL the-NOM.PL gastas […] ðe me mid […] tangan tobeotodan ghost-NOM.PL REL I-ACC with tong-DAT.PL beat-3PL.PRET (Bede V.XIII.428.22) ‘Away fled all the ghosts that had beaten/were beating me with tongs.’ In this example, the act of beating may be envisioned as something that was terminated by the act of fleeing or as something that was still in progress at an earlier point in time; the analysis of the Old English text in accordance with the principles described in Section 3.3.2.2 results in its inclusion in the data of the present study. The evidence from the Latin text may be taken to indicate that in these cases an imperfect-type reading of the Old English verb may in fact have been intended by the translator. However, other examples in which a Latin imperfect corresponds to an Old English periphrastic pluperfect (e.g. (100) below) show that in some cases translators did intend to represent events differently from the Latin original, and while it would be unwise to assume that such deviations from the Latin original were intended where no explicit evidence exists of the sort provided by the above example, the possibility that similar differences in perspective between the English and Latin authors may exist even when they are 142 not perceptible must be borne in mind. As will be seen below, where such evidence is unambiguously present, the ‘recast’ category is used. 3.4.2.5 Present The category ‘present’ is used when a Latin present tense is translated in Old English by a preterite or periphrastic construction. As might be expected, this category is only found for perfects, and not for pluperfects; it can be seen in the following example: (93) Huius multa miracula […] this-MASC.GEN.SG many-NEUT.ACC.PL miracle-ACC.PL Gaudentius presbiter narrat Gaudentius-NOM priest-NOM narrate-3SG.PRES (GD I.IX.1, p. 110) ‘The priest Gaudentius tells many miracles of this man’s.’ Þyses bisceopes manega wundru this-MASC.GEN.SG bishop-GEN.SG many-NOM.PL wonder-NOM.PL me rehte se mæssepreost I-ACC tell-3SG.PRET the-MASC.NOM.SG mass.priest-NOM Gaudentius. Gaudentius-NOM (GD MS H I.XX.56.20) ‘The priest Gaudentius has told me many miracles of this bishop’s.’ As discussed in Section 2.5.2.3 above, the use of the Old English preterite to denote an event that in the original Latin text was described as present provides support for the assumption in the present study that the semantic range of the Old English preterite could include events for which present relevance was salient, such as would be denoted in Modern English by a present perfect; if the preterite invariably represented events simply as past, with no further qualification, the Old English present might be expected to have appeared more suitable for a situation that was explicitly marked in the original texts as continuing. 143 (94) Hoc est quod this-NEUT.GEN.SG be-3SG.PRES what-NEUT.NOM.SG recolentem vehementius coquit revisit-PRES.PPL-MASC.ACC.SG vehemently-COMP cook-3SG.PRES (Boece II.P4.2) ‘This is what more vehemently troubles him who returns to it.’ Þæt me hæfð eallra that-NEUT.NOM.SG I-ACC have-3SG.PRES all-GEN.PL swiðost gedrefed severe-SUP trouble-PA.PPL (Boece X.258.6) ‘That has troubled me most severely of all.’ In cases such as this a Latin present is rendered in Old English by a periphrastic perfect; examples of this sort provide further support for the notion of semantic similarity between periphrastic perfects such as (94) and preterites such as (93). 3.4.2.6 Perfect Participle The category ‘perfect participle’ is used for constructions in which the Old English form in question translates a Latin perfect participle; when the finite verb with which the participle is construed is in a primary tense the participle has the sense of a perfect, while is has the sense of a pluperfect when the finite verb is in a secondary tense (see Gildersleeve and Lodge 1895, 426). The effect of such differences in tense can be seen in the following examples: (95) Devictis magnis hostis conquer-PA.PPL-MASC.ABL.PL great-ABL.PL host-ABL.PL mentem non erigant mind-ACC not erect-3SG.PRES.SUBJ (GD III.XVI.13, p. 250) ‘Great foes having been conquered, they shall not raise their spirits.’ Hi hyra mod upp ne aræran […] þeah they-NOM they-GEN mood-ACC up NEG rear-3PL.PRES.SUBJ though 144 þe hi habban heora feond […] that they-NOM have-3PL.PRES.SUBJ they-GEN fiend-ACC oferswiðede conquer.PA-PPL-ACC.PL (GD MS C III.XIV.204.20) ‘They shall not raise up their spirits though they may have conquered their enemy.’ (96) Contigit […] ut […] cum uno […] fratre[…] happen-3SG.PF that with one-MASC.ABL brother-ABL commoraretur, ceteris eius remain-3SG.IMPF.SUBJ.MP remaining-MASC.ABL.PL he-GEN sociis pro causa opportuna fellow-ABL.PL for cause-ABL opportune-FEM.ABL.SG ad ecclesiam reversis to church-ACC return-PA.PPL-MASC.ABL.PL (Bede IV.III, II, p. 18) ‘It happened that he was remaining with one brother, the rest of his fellows having gone to church as was suitable.’ Þa gelomp […] þæt he wæs […] mid then happen-3SG.PRET that he.NOM be-3SG.PRET with ane breðer wuniende. His oðre one-MASC.DAT brother-DAT dwell.PRES-PPL he-GEN other- NOM.PL geferan fore gelimplicum intingum fellow-NOM.PL for fitting-DAT.PL matter-DAT.PL hwurfon to cirican turn-3PL.PRET to church-DAT (Bede IV.III.262.29) ‘Then it happened that he was remaining with one brother. His other fellows had gone to church as was suitable.’ In both these examples, a Latin participle has been rendered by an Old English finite verbal phrase; a periphrastic perfect is used in the first case, in which the Latin finite verb is in the present, and a preterite of pluperfect meaning in the second, in which the finite verb is an imperfect. 145 3.4.2.7 Other The category ‘other’ is used for Latin grammatical forms that do not fit into any of the above categories. One use of this category is for the participial constructions discussed in Section 4.3.3.2 below. Another type of construction placed in this category is the Latin perfect infinitive; as the following examples show, constructions with the infinitive are generally similar in their syntax to the participial constructions described above, and are similarly dependent on a finite verb: (97) Dixit frequenter se cellam Equitii say-3SG.PF frequently self-ACC cell-ACC Equitius-GEN magicis artibus in area suspendisse magic-FEM.ABL.PL art-ABL.PL in air-ABL suspend-PF.INF (GD I.IV.6, p. 90) ‘He said that he had frequently suspended Equitius’ monastery in the air by magic arts.’ He […] sæde, þæt he mid his he-NOM say-3SG.PRET that he-NOM with he-GEN drycræfte gelomlice on þa lyfte wizard.craft-DAT habitually on the-FEM.ACC.SG air-DAT ahenge Æquities mynster hang-3SG.PRET.SUBJ Equitius-GEN minster-ACC (GD MS H I.VIII.30.18) ‘He said that he had frequently suspended Equitius’ monastery in the air by magic.’ Such infinitive constructions are much less common in the data than participial constructions, and their low frequency, coupled with their uneven distribution among the different texts, would preclude their analysis within a separate category from giving meaningful results. 146 3.4.2.8 Recast The category ‘recast’ is used for cases in which the temporal content of the Old English text is substantially different from that of the Latin, so that the Old English tense does not directly correspond to that of the Latin. This category is used for changes in tense such as a shift from present to past, as in the following example: (98) Ipsam quam […] impetrare potuit same-FEM.ACC.SG REL-FEM.ACC.SG impetrate-INF can-3SG.PF veniam contemnit indulgence.ACC contemn-3SG.PRES (CP III.30, II, p. 478) ‘He despises the same indulgence that he was able to obtain.’ He forhogde ða forgifnesse he-NOM despise-3SG.PRET the-FEM.ACC.SG forgiveness-ACC ðe he […] begiten hæfde REL he-NOM obtain-PA.PPL have-3SG.PRET (CP LIV.421.7) ‘He despised the forgiveness that he had obtained.’ In this case, the correspondence of the Old English periphrastic pluperfect to a Latin perfect is the result of a change in the tense of the main verb, from a present in Latin to a preterite in Old English; a similar effect is produced by other changes, such as shifts from direct to indirect speech. The ‘recast’ category is also used for the effects of changes in the subordination of events to other events, as in the following example: (99) Ruinae suae totas Graeciae ruin-DAT their-FEM.DAT.SG all-FEM.ACC.PL Greece-GEN vires inplicuerunt: qui, cum se force-ACC.PL enfold-3PL.PF REL-NOM.MASC.PL when self-ACC.PL exsecrationibus devovissent, sacramentisque execration-ABL.PL devote-3PL.PLP.SUBJ sacrament-ABL.PL.and obstrinxissent, domum nisi Messena bind-3PL.PLP.SUBJ home-ACC not.if Messena-ABL 147 expugnata, numquam esse defeat.PA-PPL.FEM.ABL.SG never be-INF redituros, […] eos[…] Spartam remittunt return.FT-PPL.MASC.ACC.PL them-ACC Sparta-ACC remit-3PL.PF (Oros I.XXI, p. 57) ‘They brought all the forces of Greece to their ruin, who, when they had devoted themselves with curses and bound themselves with oaths never to return home without Messena’s having been defeated, sent them back to Sparta [who had come after the oath- taking].’ Þa æt nihstan hie hæfden when at next-NEUT.DAT.SG they-NOM have-3PL.PRET.SUBJ getogen eal Creca folc to ðæm draw-PA.PPL all Greek-GEN.PL folk-ACC to the-NEUT.DAT.SG gewinnum þa Læcedemonia […] aðas strife-DAT then Lacedaemonian-NOM.PL oath-ACC.PL gesworan þæt hie næfre noldon swear-3PL.PRET that they-NOM never NEG.will-3PL.PRET æt ham cuman ær hie þæt at home-DAT come-INF ere they-NOM that-NEUT.ACC.SG gewrecen hæfden. […] Gecwædon þa […] wreak-PA.PPL have-3SG.PRET.SUBJ say-3PL.PRET then þæt þa ham gelendon that that-NOM.PL home-ACC land-3PL.PRES.SUBJ (Oros I.XIV.56.17) ‘Next, when they had drawn all the people of Greece into the war, the Lacedaemonians swore oaths that they would never go home before they had avenged it. They said that those [who had come after the oath-taking] would return home.’ From this example it can be seen that the two pluperfects in the Old English text refer to events denoted in Latin by a perfect and a participial clause, whereas the two pluperfects in the Latin are translated in Old English by a preterite that shares no semantic properties of the pluperfect and does not mark the event with respect to its anteriority to other events. Because the relationship between the Old English pluperfects and their Latin originals is so closely dependent on other changes to the syntax, they are placed in the ‘recast’ category rather than the 148 ‘perfect’ and ‘participle’ categories. Sentences such as (98) and (99) differ in this respect from previously discussed examples such as (93) and (94). In the latter two examples, the Old English and the Latin text refer to the same event, whose temporal relation to other elements of the discourse context is essentially the same in both languages; for example, in (93) the events of telling that have occurred are already in the past, and the present in Latin and the present perfect in Old English express the notion that these events are part of the present inasmuch as their future recurrence is not precluded. It is only where semantic similarity of this nature does not exist that verb forms are placed in the ‘recast’ category. 3.4.2.9 Interchange When the Old English text has undergone more substantial alteration from the Latin original, the ‘interchange’ category is used. An example of such a substantial alteration can be seen in the following example: (100) A legione, quae hominem tenebat, of legion-ABL REL-FEM.NOM.SG man-ACC hold-3SG.IMPF dictum est […] say-PA.PPL be-3SG.PRES (GD III.XXI.4, p. 276) ‘By the legion that was occupying the man it was said […].’ Wæs gecweden fram þam deofla be-3SG.PRET speak-PA.PPL from the-MASC.DAT.SG devil-GEN.PL heapa, þe þone […] man ofseten hæfde […] heap-DAT REL the-MASC.ACC.SG man-ACC beset-PA.PPL have- 3SG.PRET (GD MS C III.XXI.223.22) ‘It was said by the host of devils that had besieged the man […]’ In this example, whether hæfde is interpreted as an auxiliary or a lexical verb (‘… that had him besieged…’), the Old English translator has conceived of the event in terms of an anterior act of besieging rather than merely an ongoing state of 149 occupation. Such interchanges of cause and effect, which are not uncommon, result in Old English forms whose temporal content does not directly correspond to that of the Latin. (101) post culpam after fault-ACC.SG (CP III.30, II, p. 460) ‘after the sin’ æfterðæmðe hie gesyngod habbað after they-NOM sin-PA.PPL have-3PL.PRES (CP LII.405.23) ‘after they have sinned’ As the above example shows, this category is also used where an Old English verb translates a different part of speech; in this case, a noun in the Latin original is translated by a verb phrase in Old English. 3.4.2.10 Interpolation The ‘interpolation’ category, unlike the ‘interchange’ category, is used when the Old English verb form does not correspond to any wording in the original Latin text. This includes content that was not present in the original text in any form, but which was added by the translator; many such cases involve brief expository notes, as in the following example: (102) Malis ante noverat pie parcere bad-DAT.PL before know-3SG.PLP piously spare-INF (CP III.40, I, p. 140) ‘He previously knew how to spare the evil dutifully.’ David […] forbær ðæt he ðone David forbear-3SG.PRET that he-NOM the-MASC.ACC.SG kyning ne yfelode, ðe hine on sua king-ACC not harm-3SG.PRET REL he-ACC on so 150 heardum wræce gebrohte hard-NEUT.DAT.SG exile-DAT bring-3SG.PRET (CP III.37.3) ‘David forbore to do ill to the king who had brought him into such arduous exile.’ In the above example the relevant verb, gebrohte, occurs in material not present in the original text and is therefore placed in the ‘interpolation’ category. 3.4.2.11 Expansion As stated above, the category ‘expansion’ is used when a single Latin verb is rendered into Old English by two nearly synonymous verbs, as in the following example: (103) ista omnia quae vidisti that-NEUT.NOM.PL all-NEUT.NOM.PL REL-NEUT.ACC.PL see-2SG.PF (Bede V.XII, II, p. 262) ‘all these things that you have seen’ ðas þing ealle […] ðe þu this- NOM.PL thing-NOM.PL all-NOM.PL REL thou-NOM sceawadest & gesawa show-2SG.PRET and see-2SG.PRET (Bede V.XIII.430.29) ‘all these things that you have observed and seen’ In such cases, the first verb in the Old English passage is categorized according to the tense of the Latin verb, while the second verb is placed in the ‘expansion’ category. It was felt advisable to have a separate category for such cases to avoid biasing the data by counting multiple instances of the same translation, given that both verbs in such cases are generally of the same form; members of the ‘expansion’ category are therefore excluded from statistical analysis relating to translation practices. 151 3.5 Conclusion From the discussion in this chapter it should be possible to see the nature of the data on which the analysis in the present study is based. A selected body of Old English and Old Saxon texts is examined to produce a list of verb tokens, comprising periphrastic perfects, periphrastic pluperfects, and perfect-like and pluperfect-like preterites. The discourse context of each verb is also recorded, as is the category of the original Latin form that it renders, where applicable. These data are intended to provide information about the use and distribution of the two formal categories, the preterite and the periphrastic constructions, as expressions of similar perfect and pluperfect semantic content. In the next chapter the data obtained in this way will be described and the results of their statistical analysis discussed. 153 4. Results 4.1 Introduction In this chapter the results of the present research will be presented. The Old Saxon data will be discussed first; it will be seen that the Old English data are greater not only in their quantity but in their complexity, and that the Old Saxon data can provide a valuable point of reference in the interpretation of the Old English data. The relative distribution of the simple and periphrastic forms will be shown for the individual texts analysed, and the possible influence on this distribution of variables such as chronology, translation practice, and discourse context will be evaluated. Three Old English texts, the Dialogues, Bede, and Boethius, will be examined in more detail in order to illustrate the range of environments in which simple and periphrastic forms are found. It will be seen from the data that the Old English data display a considerable degree of synchronic variation among texts; the possible causes of the distributional patterns observed will be discussed. 4.2 Old Saxon 4.2.1 Distribution of Forms In the two Old Saxon texts analysed, the distribution of simple and periphrastic forms was as follows: 154 Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic Genesis 14.3% (n=1) 85.7% (n=6) 31.3% (n=5) 68.8% (n=11) Heliand 48.7% (n=19) 51.3% (n=20) 44.3% (n=35) 55.7% (n=44) Total: 43.5% (n=20) 56.5% (n=26) 42.1% (n=40) 57.9% (n=55) Table 3: Distribution of Forms in Old Saxon It can be seen from this table that the periphrastic forms are slightly more frequent in Genesis than in the Heliand; however, the differences between the texts are not statistically significant 1 (perfects: χ²(1)=2.863, p>.05 (Fisher’s Exact Test); pluperfects: χ²(1)=0.930, p>.05 (Fisher’s)). Periphrastic forms are also more common for the pluperfect than for the perfect; this difference too is statistically insignificant (Genesis: χ²(1)=0.727, p>.05 (Fisher’s); Heliand: χ²(1)=0.205, p>.05 (Fisher’s)). In keeping with the discussion in Section 3.3.3.1, the data were also analysed without the preterites from verbs forming periphrastic perfects with uuesan. The results of this analysis can be seen below: Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic Genesis 14.3% (n=1) 85.7% (n=6) 31.3% (n=5) 68.8% (n=11) Heliand 44.4% (n=16) 55.6% (n=20) 39.7% (n=29) 60.3% (n=44) Total: 39.5% (n=17) 60.5% (n=26) 38.2% (n=34) 61.8% (n=55) Table 4: Distribution of Forms in Old Saxon (Adjusted) The differences between the texts remain statistically insignificant (perfects: χ²(1)=2.230, p>.05 (Fisher’s); pluperfects: χ²(1)=0.399, p>.05 (Fisher’s)), as do those between the perfect and the pluperfect (Genesis: χ²(1)=0.727, p>.05 (Fisher’s); Heliand: χ²(1)=0.221, p>.05 (Fisher’s)). 1 For discussion of the statistical techniques used in the present study, see e.g. Siegel (1956) and Woods et al. (1986). 155 It should be noted that although the periphrastic forms were available in the language as expressions of perfect and pluperfect meaning, the preterite continued to be used for the same purposes. This can be seen in examples such as the following: (104) Nu uuet ik that ik scal an now know-1SG.PRES I-NOM that I-NOM shall-1SG.PRES on thinum heti libbian, […] nu ik mi thy-MASC.DAT.SG hate-DAT live-INF now I-NOM I-DAT thesa firina gideda. this-FEM.ACC.SG evil-ACC do-1SG.PRET (Genesis II.239.60) ‘Now I know that I must live in Thy enmity, now I have done this evil.’ (105) That uuîti uuas thô the-NEUT.NOM.SG punishment-NOM be-3SG.PRET then agangan, […] the im hêlag god go-PA.PPL REL he-DAT holy-MASC.NOM.SG God-NOM mahtig macode[…]. mighty-MASC.NOM.SG make-3SG.PRET (Heliand III.15.239) ‘Then the punishment was gone which holy God Almighty had inflicted upon him.’ Example (104) is the sole occurrence of a perfect-like preterite in Genesis; the semantic affinity of this preterite with the perfect can nevertheless be seen in its co-occurrence with the temporal adverb nu ‘now’ and the identification of the event denoted by the verb with an explicitly identified, pragmatically salient result state (see Section 3.3.2.1 above). Example (105) shows the use of the preterite to refer to a past event anterior to that denoted by another preterite; it should be noted that in this instance the periphrasis with uuas and the past participle is most plausibly interpreted in a stative sense rather than as a pluperfect. 156 4.2.2 Discourse Context As discussed in Section 3.4.1, the decision was made to identify the discourse contexts in which verb forms occurred. The association of periphrastic perfects with direct speech in Old Saxon has previously been observed (Watts 2001, 131); an analysis of the discourse contexts in which these forms occur has the potential to provide information about the role of stylistic factors in the distribution of these forms. The surviving portion of Genesis is too short to provide meaningful information about associations of this sort, whose identification requires a large enough number of verb forms to provide a representative range of contexts; accordingly, the following analysis is based on the combined data for both texts. Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic Direct speech 31.6% (n=12) 68.4% (n=26) 33.3% (n=1) 66.7% (n=2) Personal — — — — Exposition 100.0% (n=7) 0.0% (n=0) 0.0% (n=0) 100.0% (n=1) Narrative 100.0% (n=1) 0.0% (n=0) 39.7% (n=27) 60.3% (n=41) Indirect speech — — 52.2% (n=12) 47.8% (n=11) Total: 43.5% (n=20) 56.5% (n=26) 42.1% (n=40) 57.9% (n=55) Table 5: Discourse Context in Old Saxon There is a statistically significant association of periphrastic forms with direct speech for the perfect (χ²(2)=12.589, p<.001); no significant association between grammatical form and discourse type exists for the pluperfect (χ²(3)=1.939, p>.05). This remains the case even if the same preterites are excluded as in Section 4.2.1: 157 Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic Direct speech 27.8% (n=10) 72.2% (n=26) 33.3% (n=1) 66.7% (n=2) Personal — — — — Exposition 100.0% (n=7) 0.0% (n=0) 0.0% (n=0) 100.0% (n=1) Narrative — — 35.9% (n=23) 64.1% (n=41) Indirect speech — — 47.6% (n=10) 52.4% (n=11) Total: 39.5% (n=17) 60.5% (n=26) 38.2% (n=34) 61.8% (n=55) Table 6: Discourse Context in Old Saxon (Adjusted) The association is still statistically significant for the perfect (χ²(1)=12.788, p<.01) and not for the pluperfect (χ²(3)=1.576, p>.05). However, it will be seen below that the interpretation of this pattern is more complex than the statistical analysis might suggest. The contrast between the use of periphrastic forms in direct speech and preterites in other contexts can be seen in examples such as the following: (106) Thea liudi sind the-FEM.NOM.PL people-NOM.PL be-3PL.PRES farlorane, farlâten habbiad uualdandes lose-PA.PPL-FEM.NOM.PL leave-PA.PPL have-3PL.PRES ruler-GEN uuord[…]. word-ACC (Heliand XXXVI.3003.110) ‘The people are lost; they have forsaken the word of the Lord.’ (107) Ni uuarð sið noh êr giâmarlîcara NEG become-3SG.PRET since yet ere wretched-MASC.GEN.PL forgang iungaro manno[…]. decease-NOM young-MASC.GEN.PL man-GEN.PL (Heliand IX.734.32) ‘Never since has there yet occurred such a slaughter of wretched young people.’ The periphrastic form in (106) is taken from direct speech in quotation, while the preterite in (107) occurs in an expository passage from the viewpoint of the author’s own present time. The form uuarð in the latter example is one of the verbs excluded from the adjusted totals in Table 6; it nevertheless provides an 158 example of a preterite semantically comparable to the perfect. Such preterites also occurred in direct speech, as well as the periphrastic forms; this can be seen from examples such as (104) above. The question remains of how these results are to be interpreted. It was suggested in Section 3.4.1 above that differences in the distribution patterns for the perfect and the pluperfect typically reflect the operation of content-based factors such as differing proportions of primary and secondary tenses; however, the significant association of periphrastic forms with the particular discourse context of direct speech, as opposed to a random distribution of these forms among contexts in which this temporal content is desired, seems incompatible with such an explanation. Although it might appear improbable that the present perfect and pluperfect periphrases would differ synchronically in their perceived stylistic or register values, such a situation would be compatible with the data and cannot be dismissed out of hand. However, all the examples from the ‘exposition’ category represent occurrences of a single narrative device, gifragn ik ‘I [the author] have heard’. The possibility exists that this construction may not have been subject to variation; for example, it may have been a fixed phrase that was customarily used in this specific form. In such a case, these constructions would not be representative of the author’s choice of a formal representation for perfect semantics, and there would be insufficient data regarding the expression of perfect semantics outside direct-speech contexts for conclusions to be drawn regarding the preferred form in such environments. It is therefore uncertain whether the aforementioned association of periphrastic forms with direct speech simply reflects the absence of any occasion to express perfect-like semantic content in 159 other environments within these two texts. The issues involved in interpreting the effect of discourse variables will be discussed in more detail below in connection with the Old English data. 4.2.3 Summary These results for Old Saxon seem compatible with the standard assumptions regarding the perfect in early Germanic languages, as discussed in Section 1.3 above. There is no significant synchronic variation, and the periphrastic constructions, having been grammaticalized as expressions of perfect meaning, are freely used. It is important to note, however, that the degree of synchronic variation is difficult to estimate precisely given the brevity and fragmentary nature of Genesis; it is possible that with a sample encompassing a greater variety of contemporary Old Saxon usage, unsuspected variation would emerge in the use of these forms. Although the distribution of these forms may have been influenced by stylistic considerations, any such factors that may have existed failed to retard the use of the periphrastic forms to an appreciable extent. 4.3 Old English 4.3.1 Distribution of Forms The apportionment of the perfect and pluperfect semantic domains between simple and periphrastic forms in the Old English texts analysed was as follows: 160 Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic ChronA1 I 100.0% (n=1) 0.0% (n=0) 83.9% (n=26) 16.1% (n=5) Bede 89.5% (n=17) 10.5% (n=2) 91.9% (n=57) 8.1% (n=5) Boece 30.2% (n=16) 69.8% (n=37) 6.3% (n=1) 93.8% (n=15) CP 38.2% (n=13) 61.8% (n=21) 57.7% (n=15) 42.3% (n=11) GD MS C I 100.0% (n=24) 0.0% (n=0) 89.2% (n=58) 10.8% (n=7) ChronA1 II 100.0% (n=2) 0.0% (n=0) 39.5% (n=15) 60.5% (n=23) Oros 71.4% (n=10) 28.6% (n=4) 59.3% (n=51) 40.7% (n=35) ChronA2 — — 0.0% (n=0) 100.0% (n=2) GD MS H I 95.0% (n=19) 5.0% (n=1) 87.7% (n=50) 12.3% (n=7) GenB 21.7% (n=5) 78.3% (n=18) 48.0% (n=12) 52.0% (n=13) WS 100.0% (n=50) 0.0% (n=0) 95.1% (n=39) 4.9% (n=2) ÆCHom 94.1% (n=16) 5.9% (n=1) 90.0% (n=9) 10.0% (n=1) ÆLS 100.0% (n=11) 0.0% (n=0) 91.4% (n=32) 8.6% (n=3) OT 100.0% (n=48) 0.0% (n=0) 93.2% (n=41) 6.8% (n=3) Wulf 75.0% (n=30) 25.0% (n=10) 100.0% (n=7) 0.0% (n=0) ChronE 75.0% (n=3) 25.0% (n=1) 59.3% (n=32) 40.7% (n=22) Total: 73.6% (n=265) 26.4% (n=95) 74.3% (n=445) 25.7% (n=154) Table 7: Distribution of Forms in Old English In contrast to the uniform nature of the Old Saxon data presented above, the Old English data reveal a striking amount of variation. The variation among texts is statistically significant (perfects: χ²(14)=164.892, p<.001;2 pluperfects: χ²(15)=149.187, p<.001). As in Old Saxon, there is generally no significant difference between the perfect and the pluperfect in the use of periphrastic forms; the sole exception is Boethius (χ²(1)=3.793, p<.05). There are a number of potential factors that merit exploration to determine their contribution to the range of variation observed; these include the possibility of diachronic variation, the influence of original texts upon translations, and pragmatic and stylistic variation influenced by variables such as discourse context. 2 For the Old English data, significance values are normally calculated using the Monte Carlo method with a confidence interval of 99% and a sample size of 100,000; however, exact methods continue to be used for χ² tests with one degree of freedom. 161 4.3.2 Diachronic Variation As discussed in Section 1.3.3, authors such as Denison (1993) and Carey (1994) have suggested that the grammatical status of the periphrastic constructions underwent diachronic change over the course of the Old English period. Although considerable variation existed among Old English texts in their use of the periphrastic constructions, analysis of the data from the present study nevertheless suggests that chronology is not a relevant factor in the observed variation. A relatively simple chronological classification, grouping texts into Early Old English and Late Old English, was used as the basis for Carey’s work (1994); if such a division is made of the texts in the present study, grouping the chronological divisions described in Section 3.2.4 in pairs, the results are as follows: Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic Early 65.4% (n=157) 34.6% (n=83) 72.2% (n=324) 27.8% (n=125) Late 90.0% (n=108) 10.0% (n=12) 80.7% (n=121) 19.3% (n=29) Table 8: Comparison of Early and Late Old English Texts From this table it can be seen that there is in fact a decrease in the use of the periphrastic forms in later texts, a decrease which is moreover statistically significant (χ²(1)=24.889, p<.001 for perfects; χ²(1)=4.260, p<.05 for pluperfects). However, a finer-grained chronological analysis shows that these data are not accurate evidence of a general diachronic trend. 162 Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic Late 9th c.– early 10th c. (ChronA1, Bede, CP, Boece, Oros, GD MS C I) 56.5% (n=83) 43.5% (n=64) 68.8% (n=223) 31.2% (n=101) Mid 10th c. (ChronA2, GD MS H I, GenB, WS) 79.6% (n=74) 20.4% (n=19) 80.8% (n=101) 19.2% (n=24) Late 10th c.– early 11th c. (ÆCHom, ÆLS, OT, Wulf) 90.5% (n=105) 9.5% (n=11) 92.7% (n=89) 7.3% (n=7) 11th c.– 12th c. (ChronE) 75.0% (n=3) 25.0% (n=1) 59.3% (n=32) 40.7% (n=22) Table 9: Chronological Comparison of Old English Texts With the texts broken down into the above chronological groupings, it can be seen that the use of periphrastic forms shows neither a regular increase nor a regular decrease over time. It should be noted that the picture for perfects is less representative than that for pluperfects, due to the extreme scarcity of the perfect as a semantic category within the latest text, ChronE. Statistical analysis confirms the absence of any significant correlation between the period of the texts and the proportion of periphrastic forms used in them (Spearman’s ρ=-.400 (perfects), .200 (pluperfects), p>.05). Moreover, these chronological groupings conceal considerable synchronic variation among the individual texts, as shown above by Table 7. This is the case especially within the first two periods, which contain texts differing to the extent of Bede and Boethius, and the West-Saxon Gospels and Genesis B, respectively; statistically significant variation exists both among texts of the first period (perfects: χ²(6)=49.996, p<.001; pluperfects: χ²(6)=80.915, 163 p<.001) and among texts of the second period (perfects: χ²(2)=63.085, p<.001; pluperfects: χ²(3)=32.934, p<.001). The third period shows a greater degree of consistency, attributable in part to the fact that only two authors are represented; moreover, these are the ‘Winchester’ authors Ælfric and Wulfstan, who worked within a coherent scholastic and textual milieu (e.g. Gretsch 1999; Bethurum 1957, 30–96). There is no statistically significant variation among the works of Ælfric analysed here (perfects: χ²(2)=3.517, p>.05; pluperfects: χ²(2)=0.154, p>.05); although the variation between Ælfric and Wulfstan is not significant in the case of pluperfects (χ²(1)=0.594, p>.05), it is significant for perfects (χ²(1)=17.126, p<.001). It can be seen from this analysis that although synchronic variation in the expression of perfect and pluperfect semantics exists at all periods, this variation displays no readily discernible diachronic trend over the course of the Old English period. Another potential hypothesis regarding the data presented above is that they reflect a mixture of a more primitive form of the language with a more advanced form in which the periphrastic constructions have attained a more modern state. Approaches of this kind have been taken in studies of certain other issues in Old English syntax, such as word order (e.g. van Kemenade 1997; Pintzuk and Taylor 2006). However, such a hypothesis would also seem to be incompatible with the evidence. Texts that generally avoid the periphrastic constructions, even those from an early period, include examples of unambiguously modern periphrastic perfects and pluperfects, such as the following: 164 (108) Þu nu hafast þurh Godes gife thou-NOM now have-2SG.PRES through God-GEN gift-DAT þinra feonda hond beswicade[…]. thy-GEN.PL fiend-GEN.PL hand-ACC.PL elude-PA.PPL-ACC.PL (Bede II.9.132.22) ‘You have now through God’s gift eluded the hands of your enemies.’ (109) He for þæs mynstres þearfe, swa he-NOM for the-NEUT.GEN.SG minster-GEN need-DAT so swa he ær gecweden hæfde, so he-NOM ere bespeak-PA.PPL have-3SG.PRET wæs utfarende[…]. be-3SG.PRET out.fare-PRES.PPL (GD MS C I.IV.22.11) ‘He was going out for the needs of the monastery, just as he had arranged previously.’ The interpretation of such examples as genuine perfects and pluperfects in the modern sense, rather than as the stative constructions from which these forms were derived, is supported by evidence of different kinds. In the case of (108), it is pragmatically improbable that the sentence was intended to refer to having the hands of one’s enemies in an eluded state, rather than to the act of eluding them;3 additional evidence for the perfect interpretation is provided by the fact that the verb phrase in question translates a Latin perfect tense, which could not have been interpreted as stative in sense. In (109) there is also syntactic evidence against a stative interpretation; the sentence lacks a direct object for the participle to modify, the verb being instead construed with an adverb of manner. (110) Þa hig hæfdon gehrowen swylce twentig when they-NOM have-3PL.PRET row-PA.PPL such twenty furlanga […] þa gesawon hig furlong-ACC.PL then see-3PL.PRET they-NOM 3 See the discussion in Section 3.3.3.2. 165 þone Hælend[…]. the-MASC.ACC.SG healer-ACC (WS John 6. 19, p. 56) ‘When they had rowed twenty furlongs thus, then they saw the Saviour.’ (111) Ure Drihten […] wile þonne witan […] our-MASC.NOM.SG Lord-NOM will-3SG.PRES then know-INF hu we urne cristendom how we-NOM our-MASC.ACC.SG Christendom-ACC gehealden habban. hold-PA.PPL have-3PL.PRES (Wulf II.121.69) ‘Our Lord will then know how we have kept our Christianity.’ The above examples illustrate further the use of semantically modern periphrastic constructions by authors who normally prefer the preterite. In (110), which contains one of the very few periphrastic pluperfects from the West-Saxon Gospels, the grammaticalized state of this form is shown by its construction with an intransitive verb; such intransitive constructions are already found in earlier texts, as shown by examples such as (78) above. In (111), repeated from (41) above, the temporal nature of the periphrasis is clearly shown by its use with other temporal expressions such as wile and þonne to refer to a span of time extending into the future. From the evidence discussed above it would appear that the low frequency of periphrastic constructions in some Old English texts is due neither to the greater age of these texts nor to a mode of speech in which the periphrastic constructions are restricted to their primitive stative sense. In this connection it might also be noted that at the period in question the use of the preterite to express meaning belonging to the semantic domains of the perfect and pluperfect was not in itself an archaism; preterites of this sort are found throughout the body of Old 166 English texts analysed here, and also in Old Saxon, in which the periphrastic forms are used with a more uniform freedom. The variation among Old English texts in their use of these forms, therefore, is unlikely to be the effect of a diachronically heterogeneous sample, being due rather to some form of synchronic variation which undergoes little diachronic change within the Old English period. 4.3.3 Translation Practices 4.3.3.1 Introduction As discussed in Section 3.4.2, one possibility to be investigated is that the choice of grammatical form in Old English translations from Latin was influenced by the form in the original Latin text. The different categories of Latin original, whether translated by preterites or periphrastic forms, are represented in the Old English texts as follows: 1 6 7 Perfect Perf. Ppl. Imperfect Present Other Interpolation Recast Interchange Bede 70.6% (n=12) 5.9% (n=1) 11.8% (n=2) 11.8% (n=2) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) Boece 17.3% (n=9) 1.9% (n=1) 0.0% (n=0) 5.8% (n=3) 1.9% (n=1) 51.9% (n=27) 5.8% (n=3) 15.4% (n=8) CP 20.6% (n=7) 5.9% (n=2) 0.0% (n=0) 2.9% (n=1) 8.8% (n=3) 23.5% (n=8) 32.4% (n=11) 5.9% (n=2) GDC I 65.2% (n=15) 0.0% (n=0) 0.0% (n=0) 26.1% (n=6) 0.0% (n=0) 4.3% (n=1) 4.3% (n=1) 0.0% (n=0) GDC 60.6% (n=20) 3.0% (n=1) 0.0% (n=0) 6.1% (n=2) 0.0% (n=0) 12.1% (n=4) 12.1% (n=4) 6.1% (n=2) Oros 14.3% (n=2) 7.1% (n=1) 7.1% (n=1) 7.1% (n=1) 0.0% (n=0) 35.7% (n=1) 7.1% (n=1) 21.4% (n=3) GDH I 60.0% (n=12) 0.0% (n=0) 0.0% (n=0) 30.0% (n=6) 0.0% (n=0) 5.0% (n=1) 5.0% (n=1) 0.0% (n=0) GDH — — —. — — — — — WS 98.0% (n=49) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) 2.0% (n=1) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) OT 85.4% (n=41) 0.0% (n=0) 0.0% (n=0) 2.1% (n=1) 4.2% (n=2) 0.0% (n=0) 2.1% (n=1) 6.3% (n=3) Wulf 10.0% (n=1) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) 90.0% (n=9) 0.0% (n=0) 0.0% (n=0) Total 55.8% (n=168) 2.0% (n=6) 1.0% (n=3) 7.3% (n=22) 2.3% (n=7) 18.3% (n=55) 7.3% (n=22) 6.0% (n=18) Table 10: Latin Forms Translated in Old English Texts (Perfects) Pluperfect Perfect Perf. Ppl. Imperfect Other Interpolation Recast Interchange Bede 46.7% (n=28) 15.0% (n=9) 8.3% (n=5) 11.7% (n=7) 5.0% (n=3) 8.3% (n=5) 0.0% (n=0) 5.0% (n=3) Boece 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) 87.5% (n=14) 0.0% (n=0) 12.5% (n=2) CP 26.9% (n=7) 7.7% (n=2) 0.0% (n=0) 3.8% (n=1) 0.0% (n=0) 34.6% (n=9) 23.1% (n=6) 3.8% (n=1) GDC I 50.8% (n=32) 11.1% (n=7) 7.9% (n=5) 12.7% (n=8) 0.0% (n=0) 4.8% (n=3) 3.2% (n=2) 9.5% (n=6) GDC 53.7% (n=51) 4.2% (n=4) 6.3% (n=6) 8.4% (n=8) 4.2% (n=4) 7.4% (n=7) 8.4% (n=8) 7.4% (n=7) Oros 4.7% (n=4) 0.0% (n=0) 12.8% (n=11) 2.3% (n=2) 0.0% (n=0) 61.6% (n=53) 11.6% (n=10) 7.0% (n=6) GDH I 54.4% (n=31) 8.8% (n=5) 7.0% (n=4) 10.5% (n=6) 1.8% (n=1) 5.3% (n=3) 3.5% (n=2) 8.8% (n=5) GDH 50.0% (n=2) 0.0% (n=0) 25.0% (n=1) 0.0% (n=0) 25.0% (n=1) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) WS 51.2% (n=21) 36.6% (n=15) 0.0% (n=0) 9.8% (n=4) 2.4% (n=1) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) OT 45.5% (n=20) 6.8% (n=3) 0.0% (n=0) 0.0% (n=0) 0.0% (n=0) 27.3% (n=12) 13.6% (n=6) 6.8% (n=3) Wulf — — — — — — — — Total 39.8% (n=196) 9.1% (n=45) 6.5% (n=32) 7.3% (n=36) 2.0% (n=10) 21.5% (n=106) 6.9% (n=34) 6.7% (n=33) Table 11: Latin Forms Translated in Old English Texts (Pluperfects) 168 As discussed in Section 3.2.3, the Dialogues were analysed in their entirety only for the first book; for the remaining books, those sections that included at least one periphrastic form were also analysed. This selection procedure would have affected the ratio between simple and periphrastic forms within those books; accordingly, these data were not used for purposes to which this ratio would have been relevant, and were therefore excluded from the Old English data presented in Sections 4.3.1 and 4.3.2. However, the bias introduced by this selection procedure would not have affected the relationship between the grammatical categories in question and other variables, such as the original Latin form to which they correspond; these data have therefore been included in the analysis in this section. In the above tables, the labels GDC I and GDH I refer to the data from Book I of MSS C and H respectively, while the labels GDC and GDH refer only to the data from the subsequent books. In the case of MS H, this data represents the short fragment of Book II which is all that survives of the later portions of this text (see further Yerkes 1986). The homilies of Wulfstan present a special case with regard to translation practices; in a number of instances the Old English homilies are accompanied by a Latin text composed by Wulfstan, often comprised largely of quotations from the Bible and from theological writings, which he then developed into a homily in the vernacular (see Bethurum 1957, 24–49). However, not all his homilies are accompanied by such Latin matter; for those which are not, including those in which all the pluperfect-type forms enumerated above occur, no translation data were recorded. It may be noted here that of all relevant occurrences of preterites 169 or periphrastic constructions in these homilies, only one corresponds directly to a verb in the original Latin: (112) Ego sum Dominus Deus tuus, I-NOM be-1SG.PRES Lord-NOM God-NOM thy-NOM.MASC.SG qui eduxit te de terra REL-NOM.MASC.SG out.lead-3SG.PF thou-ACC of land-ABL Ægipti. Egypt-GEN (Wulf Xb.194.11) ‘I am the Lord thy God, who has led thee out of the land of Egypt.’ Ic eom ðin Drihten, […] I-NOM be-1SG.PRES thy-NOM.MASC.SG Lord-NOM þe gelædde þe ut of Egyptum REL lead-3SG.PRET thou-ACC out of Egypt-DAT (Wulf Xc.201.23) ‘I am thy Lord, who has led thee out of Egypt’ Accordingly, the potential of this text to provide information regarding the influence of Latin on the choice of Old English form is limited; it has nevertheless been included in the figures given here, for informational purposes. 4.3.3.2 Latin Periphrastic Constructions One way in which it is theoretically possible for the form of the Latin text to have exerted an influence upon Old English is through the use in Latin of periphrastic constructions formally comparable to those found in Old English. As the discussion in Section 1.2.3.2 indicates, periphrastic perfects using an auxiliary with the original meaning ‘have’ are also found in the Romance languages, a construction that had its origins in Late Latin. Such Late Latin and vernacular usages had an effect upon the Latin used as a literary language in late antiquity and the Middle Ages; in Medieval Latin texts there are a number of periphrastic constructions with habere that seem from the context to be expressions of tense 170 and aspect alone without any semantic contribution from the notion of ‘having’ expressed by habere as a lexical verb. Some such examples are ambiguous, and there are many collocations of habere with a participle from Late Latin and even Classical Latin texts whose semantic similarity to the perfects that eventually emerged has been the subject of controversy (see Thielmann 1885; Jacob 1995). By the early Middle Ages, however, incontrovertible examples of the use of such constructions to express temporal and aspectual meaning come to be found; one of the earliest sources in which such constructions are found is the sixth-century writing of Gregory of Tours, which includes examples such as the following (from Jacob 1995, 378): (113) Sicut domnus imperator mandatum habet so.as lord-NOM.SG emperor-NOM.SG order-PA.PPL have-3SG.PRES ‘just as the Lord Emperor has commanded’ In this example, the absence of a noun or pronoun that the past participle mandatum could be understood to modify seems to provide clear indication of the grammaticalized status of the construction. In contexts such as this it is difficult to identify the agreement of the past participle, which morphologically can be interpreted as masculine accusative, neuter nominative, or neuter accusative; however, similar periphrases in the same text show participial agreement with the object (see Jacob 1995, 377–8), and so the participle might best be interpreted as an accusative, receiving a default neuter gender. Periphrastic constructions continued to occur in Latin texts throughout the Middle Ages, undoubtedly reinforced by the presence of similar constructions in Romance and even Germanic vernaculars (see Stotz 1998, 329–31); they occurred 171 not only as present perfects, but future perfects and pluperfects, as in the following example from the Annals of Einhard (from Stotz 1998, 331): (114) In Saxoniam, sicut dispositum habebat, in Saxony-ACC so.as arrange-PA.PPL have-3SG.IMPF duxit exercitum lead-3SG.PERF army-ACC.SG ‘Into Saxony, just as he had arranged, he led the army.’ Given the pre-eminence of Latin as a medium for written texts in the Middle Ages, the possibility that these Medieval Latin constructions themselves exerted an influence on the vernaculars and their nascent literary languages cannot be summarily dismissed. As discussed in Section 1.3.1, it has even been suggested by some that Latin periphrastic constructions of this sort are the origin of the Germanic perfect forms; while this may not be the case, it is not improbable that the use of such Latin constructions could have encouraged the use of similar forms already existing in Old English. Although the influence of such constructions on Old English merits consideration as a possibility, upon analysis the data included in the present study contain only two cases for which such an interpretation is even conceivable: (115) Regem crudelissimum Totilam infensum king-ACC cruelest-MASC.ACC.SG Totila-ACC enraged-MASC.ACC.SG omnimodo habebat every.way have-3SG.IMPF (GD III.XII.2, p. 240) ‘He had completely enraged the most cruel king Totila/had the most cruel king Totila completely enraged.’ Se hæfde swiðe abelged that-MASC.NOM.SG have-3SG.PRET severely enrage-PA.PPL þone wælhreowan cyning Totila the-MASC.ACC.SG bloodthirsty-MASC.ACC.SG king.ACC Totila.ACC (GD MS C III.XII.196.16) ‘He had completely enraged the bloodthirsty king Totila/had the bloodthirsty king Totila completely enraged.’ 172 (116) Eorum nomina […] apud se fixa they-GEN name-NOM.PL near self-ACC fix-PA.PPL-NEUT.NOM.PL aeterna claritas habebat eternal-FEM.NOM.SG clarity-NOM have-3SG.IMPF (GD IV.XXVII.8, p. 378) ‘The eternal brightness had their names fixed/had fixed their names with itself.’ Seo ece beorhtnes […] the-FEM.NOM.SG eternal-FEM.NOM.SG brightness-NOM hæfde heora naman gefæstnode have-3SG.PRET they-GEN name-NOM.PL fasten.PA-PPL-NOM.PL mid hy sylfe with it-INST self-INST.SG (GD MS C IV.XVII.299.21) ‘The eternal brightness had their names fixed/had fixed their names with itself.’ It should be noted that for both these examples the interpretation with a lexical verb meaning ‘have’ is possible both in Latin and in Old English; moreover, in the case of (115), infensus is in origin participial, but other forms of the verb to which it corresponds are virtually unknown in the classical language (see Lewis and Short 1879), although occurrences in medieval texts are more frequent (Latham and Howlett 1997). However, even if the Latin constructions were not true perfects, the same is not necessarily true of their Old English counterparts, both of which would normally qualify for inclusion under the criteria described in Section 3.3.3. To avoid a possibly unjustified commitment to a particular interpretation of the relation between the Latin and Old English forms, both examples have been placed in the category ‘other’. In general, though, it seems reasonable to conclude that in the absence of any other potential examples in the data of Latin periphrastic perfects or pluperfects, the opportunity of these forms for influence on Old English usage may have been limited. 173 4.3.3.3 Translation Practices and Old English Verb Forms The relationship between the different types of Latin original and the choice in Old English between preterites and periphrastic constructions as expressions of given semantic content can be seen for all texts together in the following table: Perfect Pluperfect Simple Periphrastic Simple Periphrastic Pluperfect — — 92.9% (n=182) 7.1% (n=14) Perfect 89.9% (n=151) 10.1% (n=17) 91.1% (n=41) 8.9% (n=4) Perfect Participle 33.3% (n=2) 66.7% (n=4) 50.0% (n=16) 50.0% (n=16) Imperfect 100.0% (n=3) 0.0% (n=0) 91.7% (n=33) 8.3% (n=3) Present 81.8% (n=18) 18.2% (n=4) — — Other 42.9% (n=3) 57.1% (n=4) 70.0% (n=7) 30.0% (n=3) Interpolation 56.4% (n=31) 43.6% (n=24) 58.5% (n=62) 41.5% (n=44) Recast 50.0% (n=11) 50.0% (n=11) 70.6% (n=24) 29.4% (n=10) Interchange 27.8% (n=5) 72.2% (n=13) 48.5% (n=16) 51.5% (n=17) Total 74.4% (n=224) 25.6% (n=77) 77.4% (n=381) 22.6% (n=111) Table 12: Translation and Old English Verb Forms (All Texts) It can be seen from the above table that periphrastic forms are generally less common in the more literal translation categories, such as ‘perfect’ and ‘pluperfect’, than in less literal categories, such as ‘recast’ and ‘interchange’. When all texts are analysed together, the variation found is statistically significant (perfects: χ²(7)=68.622, p<.001; pluperfects: χ²(7)=88.294, p<.001). When the different texts are analysed individually, however, a different picture emerges. The only text that displays a statistically significant association between the form of the Latin original and that of the Old English translation is the Dialogues. 174 Perfect Pluperfect Simple Periphrastic Simple Periphrastic Pluperfect — — 100.0% (n=32) 0.0% (n=0) Perfect 100.0% (n=15) 0.0% (n=0) 100.0% (n=7) 0.0% (n=0) Perfect Participle — — 40.0% (n=2) 60.0% (n=3) Imperfect — — 100.0% (n=8) 0.0% (n=0) Present 100.0% (n=6) 0.0% (n=0) — — Interpolation 100.0% (n=1) 0.0% (n=0) 66.7% (n=2) 33.3% (n=1) Recast 100.0% (n=1) 0.0% (n=0) 100.0% (n=2) 0.0% (n=0) Interchange — — 50.0% (n=3) 50.0% (n=3) Table 13: Translation and Old English Verb Forms (GD MS C, I) Perfect Pluperfect Simple Periphrastic Simple Periphrastic Pluperfect — — 90.2% (n=46) 9.8% (n=5) Perfect 80.0% (n=16) 20.0% (n=4) 75.0% (n=3) 25.0% (n=1) Perfect Participle 0.0% (n=0) 100.0% (n=1) 16.7% (n=1) 83.3% (n=5) Imperfect — — 100.0% (n=8) 0.0% (n=0) Present 100.0% (n=2) 0.0% (n=0) — — Other — — 50.0% (n=2) 50.0% (n=2) Interpolation 25.0% (n=1) 75.0% (n=3) 57.1% (n=4) 42.9% (n=3) Recast 75.0% (n=3) 25.0% (n=1) 87.5% (n=7) 12.5% (n=1) Interchange 0.0% (n=0) 100.0% (n=2) 0.0% (n=0) 100.0% (n=7) Table 14: Translation and Old English Verb Forms (GD MS C, Other Books) Perfect Pluperfect Simple Periphrastic Simple Periphrastic Pluperfect — — 96.8% (n=30) 3.2% (n=1) Perfect 91.7% (n=11) 8.3% (n=1) 100.0% (n=5) 0.0% (n=0) Perfect Participle — — 25.0% (n=1) 75.0% (n=3) Imperfect — — 100.0% (n=6) 0.0% (n=0) Present 100.0% (n=6) 0.0% (n=0) — — Other — — 100.0% (n=1) 0.0% (n=0) Interpolation 100.0% (n=1) 0.0% (n=0) 66.7% (n=2) 33.3% (n=1) Recast 100.0% (n=1) 0.0% (n=0) 100.0% (n=2) 0.0% (n=0) Interchange — — 60.0% (n=3) 40.0% (n=2) Table 15: Translation and Old English Verb Forms (GD MS H, I) Perfect Pluperfect Simple Periphrastic Simple Periphrastic Pluperfect — — 50.0% (n=1) 50.0% (n=1) Perfect Participle — — 0.0% (n=0) 100.0% (n=1) Table 16: Translation and Old English Verb Forms (GD MS H, Other Books) 175 As the above tables show, Book I contains no periphrastic perfects in MS C and only one in MS H, a phenomenon which will be discussed in more detail in Section 4.3.5 below; however, in the remaining books of MS C a significant association exists between the use of periphrastic forms and the relationship of the Old English and Latin texts (χ²(5)=11.850, p<.05). For the pluperfect, which is represented by a greater variety of forms, significant associations exist for Book I of MS C (χ²(6)=28.912, p<.01), Book I of MS H (χ²(7)=23.726, p<.01), and the remaining books of MS C (χ²(7)=43.718, p<.001); the fragment of Book II from MS H is too short to allow for meaningful statistical analysis. In this case the association of periphrastic forms with less literal translation contexts may be genuine. One possible explanation is that Old English perfect-like and pluperfect- like preterites are associated with the Latin perfect and pluperfect tenses by means of a simple tendency towards iconicity, the representation of one synthetically inflected verb form by another verb form of the same kind; passages where the Latin model was not followed so strictly would therefore be more representative of the translator’s usual practice. Although statistical evidence for the influence of translation upon the form of an Old English text is present only for the Dialogues, similar factors may also be responsible for the observed variation in the expression of perfect and pluperfect semantics in some other cases. One such case is that of Genesis B, the Old English translation of the Old Saxon poem Genesis; the literalism of the translation and the influence of Old Saxon on the language of the Old English text have long been remarked (see Doane 1991, 47–54). However, an exhaustive comparison of the translation with its original, such as was performed here for the 176 Latin texts, is in this case impossible; both poems survive only in fragmentary form, with only 27 lines found in both texts. This overlapping portion contains only two forms relevant to the present study, about which all that can be said is that the choice of preterite or periphrastic tense is the same for both languages in each case: (117) Uuela that thu nu eua habas[…] ubilo alas that thou-NOM now Eve-NOM have-2SG.PRES evilly gimarakot unkaro selbaro siđ. mark-PA.PPL us-GEN.DUAL self-GEN.PL journey-ACC (Genesis I.232.1) ‘Alas that you, Eve, have now evilly marked our own path.’ Hwæt, þu eue, hæfst yfele what thou-NOM Eve-NOM have-2SG.PRES evilly gemearcod uncer sylfra sið. mark-PA.PPL us-GEN.DUAL self-GEN.PL journey-ACC (GenB XIII.229.791) ‘Oh you, Eve, have evilly marked our own path.’ (118) Thit uuas alloro lando this-NEUT.NOM.SG be-3SG.PRET all-NEUT.GEN.PL land-GEN.PL sconiust that uuit […] hebbian muostun fairest-NEUT.NOM.SG REL we-NOM.DUAL have-INF must-1PL.PRET thar thu them ni hordis[…]. where thou-NOM that-MASC.DAT.SG NEG hear-2SG.PRET.SUBJ (Genesis I.232.5) ‘This was the fairest of all lands, which we were to have had when you had not listened to him.’ Þis is landa betst this-NEUT.NOM.SG be-3SG.PRES land-GEN.PL best-NEUT.NOM.SG þæt wit […] habban moston þær REL we-NOM.DUAL have-INF must-1PL.PRET where þu þam ne hierde[…]. thou-NOM that-MASC.DAT.SG NEG hear-2SG.PRET.SUBJ (GenB XIII.229.795) ‘This is the best of lands, which we were to have had when you had not listened to him.’ 177 From Table 7 it can be seen that Genesis B is unusual among Old English texts in its extensive use of the periphrastic constructions, with 78.3% of perfects and 52.0% of pluperfects being expressed in this manner. As Tables 3 and 4 show, in Genesis, 85.7% of perfects and 68.8% of pluperfects are periphrastic; while Genesis B is anomalous among Old English texts, there is no significant difference between Genesis and Genesis B in their use of these forms (perfects: χ²(1)=0.186, p>.05; pluperfects: χ²(1)=1.128, p>.05). If translation influences are responsible for the differences between Genesis B and other Old English texts, this could be explained through the operation of the same iconic tendencies suggested for the Dialogues; in this case the desire to render a periphrastic form by another periphrastic form would produce the opposite effect to that seen in the Dialogues and would increase the use of the periphrastic tenses. It should be noted that in both cases the fact that the observed translations were felt to be suitable equivalents of the original texts suggests that different translators were simply making different use of a range of variation already possible within Old English. In other cases, the data are not sufficient to determine whether or not translation practices influenced the choice of form in Old English. For example, the West-Saxon Gospels are among the most literal of translations, with 98.0% of semantically perfect-like forms and 97.6% of pluperfect-like forms rendering Latin inflected tenses; they are also among the lowest in periphrastic forms, with only 2 periphrastic pluperfects, or 4.9% of the total, and no periphrastic perfects. It is not impossible that tendencies similar to those operating in the Dialogues are operating here, and that the absence of periphrastic forms is an effect of this 178 literalism in translation; however, it is also possible that this text represents a pattern similar to that seen in Ælfric’s Old Testament translations, in which periphrastic forms are avoided even in passages of relatively free translation. In the absence of similarly free passages from the West-Saxon Gospels, it is impossible to prove or disprove either of these hypotheses. If there are many individual texts for which the association between the form of the Latin original and that of the Old English translation does not hold, the question remains of why such an association should be found when the data from all Old English texts is combined. A likely explanation is provided by the data presented above in Tables 10 and 11. It can be seen from these tables that the Old English texts analysed differ widely in their fidelity to the Latin original, in the manner discussed in Section 3.4.2. The sample includes texts such as Bede, in which 0.0% of perfects and 13.3% of pluperfects come from less literal categories such as ‘interpolation’, ‘recast’, and ‘interchange’, and texts such as Boethius, in which 73.1% of perfects and 100.0% of pluperfects fall into these categories. As shown in Table 7, these texts also differ in their use of the periphrastic constructions, which are used in Bede for only 10.5% of perfects and 8.1% of pluperfects, but in Boethius for 69.8% of perfects and 93.8% of pluperfects. Although Bede avoids the periphrastic forms regardless of the literalism of the translation and Boethius uses them freely, again regardless of literalism, combining the data from texts such as these may create a spurious association between literalism and the avoidance of periphrastic forms. If little of the observed difference among texts can be ascribed to the direct influence of translation, it may be better to consider this variation as deriving not from a 179 simple iconic tendency, such as that suggested above for the Dialogues, but from more subtle and generalized stylistic principles. In the following sections, possibilities of this sort will be explored in greater depth. 4.3.4 Discourse Context An analysis of the association between the different grammatical forms and specific discourse contexts has been performed for the Old English data, as for the Old Saxon data. The results for all Old English texts combined can be seen in the following table: Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic Direct speech 75.4% (n=212) 24.6% (n=69) 60.5% (n=26) 39.5% (n=17) Personal 80.0% (n=12) 20.0% (n=3) 83.3% (n=5) 16.7% (n=1) Exposition 61.6% (n=61) 38.4% (n=38) 62.1% (n=18) 37.9% (n=11) Narrative 100.0% (n=4) 0.0% (n=0) 74.5% (n=380) 25.5% (n=130) Indirect speech — — 81.4% (n=92) 18.6% (n=21) Total: 72.4% (n=289) 27.6% (n=110) 74.3% (n=521) 25.7% (n=180) Table 17: Discourse Context in Old English (Combined) The variation among these discourse contexts is statistically significant (perfects: χ²(3)=9.030, p<.05; pluperfects: χ²(4)=9.852, p<.05). However, no clear association of a periphrastic form with direct speech, such as that found above for Old Saxon, emerges here; in fact, the periphrastic perfect is actually most common in expository contexts. It should be noted that the existence of such a clear association is dependent on the interaction of two independent factors, an association of a grammatical form such as the periphrastic perfect with a particular stylistic value and the association of this stylistic value with a particular discourse context; where these two factors vary independently, an obvious association of this sort may not be found. To take a purely hypothetical example, 180 if the periphrastic perfect were seen as markedly colloquial, it would exhibit an association of this sort with direct speech only if the direct speech in a particular text were significantly more colloquial than the surrounding narrative. As in the case of translation factors, in order to elucidate the causes of the statistically significant variation described above, it is necessary to view the texts analysed individually. When the same statistical analysis is performed on individual texts, a significant association is again found within only one text, in this case the Pastoral Care. The occurrences of the relevant forms within different discourse contexts can be seen in the following table: Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic Direct speech 80.0% (n=4) 20.0% (n=1) 33.3% (n=1) 66.7% (n=2) Personal 100.0% (n=2) 0.0% (n=0) 100.0% (n=1) 0.0% (n=0) Exposition 25.9% (n=7) 74.1% (n=20) 40.0% (n=6) 60.0% (n=9) Narrative — — 100.0% (n=7) 0.0% (n=0) Indirect speech — — — — Total: 38.2% (n=13) 61.8% (n=21) 57.7 % (n=15) 42.3% (n=11) Table 18: Discourse Context in Old English (CP) It can be seen that in this text periphrastic forms are preferred only in expository contexts, an association which is statistically significant (perfects: χ²(2)=8.656, p<.01; pluperfects: χ²(3)=8.520, p<.05). The strength of this association may have biased the data for Old English as a whole. 181 Perfect Pluperfect Preterite Periphrastic Preterite Periphrastic Direct speech 75.4% (n=208) 24.6% (n=68) 62.5% (n=25) 37.5% (n=15) Personal 76.9% (n=10) 23.1% (n=3) 80.0% (n=4) 20.0% (n=1) Exposition 75.0% (n=54) 25.0% (n=18) 85.7% (n=12) 14.3% (n=2) Narrative 100.0% (n=4) 0.0% (n=0) 74.2% (n=373) 25.8% (n=130) Indirect speech — — 81.4% (n=92) 18.6% (n=21) Total: 75.6% (n=276) 24.4% (n=89) 75.0% (n=506) 25.0% (n=169) Table 19: Discourse Context in Old English (Excluding CP) When the Pastoral Care is excluded, the variation among discourse contexts is no longer statistically significant (perfects: χ²(3)=1.326, p>.05; pluperfects: χ²(4)=6.922, p>.05). This would seem to support the hypothesis that the significance of the association found when data from all texts were analysed together was due to bias from the Pastoral Care. If periphrastic forms are used significantly more often in the Pastoral Care in expository passages, the question arises of what meaning is to be attached to this association. At first sight, there might seem to be little reason why the periphrastic forms should be seen as more suitable for expository purposes. However, as the Pastoral Care is a work of instruction, exposition in the sense in which the term is used here forms the bulk of its content; the passages of direct speech and narrative enumerated in Table 18 consist entirely of Biblical quotations and paraphrases, such as the following: (119) He cwæð: Ic wille secgan ongean he-NOM speak-3PL.PRET I-NOM will-1SG.PRES say-INF against me selfne min unryht, Dryhten, me-ACC self-MASC.ACC.SG my-NEUT.ACC.SG unright-ACC Lord-NOM forðæm ðu forgeafe ða because thou-NOM forgive-2SG.PRET the-FEM.ACC.SG 182 arleasnesse minre heortan. wickedness-ACC my-FEM.GEN.SG heart-GEN (CP LIII.419.8) ‘He [the Psalmist] said, “I will tell against myself my sin, Lord, because You have forgiven the wickedness of my heart.”’ (120) Saul […] hine bealg wið ðone Saul-NOM he-ACC anger-3SG.PRET with the-MASC.ACC.SG ilcan Samuel ðe hine ær on same-MASC.ACC.SG Samuel-ACC REL he-ACC ere on ðæm rice gebrohte[…]. the-NEUT.DAT.SG kingdom-DAT bring-3SG.PRET (CP III.35.14) ‘Saul was enraged against the same Samuel who had previously brought him to the throne.’ This might suggest that the expository pattern represents the norm for this text, from which the Biblical passages represent a deviation. In such a case, the observed variation would not be directly connected to discourse contexts as such, but it would nevertheless represent a genuine stylistic differentiation; it may be recalled that Table 7 shows a similarly sparse use of periphrastic forms in other Biblical translations such as the West-Saxon Gospels and Ælfric’s Old Testament work. Even if stylistic factors are operating to produce this variation, without knowing the exact value attached to these constructions it is difficult to predict why they should have been avoided, without being entirely excluded, in Biblical contexts. In this way, the analysis of discourse contexts provides data that are suggestive, but not conclusive, with regard to some of the factors motivating the observed variation in the use of simple and periphrastic forms of comparable meaning. 183 4.3.5 Textual Examination 4.3.5.1 Introduction It can be seen from the foregoing discussion that although statistical analysis of the Old English data can identify some of the factors responsible for certain cases of variation, much of the observed variation is left unexplained; in their use of the preterite and the periphrastic constructions to express similar semantic content, the texts analysed differ more greatly than would be predicted on the basis of any of the factors discussed above. Accordingly, it is necessary to examine the data more closely in their original context in order to provide a more detailed picture of each text and to ensure that the observed variation is not the product of any unidentified grammatical factors. Three texts have been chosen as the basis for such a detailed investigation: the Dialogues, Bede, and Boethius. As discussed in Section 3.2.4, these texts are roughly contemporary; despite this, they exhibit considerable variation in their use of the periphrastic constructions. As the data presented above show, in the Dialogues the use of the periphrastic constructions is associated with the degree to which the text departs from the Latin original, while Bede and Boethius, though showing no significant correlation between the use of these constructions and their translation practices, differ significantly in their use of the periphrastic constructions, with Boethius making much more copious use of these forms than Bede. Although there would be considerable redundancy in an exhaustive discussion of every occurrence of the periphrastic tenses and of semantically comparable preterites within the analysed samples of these texts, cases are only omitted from the following discussion when they are considered to be substantially identical to those discussed in all salient 184 syntactic, semantic, and pragmatic respects; references are provided in some instances to similar examples not discussed in greater detail. For each work, the present perfect and pluperfect will be discussed separately. 4.3.5.2 Dialogues 4.3.5.2.1 The Perfect As the data in Section 4.3.3.3 show, although there is a significant correlation in the Dialogues between the choice of verb form used in Old English and the relationship of this form to the original Latin text, within the first book this correlation only applies to the pluperfect. Periphrastic present perfects are in fact extremely uncommon in the first book of the Dialogues; there are no examples in MS C and only one in the revision in MS H. This example, together with the corresponding passages in the Latin original and in MS C, can be seen below: (121) Tradidit te mihi Deus. deliver-3SG.PF thou-ACC I-DAT God-NOM (GD I.III.4, p. 86) ‘God has delivered you to me.’ Forþon þe þu þis dydest, because that thou-NOM this-NEUT.ACC.SG did-2SG.PRET God þe me on geweald sealde. God-NOM thou-ACC I-DAT on power-ACC sell-3SG.PRET (GD MS C I.6.25.6) ‘Because you did this, God has given you into my power.’ Nu hæfð God þe geseald now have-3SG.PRES God.NOM thou.ACC sell-PA.PPL me on geweald. I-DAT on power-ACC (GD MS H I.6.25.6) ‘Now God has given you into my power.’ 185 This example, which is drawn from dialogue quoted as direct speech in the text, illustrates several tendencies found within the data. 189 , the use of the simple preterite to translate the Latin perfect tense in its perfect sense, and not merely in its aorist sense, 4 is characteristic of MS C. The replacement of this preterite by the periphrastic construction seen in MS H might be interpreted as an instance of the tendency of this revised text toward a less literal translation incorporating more modern elements in its syntax (see Yerkes 1982); however, as comparison with similar examples below will show, the question remains of why a periphrastic form should have been used only in this instance. It might also be noted that MS C contains an interpolated causal clause with forþon, which is omitted from MS H. Elsewhere in the Dialogues, both MS C and MS H use the simple preterite to convey semantic content similar to that denoted by the perfect in Example (121). (122) Felix […] qui eiusdem monasterii Felix-NOM REL-MASC.NOM.SG same-NEUT.GEN.SG monastery-GEN nuper praepositus fuit newly before.place-PA.PPL-MASC.NOM.SG be-3SG.PF (GD I.III.1, p. 84) ‘Felix, who has recently been made provost of the same monastery’ Felix […] se nu niwan wearð Felix-NOM REL-MASC.NOM.SG now newly become-3SG.PRET prafost þæs ylcan mynstres provost-NOM the-NEUT.GEN.SG same-NEUT.GEN.SG minster-GEN (GD MS C, MS H I.6.23.10) ‘Felix, who has now recently become provost of the same monastery’ 4 See Section 3.4.2.2 186 This example is similar to (121) in terms of temporal and aspectual semantic content; the two examples also have in common shared contextual features such as the occurrence of present-time adverbs such as nu ‘now’ (see Section 3.3.2.1). Despite the similarity between (121) and (122), the verb phrase in the latter, unlike that in the former, has undergone no revision in MS H to a periphrastic construction. (123) Quid miraris, Petre[…]? what-ACC wonder-2SG.PRES.MP Peter-VOC An menti excidit quod […]? or mind-DAT out.fall-3SG.PF that (GD I.IV.19, p. 96) ‘Why do you wonder, Peter? Or has it slipped your mind that […]’ To hwan wundrast þu, Petrus[…]? Hwæþer to what-INST wonder-2SG.PRES thou-NOM Peter-NOM whether þe of mode abeah, þæt þu thou-ACC of mood-DAT depart-3SG.PRET that thou-NOM ne gemundest, þæt […]. NEG remember-2SG.PRET that (GD MS C I.12.40.20) ‘Why do you wonder, Peter? Has it departed from your mind, that you do not remember that […]?’ Hwæt wundrast þu, Petrus[…]? Hwæðer what-ACC wonder-2SG.PRES thou-NOM Peter-NOM whether þe þe of mode gewat, þæt þu that thou-ACC of mood-DAT leave-3SG.PRES that thou-NOM ne gemundest, þæt […]. NEG remember-2SG.PRET that (GD MS H I.12.40.20) ‘Why do you wonder, Peter? Has it departed from your mind, that you do not remember that […]?’ Example (123) provides a further example of an Old English preterite that translates a Latin perfect; like (122), in the text it is spoken by the author, St. Gregory. Example (123) is similar to Example (121) but not Example (122) in that the passage in question has undergone revision in MS H. Nevertheless, in 187 this case no periphrastic perfect is used; instead, as is most frequent in this text, the preterite alone is used when translating a Latin perfect. Many other examples from the Dialogues are similarly uncomplicated translations of Latin perfects by Old English preterites, such as the following: (124) Postquam facti illius tale after deed-GEN that-MASC.GEN.SG such-NEUT.ACC.SG miraculum dixisti, superest ut me etiam de miracle-NOM say-2SG.PF over.be-3SG.PRES that I-ACC also of humilitate mentis eius aedifices. humilty-GEN mind-GEN he-GEN edify-2SG.PRES.SUBJ (GD I.V.3, p. 100) ‘After you have told me such a miracle from his acts, it is most important that you edify me about his humility of mind.’ Æfter þan þu me sædest after that-NEUT.DAT.SG thou-NOM I-ACC say-2SG.PRET hwilc wundor his dæda, ofer þæt which wonder-ACC he-GEN deed-GEN.PL over that-NEUT.ACC.SG me lysteþ get, þæt þu me hwæthugu I-DAT list-3SG.PRES yet that thou-NOM I-ACC somewhat lære be his modes eadmodnesse. teach-2SG.PRES.SUBJ by he-GEN mood-GEN humility-DAT (GD MS C I.XV.45.11) ‘After you have told me such a wonder from his acts, beyond that yet it will please me for you to teach me something of his spirit’s humility.’ Æfter þam þe þu me sædest after that-NEUT.DAT.SG REL thou-NOM I-ACC say-2SG.PRET swylc wundor his weorces, ofer þæt which wonder-ACC he-GEN work-GEN over that-NEUT.ACC.SG me lysteð, þæt þu me eac secge I-DAT list-3SG.PRES that thou-NOM I-ACC also say-2SG.PRES.SUBJ sum þing be his modes eadmodnysse. some thing-ACC by he-GEN mood-GEN humility-DAT (GD MS H I.XV.45.11) ‘After you have told me such a wonder from his works, beyond that it will please me for you to tell me something of his spirit’s humility.’ 188 (125) Tu solus es qui in thou-NOM alone-MASC.NOM.SG be-2SG.PRES REL-MASC.NOM.SG in me oculos apertos habuisti. I-ABL eye-ACC.PL open-MASC.ACC.PL have-2SG.PF (GD I.V.5, p. 100) ‘You are the only one who has had open eyes upon me.’ Þu ana hæfdest ontynde thou-NOM one-MASC.NOM.SG have-2SG.PRET open-PA.PPL eagan on me & me mid rihte oncneowe eye-ACC.PL on I-DAT and I-ACC with right-DAT know-2SG.PRET (GD MS C I.XVI.46.29) ‘You alone have had open eyes on me and known me rightly.’ (Passage missing in MS H) Example (124) forms part of the dialogue between St. Gregory and his deacon Peter, which frames the text; similar examples can be seen in (169)– (173) in Appendix A. Example (125), like (172)– (175) below, is from direct speech in quotation. It may be noted that in the case of (125), it might also be possible to interpret the Latin text as an example of the Late Latin periphrastic perfect discussed in Section 4.3.3.2 above; giving it the sense ‘You are the only one who has opened his eyes to me.’ However, the Old English text makes it clear that the Latin was not interpreted in this sense by the translator, since the interpolated verb oncneowe ‘knew’ is a preterite coordinated with hæfdest rather than a participle coordinated with ontynde. Many of the preterites from the Dialogues analysed in this study correspond not to Latin perfects, but to Latin present tenses; as discussed in Section 3.3.2.1, the use of the preterite for events known to be ongoing provides support for the notion that the semantic range of this grammatical category included that of the present perfect. 189 (126) Ea […] Fortunati, qui valde they-NEUT.ACC Fortunatus-GEN REL-MASC.NOM.SG strongly mihi aetate, opere, et simplicitate placet, I-DAT age-ABL work-ABL and simplicity-ABL please-3SG.PRES relatione cognovi. account-ABL discover-1SG.PERF (GD I.IV.21, p. 98) ‘These things I have learned from the account of Fortunatus, who pleases me greatly in his age, works, and simplicity.’ Þa word […] ic oncneow, of that-NEUT.ACC.PL word-ACC.PL I-NOM discover-1SG.PRET of sægene […] Furtunates þæs yldu saying-DAT Fortunatus-GEN REL-MASC.GEN.SG age-NOM & weorc & bylwitnes me licaþ swiþe wel. and work-NOM and simplicity I-DAT like-3PL.PRES strongly well (GD MS C I.XIV.42.18) ‘These things I learned from the account of Fortunatus, whose age, works, and simplicity please me very well.’ Þa þing […] ic oncneow, swa that-NEUT.ACC.PL thing-ACC.PL I-NOM discover-1SG.PRET so swa me rehte […] Furtunatus se me so I-DAT tell-3SG.PRET Fortunatus-NOM REL-MASC.NOM.SG I-DAT swiðe wel gelicode on ylde, on weorce strongly well like-3SG.PRET on age-DAT on work-DAT & on bilwitnysse. and on simplicity-DAT (GD MS H I.XIV.42.18) ‘These things I have learned just as Fortunatus told me, who has pleased me very well in his age, his works, and his simplicity.’ (127) Peregrinum hominem de hospitio suo pilgrim-MASC.ACC.SG man-ACC of hospice-ABL own-NEUT.ABL.SG expulit. Quaero ubi requiescere expel-3SG.PF seek-1SG.PRES where rest.INF debeam, et in civitate eius non invenio. owe-1SG.PRES.SUBJ and in city-ABL he-GEN NEG find-1SG.PRES (GD I.X.6, p. 122) ‘He has expelled a foreign man from his lodgings. I seek where I am to rest, and in his city I do not find anything.’ 190 Nu he adraf me ælþeodigne now he-NOM out.drive-3SG.PRET I-ACC foreign-MASC.ACC.SG man of his huse, & ic forþon man-ACC of he-GEN house-DAT and I-NOM therefore sece, hwær ic me gerestan scyle, seek-1SG.PRES where I-NOM I-ACC rest-INF shall-1SG.PRES nu ic in his cæster nane ne fand. now I-NOM in he-GEN city-DAT none-ACC NEG find-1SG.PRET (GD MS C I.XXX.75.9) ‘Now he has driven me, a foreign man, out of his house, and therefore I seek where I am to rest myself, and now in his city I have found none.’ Nu he adraf me ælþeodine now he-NOM out.drive-3SG.PRET I-ACC foreign-MASC.ACC.SG mannan of his huse. Ic sece, hwær man-ACC of he-GEN house-DAT I-NOM seek-1SG.PRES where ic me gerestan scyle, & ic I-NOM I-ACC rest-INF shall-1SG.PRES .SUBJ and I-NOM on his ceastre nane findan ne mæg. on he-GEN city none-ACC find-INF NEG may-1SG.PRES (GD MS H I.XXX.75.9) ‘Now he has driven me, a foreign man, out of his house. I seek where I am to rest myself, and in his city I can find none.’ Example (127), like (176) below, is drawn from direct speech in quotation, while (126) is from the surrounding dialogue, like (55) and (93) above and (177) below. In both examples shown here, the two manuscripts differ in their treatment of the Latin present. In (126), the Latin present tense placet ‘pleases’ is translated literally in MS C by the Old English present licaþ, while in MS H this is changed to a preterite gelicode; it should be noted that even at this later date, the preterite could be used in preference to the periphrastic perfect as a method of expressing this meaning, in a case where neither the Latin nor the earlier Old English text have any apparent influence on the choice of form. It is difficult to determine whether the use of the prefix ge- in the preterite might have been felt to convey 191 any sense of completion, as discussed in Section 3.3.2.5; given the existence of the two nearly synonymous verbs lician and gelician (see e.g. Clark Hall 1960), this may simply be another instance of lexical interchange such as that between settan and asettan in (172). In (127), it is MS C that uses a preterite, translating Latin invenio ‘find’ by fand ‘found’; in MS H, the original present tense is restored, but an auxiliary mæg is added, so that the text reads not ‘find’ but ‘can find’. These passages exemplify the difficulty of making generalizations about the two manuscripts in regard to their respective fidelity to the original and their syntactic practices. As the statistics presented in Section 4.3.3.3 show, this avoidance of the periphrastic present perfect is typical only of the first book of the Dialogues. The remaining books contain a number of examples of the periphrastic perfect, such as the following: (128) Multum laborastis, iam quiescite. much labour-2PL.PF already rest-2PL.IMP (GD III.XIV.7, p. 246) ‘You have worked much; rest now.’ Ge wel habbað geworht & gewunnen. ye-NOM well have-3PL.PRES work-PA.PPL and win-PA.PPL Blinnað nu sume hwile. remain-2PL.IMP now some-FEM.DAT.SG while-DAT (GD MS C I.XV.44.23) ‘You have worked and toiled well. Rest now for a while.’ This sentence is similar to other examples such as (125) in that it is drawn from direct speech in quotation and translates a Latin perfect; however, in this case a periphrastic present perfect is used rather than a preterite. The different books of the Dialogues are in general similar in content, and there is no apparent internal motivation for the differences among them in their use of periphrastic 192 constructions; the survival of this text only in late copies makes it difficult to evaluate external possibilities such as multiple authorship (see Yerkes 1986). 4.3.5.2.2 The Pluperfect In comparison to the scarcity of the periphrastic perfect in Book I, periphrastic pluperfects occur with somewhat greater frequency. In accordance with the previously discussed tendencies, very few of them are used in literal translations of Latin pluperfects; examples from MS C occur only in later books, and the only such example in Book I is from the following passage of narrative in MS H: (129) Paene omne triticum, quod sibi almost all-NEUT.ACC.SG wheat-ACC REL-NEUT.ACC.SG self-DAT in stipendio totius anni paraverat, in stipend-ABL whole-GEN year-GEN prepare-3SG.PLPF invenit a filio suo pauperibus find-3SG.PF from son-ABL own-MASC.ABL.SG poor-DAT.PL expensum. expend-PA.PPL-NEUT.ABL.SG (GD I.IX.17, p. 118) ‘She found almost all the wheat that she had got ready for their support for all the year given to the poor by her own son.’ Þa gemette heo hire hwæte ealne then meet-3SG.PRET she-NOM she-GEN wheat-ACC all-MASC.ACC.SG be neah gedæledne fram hire agenum be-INF nigh deal-PA.PPL-MASC.ACC.SG from she-GEN own-DAT.SG sunu þearfendum mannum, eall þæt son-DAT needy-DAT.PL man-DAT.PL all-NOM REL-NEUT.ACC.SG heo ofer gær habban sceolde to bygleofan. she-NOM over year-ACC have-INF shall-3SG.PRET to sustenance-DAT (GD MS C I.XXVII.68.22) ‘Then she found nearly all her wheat distributed by her own son to needy men, all that she was supposed to have as support for the year’ 193 Þa afunde heo, þæt hire sunu then find-3SG.PRET she-NOM that she-GEN son-NOM hæfde þearfum gedæled forneah have-3SG.PRET needy-DAT.PL deal-PA.PPL for.nigh eallne þone hwæte, þe heo all-MASC.ACC.SG the-MASC.ACC.SG wheat-ACC REL she-NOM hæfde hire begiten to ealles have-3SG.PRET she-DAT get-PA.PPL to all-MASC.GEN.SG geares andlyfene. year-GEN sustenance-DAT (GD MS H I.XXVII.68.16) ‘Then she found that her son had distributed to the poor nearly all the wheat that she had obtained for all the year’s support.’ In MS H, the Latin pluperfect paraverat is translated by the periphrastic pluperfect hæfde begiten; in MS C, the corresponding clause is a paraphrase which makes no reference to the act of obtaining the wheat, and so there is no evidence for how the translator would have described this event. Additionally, in MS H the Latin participial clause with expensum, which is translated more literally in MS C, is expanded into a finite clause containing the periphrastic pluperfect hæfde gedæled. As the following example from a narrative passage shows, 5 the use of the periphrastic pluperfect to translate such participial clauses is more frequent: (130) Die […] altera erat pro utilitate day-ABL other-FEM.ABL.SG be-3SG.IMPF for utility-ABL monasterii causa constituta. monastery-GEN matter-NOM constitute-PA.PPL-FEM.NOM.SG Expletis […] hymnis matutinalibus, out.fill-PA.PPL-MASC.ABL.PL hymn-ABL.PL matutinal-MASC.ABL.PL Libertinus ad lectum abbatis venit[…]. Libertinus-NOM to bed-ACC abbot-GEN come-3SG.PRET (GD I.II.9, p. 84) ‘The other day, some business had been arranged for the benefit of 5 A parallel case is provided by (178) in Appendix A. 194 the monastery. Having celebrated matins, Libertinus went to the bed of the abbot.’ Þa oðre dæge hæfde Libertinus then other-MASC.DAT.SG day-DAT have-3SG.PRET Libertinus-NOM ane gemotstowe gecweden ymb one-FEM.ACC.SG meeting.place-ACC bespeak-PA.PPL about sume neodþearfe þæs mynstres, some-DAT necessity.need-DAT the-NEUT.GEN.SG minster-GEN.SG & þa þa he gefylled hæfde his and then when he-NOM fill-PA.PPL have-3SG.PRET he-GEN uhtsang & his gebedu, þa eode matins-ACC and he-GEN bead-ACC.PL then go-3SG.PRET he to ðam abbude[…]. he-NOM to the-MASC.DAT.SG abbot-DAT (GD MS C I.V.20.34) ‘The other day Libertinus had arranged a meeting about some need of the monastery, and then when he had celebrated matins and performed his prayers, then he went to the abbot.’ Þa oðre dæge hæfde Libertinus then other-MASC.DAT.SG day-DAT have-3SG.PRET Libertinus-NOM gecweden sume gemotstowe for bespeak-PA.PPL some-ACC meeting.place-ACC for sumre þarfe þæs mynstres. some-DAT.PL need-DAT.PL the-NEUT.GEN.SG minster-GEN.SG & þa þa he gefylled hæfde his and then when he-NOM fill-PA.PPL have-3SG.PRET he-GEN uhtsang & his gebedu, þa eode matins-ACC and he-GEN bead-ACC.PL then go-3SG.PRET he to þam abbode[…]. he-NOM to the-MASC.DAT.SG abbot-DAT (GD MS H I.V.20.34) ‘The other day Libertinus had arranged a meeting about some needs of the monastery, and then when he had celebrated matins and performed his prayers, then he went to the abbot.’ In addition to the periphrastic pluperfect used to render the Latin participial clause, (130) also provides an instance of the common translation practice of replacing a Latin passive construction, in this case the pluperfect constituta erat 195 ‘had been arranged’ with an Old English active construction, in this case hæfde gecweden ‘had arranged’. Most of the other periphrastic constructions in Book I are from passages of similarly non-literal translation, such as the following: (131) Statim ad viri Dei verbum ita immediately to man-GEN God-GEN word-ACC thus omnes egressae sunt, ut ne all-FEM.NOM.PL depart-FEM.NOM.PL be-3PL-PRES that NEG una quidem […] remaneret. one-FEM.NOM.SG even remain-3SG.IMPF.SUBJ (GD I.IX.15, p. 118) ‘Immediately upon the words of the man of God, they all departed so that not even one remained.’ Þa sona swa se Godes wer þa then soon so the-MASC.NOM.SG God-GEN man the-ACC.PL word gecweden hæfde, swa wæron word-ACC.PL speak-PA.PPL have-3SG.PRET so be-3PL.PRET hi sona ealle utgangende, þæt þær nan they-NOM soon all-NOM.PL out.go-PRES.PPL that there none anlipig […] to lafe ne wunode. single-NOM to remainder-DAT NEG dwell-3SG.PRET (GD MS C I.XXVI.67.15) ‘Then as soon as the man of God had spoken those words, they all immediately went out, so that not a single one remained there.’ Hi þa sona to þæs Godes weres they-NOM then soon to the-MASC.GEN.SG God-GEN man-GEN worde ealle endemes utferdon, swa þæt furðon word-DAT all-NOM together out.fare-3PL.PRET so that even þær an ne belaf[…]. there one NEG remain-3SG.PRET (GD MS H I.XXVI.67.15) ‘They then all went out together immediately upon the words of the man of God, so that not even one remained there.’ In this example, which illustrates such a use of the periphrastic pluperfect, the prepositional phrase found in the Latin original is rendered in MS C by a finite clause, while in MS H a construction more similar to the original is used. 196 Example (179) below is similar except that in both manuscripts the periphrastic pluperfect occurs in an interpolated clause. (132) Monasterii causa constituta monastery-GEN matter-NOM constitute-PA.PPL-FEM.NOM.SG est[…] quam declinare nequeo, be-3SG.PRES REL-FEM.ACC.PL decline-INF nor.go-1SG.PRES quia hesterno die me hodie because yester-MASC.ABL.SG day-ABL I-ACC today iturum promisi go-FUT.PPL-MASC.ACC.SG promise-3SG.PF (GD I.II.9, p. 84) ‘Some business for the monastery was arranged, which I cannot avoid, because yesterday I promised to go today.’ Ic hæfde gyrstandæge gecweden & I-NOM have-3SG yesterday-INST speak-PA.PPL and gehaten, þæt ic nu todæge cuman wolde promise-PA.PPL that I-NOM now today come-INF will-3SG.PRET ymb sume þearfe þises mynstres, about some-DAT need-DAT this-NEUT.GEN.SG minster-GEN forþon ic hit nu nænigra þinga for I-NOM it-ACC now NEG.any-NEUT.GEN.PL thing-GEN.PL ayldan ne mæg. delay-INF NEG may-3SG.PRES (GD MS C I.IV.21.16) ‘I had said and promised yesterday that I would now come today about some need of this monastery, wherefore I cannot now delay it in any way.’ Ic hæfde gyrstandæge gecweden, I-NOM have-3SG.PRET yesterday-INST speak-PA.PPL þæt ic nu todæge wolde cuman that I-NOM now today will-3SG.PRET come-INF ymbe sume þearfe þyses mynstres, about some-DAT need-DAT this-NEUT.GEN.SG minster-GEN & ic forþam hit nu yldan and I-NOM therefore it-ACC now delay-INF 197 ne mæg. NEG may-3SG.PRES (GD MS H I.IV.21.15) ‘I had said yesterday that I would now come today about some need of this monastery, and I therefore cannot now delay it.’ The remaining example of a periphrastic pluperfect from Book I is more difficult to characterize; it consists of the replacement within a passage of quoted dialogue of the original Latin present perfect with a periphrastic pluperfect. Such translations, which seem difficult to reconcile with the usual signification of the pluperfect, will be discussed below in connection with similar passages in Boethius. In addition to the periphrastic pluperfects cited above, Book I of the Dialogues contains many examples of preterites used in similar contexts to convey similar semantic and pragmatic content. As is shown by the following examples, all from narrative passages, preterites are favoured as translations for Latin pluperfects: (133) Unus eorum intulit quia ex culpa one-MASC.NOM.SG they-GEN infer-3SG.PF because out fault-ABL quam servo Dei in via fecerant, REL-FEM.ACC.SG slave-ABL God-GEN in way-ABL do-3PL.PLPF illa sui itineris dispendia that-FEM.ABL.SG self-GEN journey-GEN loss-ACC.PL tolerabant. tolerate-3PL.IMPF (GD I.II.3, p. 80) ‘One of them realized that because of the fault that they had done to God’s servant, they were thereby suffering the disruptions to their journey.’ Ða wæs an in ðam herge, then be-3SG.PRET one-NOM in the-MASC.DAT.SG host-DAT se þe ongæt & oncneow, that-MASC.NOM.SG REL grasp-3SG.PRET and know-3SG.PRET 198 þæt hi for þam gylte that they-NOM for the-MASC.DAT.SG guilt-DAT gehindrode wæron, þe hi þone hinder-PA.PPL be-3PL.PRET that they-NOM the-MASC.ACC.SG Godes man ær on wege his horses God-GEN man-ACC ere on way-DAT he-GEN horse-GEN bereafedon & hine his siðes bereave-3PL.PRET and he-ACC he-GEN journey-GEN agældon prevent-3PL.PRET (GD MS C I.II.15.10) ‘Then there was one of the horde who realized and knew that they were hindered because of their guilt, that they had previously on the way bereft the man of God of his horse and held him back from his journey.’ Þa ongæt & oncneow hyra an then grasp-3SG.PRET and know-3SG.PRET they-GEN one-NOM þæt hy gehindrode wæron for that they-NOM hinder-PA.PPL be-3PL.PRET for þam gylte, þe hi þone the-MASC.DAT.SG guilt-DAT that they-NOM the-MASC.ACC.SG Godes man ær on wege his horses God-GEN man-ACC ere on way-DAT he-GEN horse-GEN bereafedon & hine his siþes bereave-3PL.PRET and he-ACC he-GEN journey-GEN agældon prevent-3PL.PRET (GD MS H I.II.15.10) ‘Then one of them realized and knew that they were hindered because of their guilt, that they had previously on the way bereft the man of God of his horse and prevented him from his journey.’ (134) Buccellinus cum Francis venit. De Buccellinus-NOM with Frank-ABL.PL come-3SG.PF of monasterio […] rumor exierat quod monastery-ABL rumour-NOM out.go-3SG.PLPF that multas pecunias haberet. much-FEM.ACC.PL money-ACC.PL have-3SG.IMPF.SUBJ (GD I.II.4, p. 80) ‘Buccellinus came with the Franks. A rumour had gone out about the monastery, that it had much money.’ 199 Bucellinus wæs haten sum man, Buccellinus-NOM be-3SG.PRET call-PA.PPL some-NOM man-NOM se com mid Francum forþon REL.MASC.NOM.SG come-3SG.PRET with Frank-DAT.PL for þe he þone hlisan geacsode of that he-NOM the-MASC.ACC.SG rumour-ACC discover-3SG.PRET of þam mynstre […] & him gesægd wæs, the-NEUT.DAT.SG minster-DAT and he-DAT say-PA.PPL be-3SG.PRET þæt he mycel feoh on that he-NOM much-NEUT.ACC.SG money-ACC on þam mynstre hæfde the-NEUT.DAT.SG minster-DAT have-3SG.PRET (GD MS C I.III.16.6) ‘There was a man called Buccellinus, who came with the Franks because he had learned of the rumour about the monastery, and it was said to him that he [the provost] had much money in the monastery.’ Bucellinus com mid Francum forþam Buccellinus-NOM come-3SG.PRET with Frank-DAT.PL for þe he gehyrde þone hlisan of that he-NOM hear-3SG.PRET the-MASC.ACC.SG rumour-ACC of þam mynstre […] & him wæs gesæd, the-NEUT.DAT.SG minster-DAT and he-DAT be-3SG.PRET say-PA.PPL þæt he mycel feoh on that he-NOM much-NEUT.ACC.SG money-ACC on þam mynstre hæfde the-NEUT.DAT.SG minster-DAT have-3SG.PRET (GD MS H I.III.16.6) ‘Buccellinus came with the Franks because he had heard the rumour about the monastery, and it was said to him that he [the provost] had much money in the monastery.’ Many similar examples exist, including (180)– (188) below. It may be noted that not all such passages are uniformly literal translations throughout. For example, in (133) a reference in the Latin to having done wrong is replaced in Old English by a specific list of the injuries done, while in (134) a Latin sentence in which rumor ‘rumour’ is the subject is translated by Old English sentences in which hlisa ‘rumour’ is the object. Although within this text there is a significant 200 association between looser translations and periphrastic forms, it can be seen from the above examples that it is similarly possible to use the simple preterite in such cases. (135) Ea hora saluti restitutam that-FEM.ABL.SG hour-ABL health-DAT restitute-PA.PPL-FEM.ACC.SG Dei virginem agnovit, qua […] salutem God-GEN virgin-ACC learn-3SG.PF REL-FEM.ABL.SG health-ACC illius Dei famulus Equitius longe that-GEN.SG God-GEN servant-NOM Equitius-NOM far positus dixit. place-PA.PPL-MASC.NOM.SG say-3SG.PF (GD I.IV.6, p. 90) ‘He learned that the maiden of God was restored to health at the hour at which God’s servant Equitius, situated far away, pronounced her health.’ Se munuc […] sona ongæt, þæt the-NOM.MASC.SG monk-NOM soon discover-3SG.PRET that seo Godes fæmne wæs gehæled the-FEM.NOM.SG God-GEN maiden-NOM be-3SG.PRET heal-PA.PPL in þa ylcan tide, þe in the-FEM.ACC.SG same-FEM.ACC.SG tide-ACC REL se Godes þeow cyrde. Þeah the-MASC.NOM.SG God-GEN servant-NOM turn-3SG.PRET though he feor wære, Æquitius þa hire he-NOM far be-3SG.PRET.SUBJ Equitius-NOM then she-GEN hæle gecwæþ & gehet. health-ACC bespeak-3SG.PRET and promise-3SG.PRET (GD MS C I.VIII.29.27) ‘The monk soon found that the maiden of God was healed at the same time that the servant of God turned back. Far though he was, Equitius had then bespoken and promised her health.’ Se munuc […] sona oncneow, þæt the-NOM.MASC.SG monk-NOM soon discover-3SG.PRET that seo Godes fæmne wæs gehæled the-FEM.NOM.SG God-GEN maiden-NOM be-3SG.PRET heal-PA.PPL on þære ylcan tide, þe on the-FEM.DAT.SG same-FEM.DAT.SG tide-DAT REL 201 se Godes þeow Æquitius hyre the-MASC.NOM.SG God-GEN servant-NOM Equitius-NOM she-GEN hæle gecwæð, þeah he feorr health-ACC bespeak-3SG.PRET though he-NOM far wære. be-3SG.PRET.SUBJ (GD MS H I.VIII.29.25) ‘The monk soon found that the maiden of God was healed at the same time that God’s servant Equitius promised her health.’ (136) Percussit semel[…]. strike-3SG.PF once (GD I.II.7, p. 82) ‘He struck once.’ Þa sloh he ænes on þæt then strike-3SG.PRET he-NOM once on the-NEUT.ACC.SG wæter & wolde him weg gewyrcan water-ACC and will-3SG.PRET he-DAT way-ACC work-INF swa swa he his lareow ær geseah[…]. so so he-NOM he-GEN teacher-ACC ere see-3SG.PRET (GD MS C I.IV.19.15) ‘Then he struck at the water and wanted to make his way just as he had seen his teacher do before.’ Þa sloh he æne on þæt then strike-3SG.PRET he-NOM once on the-NEUT.ACC.SG wæter […] & wolde him weg gewyrcean water-ACC and will-3SG.PRET he-DAT way-ACC work-INF swa swa he ær geseah his lareow so so he-NOM ere see-3SG.PRET he-GEN teacher-ACC don[…]. do-INF (GD MS H I.IV.19.15) ‘Then he struck at the water and wanted to make his way just as he had seen his teacher do before.’ The foregoing examples from narrative passages provide a further illustration of the diversity of environments in which simple preterites are found. As discussed in Section 3.3.2.2, events compatible with the definition of the pluperfect as a semantic category are not necessarily marked as such; in (135), the relevant events are denoted by a Latin perfect, and translated in Old English by a simple preterite, 202 and a parallel case is provided by (189) below, from the framing dialogue. In such cases, there is no conclusive evidence for how such events were regarded by the Old English translator; however, the existence of examples such as (88) from Section 3.4.2.2, in which a periphrastic pluperfect is used to translate a Latin perfect, and the interchangeable nature of simple and periphrastic constructions within this semantic domain preclude the automatic dismissal of such examples as necessarily outside the relevant semantic sphere. Another example that illustrates the diversity of translation relationships is (97) above, which shows the preterite subjunctive used in MS H to translate a Latin perfect infinitive in indirect speech, one of the few examples of such infinitive constructions found in the data; in MS C, the passage is changed to direct speech. The use of the preterite with pluperfect signification in an interpolated passage can be seen in (136). The variety of examples discussed here show that the preterite, like the periphrastic pluperfect, can be used in a diverse range of contexts and settings as a means of conveying pluperfect-type meaning, and that there is often great similarity between the contexts in which the two forms are used; even within the Dialogues, patterns only emerge in terms of general statistical tendencies. 4.3.5.3 Bede 4.3.5.3.1 The Perfect In contrast to the Dialogues, Bede shows no significant association between the form of the Latin original and the choice in Old English between the periphrastic tenses and the preterite. Instead, as discussed above, the preterite is generally preferred as a means of expressing the relevant semantic content; although periphrastic forms occur, they are in the minority. Within the sample 203 analysed in the present study, only two instances of the periphrastic present perfect occur, both within direct speech: (137) Hostium manus[…] Domino donante enemy-GEN.PL hand-ACC.PL Lord-ABL give-PRES.PPL-MASC.ABL.SG evasisti; […] regnum[…] ipso evade-2SG.PF realm-ACC self-MASC.ABL.SG largiente percepisti. bestow-PRES.PPL-MASC.ABL.SG receive-2SG.PF (Bede II.12, I, p. 280) ‘With the Lord giving, you have escaped the hands of your enemies; with Himself granting, you have taken possession of your kingdom.’ Þu nu hafast þurh Godes gife thou-NOM now have-2SG.PRES through God-GEN gift-DAT þinra feonda hond beswicade[…] & thy-GEN.PL fiend-GEN.PL hand-ACC.PL elude-PA.PPL-ACC.PL and þurh his sylene & gife þæm through he-GEN grant-DAT and gift-DAT the- NEUT.DAT.SG rice onfenge[…]. kingdom-DAT on.take-2SG.PRET (Bede II.9.132.22) ‘You have now through God’s gift eluded the hands of your enemies and through His munificence and gift succeeded to the kingdom.’ (138) Didici[…] quia nihil omnino virtutis habet, learn-1SG.PF because nothing entirely virtue-GEN have-3SG.PRES nihil utilitatis religio illa nothing utility-GEN religion-NOM that-FEM.NOM.SG quam hucusque tenuimus[…]. REL-FEM.ACC.SG hither.to hold-1SG.PF (Bede II.13, I, p. 282) ‘I have learned that the religion which we have held until now has nothing of virtue, nothing of use at all.’ Ic cuðlice geleornad hæbbe, þæt eallinga I-NOM certainly learn-PA.PPL have-1SG.PRES that completely nawiht mægenes ne nyttnes hafað sio nought main-GEN nor use-GEN have-3SG.PRES the-FEM.NOM.SG æfæstnesse, þe we oð ðis hæfdon religion-NOM REL we-NOM until this-NEUT.ACC.SG have-1PL.PRET 204 & beeodon. and observe-1PL.PRET (Bede II.10.134.12) ‘I have certainly learned that the religion that we have held and observed until now has nothing at all of power or use.’ Each of these examples contains two occurrences of the Latin perfect, and in each case the first occurrence is translated by a periphrastic construction and the second by a preterite; as mentioned in Section 3.3.2.2, one possible explanation of such sentences is that the periphrastic perfect provides sufficient indication of the time frame in question that no further marking is needed on subsequent verbs (see further Mitchell 1975, 159–66). However, as following examples will show, pluperfect-like preterites occur in many contexts with no periphrastic constructions to provide such marking. In fact, the periphrastic present perfect is overrepresented within the sample used for the present study; a search of the entire text of Bede using the York-Toronto-Helsinki Corpus (Taylor et al. 2003) for all sentences containing a present tense of have and a past participle, regardless of syntax or semantics, identified only two other such sentences anywhere in the text, both of which were perfects (at I.16.76.10 and V.9.410.17). The scarcity of these constructions in Bede contrasts with their frequent use in other contemporary texts such as Boethius. As a means of conveying content similar to that of the present perfect, the simple preterite occurs far more frequently within Bede. (139) Vere resurrexi a morte[…]. truly resurge-1SG.PF from death-ABL (Bede V.12, II, p. 252) ‘I have truly risen again from death.’ 205 Ic soðlice fram deaðe aaras[…]. I-NOM truly from death-DAT arise-1SG.PRET (Bede III.20.246.9) ‘I have truly arisen from death.’ (140) Erat […] presbyter vocabulo Ceadda, frater […] be-3SG.IMPF priest-NOM vocable-ABL Chad-NOM brother-NOM Ceddi, cuius saepius meminimus[…]. Cedd-GEN REL-MASC.GEN.SG often-COMP remember-1PL.PF (Bede III.28, I, p. 490) ‘There was a priest by the name of Chad, brother of Cedd, of whom we have often made mention.’ Wæs mæssepreost, se wæs be-3SG.PRET mass.priest-NOM REL-MASC.NOM.SG be-3SG.PRET Ceadda haten, Ceddes broðor, þæs Chad-NOM call-PA.PPL Cedd-GEN brother-NOM REL-MASC.GEN.SG we beforan gelome gemyndgedon. we-NOM before habitually mention-1PL.PRET (Bede III.20.246.9) ‘There was a priest who was called Chad, the brother of Cedd, of whom we have often made mention before.’ (141) Quis enim ea quae who-MASC.NOM.SG for that-NEUT.ACC.PL REL-NEUT.ACC.PL per stultitiam colui nunc ad exemplum through folly-ACC cultivate-1SG.PF now to example-ACC omnium aptius quam ipse per all-GEN.PL aptly-COMP than self-MASC.NOM.SG through sapientiam mihi a Deo vero wisdom-ACC I-DAT from God-ABL true-MASC.ABL.SG donatam destruam? give-PA.PPL-FEM.ACC.SG destroy-1SG.PRES.SUBJ (Bede II.13, I, p. 287) ‘For who shall destroy, as an example to all, those things that I worshipped through folly, more fittingly than I myself through the wisdom given to me by the true God?’ Hwa mæg þa nu eað, þe ic who-NOM may-3SG.PRES that-ACC.PL now easy REL I-NOM longe mid dysignesse beeode to bysene long with dizziness-DAT observe-1SG.PRET to example-DAT oðerra monna gerisenlecor toweorpan, þonne ic other-DAT.PL man-DAT.PL aptly-COMP destroy-INF than I-NOM 206 seolfa þurh þa snytro, þe ic from self through the-FEM.ACC.SG wisdom-ACC REL I-NOM from þæm soðan Gode onfeng? the-MASC.DAT.SG true-MASC.DAT.SG God-DAT receive-1SG.PRET (Bede II.10.136.27) ‘Who can now easily destroy those things that I long served with folly as an example to other men more fittingly than I myself through the wisdom that I have received through the true God?’ Example (139), like (137) and (138) above and (190)– (192) below, is drawn from a passage of direct discourse corresponding to a Latin original containing a perfect, in which the event in question has a pragmatically salient relationship to the present; however, in all these cases the preterite is used instead of a periphrastic construction. The same is true for (140), from an expository passage in the historical narrative; in (141), which is again from a passage of dialogue, the preterite is used to translate a participial clause whose temporal significance is similar to that of the perfect. (142) Nihilominus multi sunt qui nonetheless many-MASC.NOM.PL be-3PL.PRES REL-MASC.NOM.PL ampliora a te beneficia quam ample-COMP-NEUT.ACC.PL from thou-ABL benefit-ACC.PL than ego, et maiores accipiunt dignitates, I-NOM and greater-FEM.ACC.PL receive-3PL.PRES dignity-ACC.PL magisque prosperantur in omnibus[…]. more.and prosper-3PL.PRES.MP in all-NEUT.ABL.PL (Bede II.13, I, p. 282) ‘Nonetheless there are many who receive more ample benefits, greater dignities, from you than I, and prosper more in all things.’ Noht þon læs monige syndon, nought the-NEUT.INST.SG less many-NOM.PL be-3PL.PRES þa þe maran gefe & fremsumnesse that-NOM.PL REL more-ACC.SG gift-ACC.SG and benefit-ACC.SG æt þe onfengon þonne ic, & on eallum at thou-DAT receive-3PL.PRET than I-NOM and on all-DAT.PL 207 þingum maran gesynto hæfdon. thing-DAT.PL more-ACC.SG prosperity-ACC have-3PL.PRET (Bede II.10.134.15) ‘Nonetheless there are many who have received more gifts and benefits from you than I, and in all things have had more prosperity.’ Examples such as (142) above, which shows the use of the preterite to render a Latin present tense in direct discourse, further illustrate the perceived suitability of the preterite as an expression of events connected to the present. 4.3.5.3.2 The Pluperfect As in the Dialogues, in Bede the periphrastic pluperfect is more common than the periphrastic perfect; however, as Table 7 above shows, this construction is nevertheless proportionally less frequent in Bede than in many other texts. (143) Coeperunt illi mox idolatriae, begin-3PL.PRET that-MASC.NOM.PL soon idolatry-GEN quam viventi eo aliquantulum REL-FEM.ACC.SG live-PRES.PPL-DAT he-DAT somewhat intermisisse videbantur palam servire[…]. cease-INF.PF see-3PL.PRES.MP openly serve-INF (Bede II.5, I, p. 228) ‘They soon began to practise openly the idolatry that they seemed to have abandoned to some extent with their father living.’ Þa ongunnon heo sona openlice then begin-3PL.PRET they-NOM soon openly deofolgildum þeowian, þe monnum þuhte devil.worship-DAT.PL serve-INF REL man-DAT.PL think-3SG.PRET þæt heo hwæthugu forlæten hæfde that they-NOM somewhat relinquish-PA.PPL have-3PL.PRET bi þæm fæder lifiendum. by the-MASC.DAT.SG father-DAT live-PRES.PPL-DAT (Bede II.V.112.3) ‘Then they soon began to practise devil-worship openly, which it seemed to people that they had relinquished somewhat with their father living.’ 208 (144) Iuxta quod dispositum beside what-NEUT.ACC.SG arrange-PA.PPL-NEUT.NOM.SG fuerat, ordinatur episcopus vir be-3SG.PLPF ordain-3SG.PRES.MP bishop-NOM man-NOM Deo dilectus Paulinus[…]. God-DAT pleasing-MASC.NOM.SG Paulinus-NOM (Bede II.9, I, p. 247) ‘In accordance with what had been arranged, Paulinus, the man dear to God, is ordained bishop.’ Æfter þone þe heo ær funden after that-NEUT.INST.SG REL they-NOM ere find-PA.PPL hæfdon, wæs gehalgod to biscope have-3PL.PRET be-3SG.PRET hallow-PA.PPL to bishop-DAT Gode se leofa wer Sanctus God-DAT the-MASC.NOM.SG dear-NOM.SG man-NOM Saint-NOM Paulinus[…]. Paulinus-NOM (Bede II.8.120.24) ‘In accordance with that which they had previously arranged, there was consecrated as bishop that man dear to God, Saint Paulinus.’ These examples, drawn from passages of historical narrative, illustrate some uses of the periphrastic pluperfect in relatively literal translations. In (143) this construction translates an infinitive clause of a temporal signification comparable to the pluperfect; a Latin infinitive clause is also translated by a periphrasis in (193) below. In (194) the periphrasis translates an absolute participial clause of similar temporal import, while in (90) above it correponds to a Latin pluperfect. In (144), the translation is only slightly further removed from the original. The Latin pluperfect passive is translated by a corresponding active construction in Old English; as mentioned in Section 3.3.3.3, at this period the verb wesan ‘be’ was defective and lacked a past participle, and so the periphrastic perfect and pluperfect could only be used in active constructions. Another noteworthy feature of this example is that in Latin the pluperfect is somewhat incongruously used in 209 combination with a historical present, ordinatur ‘is ordained’, whereas in Old English the corresponding verb is changed to a preterite. (145) Postquam itineris sui causam […] papae after journey-GEN own-NEUT.GEN.SG cause-ACC pope-DAT apostolico patefecit, non multo post et apostolic-MASC.DAT.SG declare-3SG.PF not much after and ipse et omnes pene qui self-MASC.NOM.SG and all-MASC.NOM.PL nearly REL-MASC.NOM.PL cum eo advenerant socii, pestilentia with he-ABL to.come-3PL.PLPF fellow-NOM.PL pestilence-ABL superveniente deleti sunt. overcome-PRES.PPL-ABL delete-PA.PPL-MASC.NOM.PL be-3PL.PRES (Bede IV.1, II, p. 4) ‘After he declared the cause of their journey to the Apostolic Pope, not long afterward both he himself and nearly all the companions who had come with him were destroyed by an overpowering pestilence.’ Æfter þon þe he þone after that-NEUT.INST.SG REL he-NOM the-MASC.ACC.SG intingan his siðfætes þæm matter-ACC he-GEN journey-GEN the-MASC.DAT.SG apostolican papan gecyðed hæfde, apostolic-DAT.SG pope-DAT inform-PA.PPL have-3SG.PRET þa æfter medmicelre tide se then after mid.much-FEM.DAT.SG tide-DAT that-MASC.NOM.SG Wigheard & lytestne alle his Wigheard-NOM and almost-MASC.ACC.SG all-NOM.PL he-GEN geferan, þa ðe mid him cwomon, fellow-NOM.PL that-NOM.PL REL with he-DAT come-3SG.PRET þy ofercumendan woole the-MASC.INST.SG overcome-PRES.PPL-INST petilence-INST fordilgade wæron & forðgeleorde. destroy-PA.PPL be-3PL.PRET and depart-3PL.PRET (Bede IV.1.252.19) ‘After he had made the purpose of his journey known to the Apostolic Pope, then after a short time this Wigheard and almost all his companions, those who had come with him, were destroyed by the overpowering pestilence and died.’ 210 (146) pugnanti adversus regem, a quo fight-PRES.PPL-DAT against king-ACC from REL-MASC.ABL.SG homicida ille, qui eum homicide-NOM that-MASC.NOM.SG REL-MASC.NOM.SG he-ACC vulneraverat, missus est wound-3SG.PLPF send-PA.PPL-MASC.NOM.SG be-3SG.PRES (Bede II.9, I, p. 250) ‘to the fight against the king from whom that murderer who had wounded him was sent’ on þæm gewinne, þe he gehogad on the-NEUT.DAT.SG strife-DAT REL he think-PA.PPL hæfde wið þam cyninge, from have-3SG.PRET with the-MASC.DAT.SG king-DAT from þæm þe se myrðra ær that-MASC.DAT.SG REL the-MASC.NOM.SG murderer ere sended wæs, se þe hine send-PA.PPL be-3SG.PRET that-MASC.NOM.SG REL he-ACC gewundade wound-3SG.PRET (Bede II.8.124.1) ‘in the war that he had contemplated against the king from whom the murderer was previously sent, he who had wounded him.’ Similarly to (137) and (138), the two examples above contain examples of the periphrastic pluperfect and the preterite used for similar purposes within the same sentence. In (145), the periphrastic pluperfect is used to translate a Latin perfect tense, which is not infrequently used in this way following postquam ‘after’, while the Latin pluperfect advenerant ‘had arrived’ is translated simply by an Old English preterite. In (146), the pluperfect is used in an interpolated clause, while the Latin pluperfect vulneraverat ‘had wounded’ is rendered by the simple preterite in Old English. The preceding examples show the diversity of the environments in which the periphrastic pluperfect is found in Bede; however, as stated previously, the simple preterite occurs far more frequently within this work in such contexts. The 211 following selection of examples should serve to illustrate the diversity of the environments in which preterites with this function could occur: (147) volens […] tuitionem eis, want-PRES.PPL-NOM protection-ACC that-MASC.DAT.PL quos et quorum doctrinam REL-MASC.ACC.PL and REL-MASC.GEN.PL doctrine-ACC susceperat, praestare under.take-3SG.PLPF provide-INF (Bede II.5, I, p. 226) ‘wishing to provide protection for those whom and whose doctrine he had received’ Wolde he ðam gescyldnysse gegearwian, will-3SG.PRET he-NOM that-DAT.PL protection-ACC prepare-INF þe he heora lare onfeng. REL he-NOM they-GEN lore-ACC receive-3SG.PRET (Bede II.5.110.15) ‘He wanted to provide protection to those whose teaching he had received.’ (148) Post annum ex quo abierunt, after year-ACC out REL-NEUT.ABL.SG off.go-3PL.PF reversi sunt. return-PA.PPL-MASC.NOM.PL be-3PL.PRES (Bede II.6, I, p. 234) ‘After a year from when they departed, they returned.’ Heo ymb an ger ham hwurfon they-NOM about one-ACC year-ACC home turn-3PL.PRET ðæs þe heo ær of Breotone that-NEUT.GEN.SG REL they-NOM ere of Britain-DAT ferdon. fare-3PL.PRET (Bede II.6.116.9) ‘They returned home a year after they had journeyed from Britain.’ (149) Anathematizato omni idolatriae anathematize-PA.PPL-MASC.ABL.SG all-MASC.ABL.SG idolatry-GEN cultu, […] suscepit fidem Christi[…]. practice-ABL under.take-3SG.PF faith-ACC Christ-GEN (Bede II.6, I, p. 234) ‘All practice of idolatry having been renounced, he received the faith of Christ.’ 212 He towearp all þa bigong he-NOM destroy-3SG.PRET all the-FEM.ACC.SG practice-ACC þara deofolgelda, þa he ær the-GEN.PL devil-worship-GEN.PL REL-ACC.PL he-NOM ere beeode, […] ond se biscop hine observe-3SG.PRET and the-MASC.NOM.SG bishop-NOM he-ACC gefulwade. baptize-ACC (Bede II.6.114.31) ‘He cast aside all the practice of devil-worship that he previously observed, and the bishop baptized him.’ In (147), the preterite is used to translate the Latin pluperfect; similar examples can be seen above in (89) and below in (195). In (148) it translates a Latin perfect of similar meaning in an ‘after’ clause comparable to that translated using a periphrastic pluperfect in (145). In (149) the relevant preterite occurs in an interpolated clause. Another relevant example is (96), which shows the preterite used to translate a Latin participial clause similar to that found in (194). It may be seen from these examples that the preterite, which can be used in Old English to denote anterior events for all the purposes seen here, is generally preferred in Bede above the periphrastic pluperfect, although periphrastic forms also occur. 4.3.5.4 Boethius 4.3.5.4.1 The Perfect Like Bede, Boethius displays no significant correlation between the form of the Latin original and the choice between simple and periphrastic forms; unlike Bede, however, Boethius makes liberal use of the periphrastic forms in all contexts. In the case of Boethius, comparison to the Latin original is complicated by the fact that Boethius is a much freer translation than Bede or the Dialogues, and in many cases the views that the translator expresses differ markedly from 213 those expressed in the Latin original (see Godden and Irvine 2009, I, 56–61). Where syntactic parallels to the original are apparent even in freely translated passages, the Latin original is given for comparison, but examples with no obvious Latin antecedent are also discussed; the present work largely follows the judgement of Godden and Irvine (2009) in deciding which passages fall into the latter category. (150) Si penitus aegritudinis tuae causas if innermost illness-GEN thy-FEM.GEN.SG cause-ACC.PL habitumque cognovi, fortunae prioris condition-ACC.and learn-1SG.PF fortune-GEN prior-FEM.GEN.SG affectu desiderioque tabescis[…]. longing-ABL desire-ABL.and waste-2SG.PRES (Boece IIp1.2) ‘If I have understood the inwardness of your illness, you are wasting away with longing and desire for your prior fortune.’ Gif ic þine unrotnesse on riht ongiten if I-NOM thy-FEM.ACC.SG grief-ACC on right understand-PA.PPL hæbbe, þonne nis þe nauht swiðor have-1SG.PRES then NEG.be-3SG.PRES thou-DAT naught stronger ðonne þæt þæt þu forloren hæfst than that that thou-NOM lose-PA.PPL have-2SG.PRES þa woruldsælþa ðe ðu ær hæfdest[…]. the-ACC.PL fortune-ACC.PL REL thou-NOM ere have-3SG.PRET (Boece VII.251.2) ‘If I have understood your grief aright, then there is nothing worse with you than that you have lost the worldly felicities that you had before.’ (151) Nulla tibi a nobis est none-NOM.FEM.SG thou-DAT of we-ABL be-3SG.PRES allata violentia off.bear-PA.PPL-NOM.FEM.SG violence-ABL (Boece IIp2.6) ‘Nothing has been taken away from you violently by us.’ 214 Hæbbe ic awer benumen þinra have-1SG.PRES I-NOM anywhere deprive-PA.PPL thy-GEN.PL gifena[…]? gift-GEN.PL (Boece VII.254.76) ‘Have I in any way deprived you of your gifts?’ (152) Hactenus mendacis formam felicitatis so.far false-FEM.GEN.SG form-ACC felicity-GEN ostendisse sufficerit indicate-INF.PF suffice-3SG.FUT.PF (Boece IIIp9.1) ‘Thus far it will have sufficed to indicate the form of false felicity.’ Genog ic þe hæbbe nu gereht enough I-NOM thou-DAT have-1SG.PRET now tell-PA.PPL ymbe þa anlicnessa and ymbe about the-FEM.DAT.PL likeness-DAT.PL and about þa sceadwa þære soðan the-FEM.DAT.PL shadow-DAT.PL the-FEM.GEN.SG true-GEN.SG gesælðe. felicity-GEN (Boece XXXIII.310.2) ‘I have said enough about the likeness and about the shadows of the true felicity.’ Example (150), like (196) and (197) below, shows the use of the periphrastic perfect to translate the Latin perfect tense; unlike Bede and the Dialogues, Boethius uses periphrastic forms freely in such contexts. In (150), an additional example of an interpolated perfect can be seen. Example (151) is similar to these except that a perfect passive in Latin is translated by an active construction in Old English, as in (144) above. A relatively uncommon replacement of a Latin future perfect by an Old English present perfect is seen in (152). Like all the other examples from Boethius discussed here, these examples are drawn from the philosophical dialogue of which most of the work consists; as examples such as 215 (153) below show, the Old English translator explicitly characterizes this material as direct discourse in quotation. (153) Atqui scis unde cuncta yet know-2SG.PRES whence whole-NEUT.NOM.PL processerint. Novi, inquam, deumque proceed-3SG.PF.SUBJ know-1SG.PF say-1SG God-ACC.and esse respondi. be-INF respond-1SG.PF (Boece Ip6.11) ‘“Nevertheless, do you know whence everything has come?” “I know,” I said, and I responded that it is God.’ Þa cwæð se wisdom. then speak-3SG.PRET the-NOM.MASC.SG wisdom-NOM Wast þu hwonan ælc wuht know-2SG.PRES thou-NOM whence each-NOM wight-NOM come? Þa andwyrde þæt come-3SG.PRET.SUBJ then answer-3SG.PRET the-NEUT.NOM.SG mod and cwæð. Ic wat þæt mood-NOM and speak-3SG.PRET I-NOM know-1SG.PRES that ælc wuht fram Gode com. each-NOM wight-NOM from God-DAT come-3SG.PRET (Boece V.249.66) ‘Then Wisdom said, “Do you know whence each creature has come?” Then the Mind answered and said, “I know that each creature has come from God.”’ (154) In omni fortunae tuae censu in all-ABL.SG fortune-GEN thy-FEM.GEN.SG property-ABL pretiosissimum possidebas id precious-SUP-NEUT.ACC.SG possess-2SG.IMPF that-NEUT.ACC.SG tibi divinitus inlaesum adhuc thou-DAT heavenly unharmed.NEUT.ACC.SG hereto inviolatumque servatur[…]. inviolate-NEUT.ACC.SG.and save-3SG.PRES.MP (Boece IIp4.4) ‘In all your fortunes you hitherto possessed the most precious of your property, that from Heaven to you, unharmed and it is kept inviolate.’ 216 Ac þu hæfst git gesund gehealde but thou-NOM have-2SG.PRES yet sound hold-PA.PPL eall þæt deorwyrðoste[…]. all that-NEUT.ACC.SG dearest-ACC (Boece X.259.18) ‘But you have yet kept unharmed all that is dearest to you.’ A variety of less literal translation practices are exemplified by passages from Boethius. In (153) the perfect results from the expansion of an ellipsis in the Latin original. Periphrastic perfects corresponding to Latin present tenses, like the semantically similar preterites seen above, occur frequently; in some cases, such as (198) and (199) below, the same event is merely depicted from two different temporal perspectives, while in others, such as (200), there is an interchange of an effect (in this case, knowing) for its cause (having found out). In (154), a reference in the Latin to something’s being kept is replaced by a reference in Old English to the addressee’s having kept it. (155) Vivit uxor[…] pudore praecellens[…]. live-3SG.PRES wife-NOM decency-ABL surpass-PRES.PPL-NOM (Boece IIp4.6) ‘Your wife lives, excellent in modesty.’ Hu ne leofoð þin wif[…]; how NEG live-3SG.PRES thy-NEUT.NOM.SG wife seo hafð ealle oþru she-NOM have-3SG.PRES all-NEUT.DAT.PL other-NEUT.DAT.PL wif oferþungen mid clennesse. wife-DAT.PL surpass-PA.PPL with cleanness-DAT (Boece X.259.27) ‘Does your wife not live? She has surpassed all other women in modesty.’ (156) Ille nuptiis felix that-MASC.NOM.SG marriage-ABL.PL happy-MASC.NOM.SG orbus liberis alieno bereft-MASC.NOM.SG child-ABL.PL alien-MASC.DAT.SG 217 censum nutrit heredi[…]. property-ACC nourish-3SG.PRES heir-DAT (Boece IIp4.14) ‘That one, happily married but lacking children, maintains his property for an alien heir.’ Manige habbað genog gesælilice gewifod many-MASC.NOM.PL have-NOM.PL enough happily wive-PA.PPL ac for bearnleste eallne þone but for childlessness-DAT all-MASC.ACC.SG the-MASC.ACC.SG welan þe hi gegaderigað hi wealth-ACC REL they-NOM gather-3PL.PRES they-NOM lætað fræmdum to brucanne let-3PL.PRES alien-DAT.PL to use-INF.DAT (Boece XI.261.21) ‘Many have married happily enough, but from childlessness they leave all the wealth that they gather to the use of an alien.’ (157) Neque enim fas est homini nor for lawful-NOM be-3SG.PRES man-DAT cunctas divinae operae machinas whole-FEM.ACC.PL divine-FEM.GEN.SG work-GEN machine-ACC.PL vel ingenio comprehendere vel explicare sermone. or wit-ABL comprehend-INF or explain-INF speech-ABL (Boece IVp6.54) ‘For neither is man permitted either to comprehend all the devices of the divine work through intelligence or to explain them in speech.’ Ac hit is nanum men but it-NOM NEG.be-3SG.PRES no-MASC.DAT.SG man-DAT alefed þæt he mæg witan grant-PA.PPL that he-NOM may-3SG.PRES know-INF eall þæt God getiohhod all-ACC REL-NEUT.ACC.SG God-NOM determine-PA.PPL hæfð, ne eac arecan þæt have-3SG.PRES NEG also tell-INF that-NEUT.ACC.SG þæt he geworht hæfð. REL-NEUT.ACC.SG he-NOM work-PA.PPL have-3SG.PRES (Boece XXXIX.359.36) ‘But it is not permitted to any man that he may know all that God has determined or tell that which He has wrought.’ 218 In (155), the periphrastic perfect is used to translate a Latin participial adjective, praecellens ‘surpassing’, perhaps to make it explicit that the excellence described is a state already attained, while (156) translates the adjective phrase nuptiis felix ‘happy in his marriage’, describing a present state, with a reference to the past event of having married happily. Such an interchange of cause and effect can also be seen in (201) below. In (157), a noun phrase referring to divine works has been expanded into a more explicit verbal construction referring to what God has planned and wrought. In (202) and (203) below, present perfects occur in passages interpolated into the Old English version. From the range of examples cited above, it can be seen that Boethius makes much more liberal use of the periphrastic perfect than the other texts examined here, both in literal translation and in original composition. Despite the free use of the periphrastic perfect in Boethius, perfect-like preterites can be found with a similar sense in a variety of contexts. Such forms are used in relatively literal translations of the Latin perfect, as in (204) and (205) from Appendix A, as well as in less literal examples such as the following: (158) Promovimus[…] aliquantum[…]. promote-1PL.PF somewhat (Boece IIp4.11) ‘We have progressed somewhat.’ Ic wene þeah þæt ic I-NOM ween-1SG.PRES though that I-NOM hwæthweganunges þe up ahofe of somewhat thou-ACC up heave-1SG-PRET.SUBJ of þære unrotnesse[…]. the-FEM.DAT.SG sorrow-DAT (Boece XI.260.2) ‘I think, though, that I have raised you up somewhat from sorrow.’ 219 (159) Quae nunc tibi est tanti what-FEM.NOM now thou-DAT be-3SG.PRES so.much-MASC.GEN.SG causa maeroris cause-NOM sorrow-GEN (Boece IIp1.12) ‘That which is now the cause of such great sorrow for you’ Þa ilcan þe ðe gedydon the-NOM.PL same-NOM.PL REL thou-DAT do-3PL.PRET nu þas gnornung now this-FEM.ACC.SG sorrow (Boece VII.252.38) ‘The same things that have now made this sorrow for you’ In (158), although there is a perfect in the Latin original, the corresponding Old English passage is a more extensive paraphrase of the Latin, having little in common with the original except the temporal frame of reference. Example (159), like the periphrastic examples seen above, paraphrases a Latin noun (causa ‘cause’) by means of a verb phrase, while in (206) below the relevant preterites occur in an interpolated passage. (160) Novum, credo, aliquid inusitatumque new-ACC believe-1SG.PRES something-ACC unusual-ACC.and vidisti [quam non viderunt alii]. see-2SG.PF [as NEG see-3PL.PF other-MASC.NOM.PL] (Boece IIp1.9) ‘You have, I believe, seen something new and unusual [such as others have not seen].’ Wenst þu þæt hit hwæt niwes ween-2SG.PRES thou-NOM that it what-NOM new-NEUT.GEN.SG sie oþþe hwæthwega ungewunelices be-3SG.PRES.SUBJ or anything-NOM unusual-NEUT.GEN.SG þæt þe on becumen is, swelce that thou-DAT on become-PA.PPL be-3SG.PRES such oðrum mannum ær þæt other-MASC.DAT.PL man-DAT.PL ere the-NEUT.NOM.SG 220 ilce ne eglede? same-NOM NEG ail-3SG.PRET (Boece VII.252.22) ‘Do you think that it is something new, or anything unusual that has happened to you, as if the same thing has not afflicted other men before?’ An illustration of the sometimes complex textual issues surrounding Boethius is provided by (160); the Latin phrase given in brackets is not part of the original text, but occurs as a gloss in a number of manuscripts (Godden and Irvine 2009, II, 276). Although the similarity in content between this gloss and the Old English text is suggestive, in the absence of any positive evidence for its use as a basis for the Old English translation the relevant Old English passage has been counted here as an interpolation for purposes of analysis. 4.3.5.4.2 The Pluperfect Due to the focus of Boethius upon general philosophical issues, verbs in primary tenses 6 predominate in this work, and accordingly verb forms of any sort falling within the semantic sphere of the pluperfect are rare. Nevertheless, most such verb forms are periphrastic constructions, as the following example illustrates: (161) Post haec paulisper obticuit atque after this-NEUT.ACC.PL shortly be.silent-3SG.PF yet.and ubi attentionem meam where attention-ACC my-FEM.ACC.SG modesta taciturnitate collegit sic modest-FEM.ABL.SG taciturnity-ABL collect-3SG.PF thus exorsa est: out.weave-PA.PPL-NOM.SG.FEM be-3SG.PRES (Boece IIp1.1) ‘Then after this she fell silent for a little while, and when she had 6 See Section 2.5.2.2 221 drawn my attention with her modest silence, she began to speak thus.’ Þa geswigode se wisdom then be.silent-3SG.PRET the-MASC.NOM.SG wisdom-NOM ane lytle hwile one-FEM.DAT.SG little-FEM.DAT.SG while-DAT.SG oððæt he ongeat þæs until he-NOM understand-3SG.PREF the-NEUT.GEN.SG modes ingeþances. Þa he hi þa mood-GEN in.thought-GEN when he-NOM they-ACC then ongiten hæfde, þa cwæð he. understand-PA.PPL have-3SG.PRET then speak-3SG.PRET he-NOM (Boece VII.251.2) ‘Then Wisdom was silent for a little while until he understood the Mind’s inward thought. When he had then understood them, then he spoke.’ One of the most common uses of the pluperfect in Boethius is in interpolated passages such as (161) and (207) below, which add explicit structure to the narrative by referring back to the preceding text. Some interpolations of this sort occur within introductory passages already present in the Latin original, such as (161); others, such as (207), stand on their own entirely. The latter example illustrates a formula often used by the Old English translator to mark the division between metrical and prose passages in the Latin original; other examples can be found at XXXIV.318.1, XXXIV.325.208, and XXXIX.358.15 inter alia. (162) Quoniam […] quae sit because which-FEM.NOM.SG be-3SG.PRES.SUBJ imperfecti, quae etiam imperfect-NEUT.GEN.SG which-FEM.NOM.SG also perfecti boni forma vidisti, nunc perfect-NEUT.GEN.SG good-GEN form-NOM see-2SG.PF now demonstrandum reor quonam demonstrate-GDV.NEUT.ACC.SG think-1SG.PRES-MP how.far haec felicitatis perfectio this-FEM.GEN.SG felicity-GEN perfection-NOM 222 constituta sit. consitute-PA.PPL-FEM.NOM.SG be-3SG.PRES.SUBJ (Boece IIIp10.1) ‘Because you have seen what form the imperfect good, as well as the perfect good, has, I think it is now to be demonstrated to what extent perfection is constituted of this felicity.’ Ic wene þæt hit sie nu I ween-1SG.PRES that it-NOM be-3SG.PRES.SUBJ now ærest þearf þæt ic þe gerecce first need that I-NOM thou-ACC tell-1SG.PRES.SUBJ hwær þæt hehste good where the-NEUT.NOM.SG highest-NEUT.NOM.SG good-NOM is, nu ic þe ær hæfde be-3SG.PRES now I-NOM thou-NOM ere have-3SG.PRET gereht hwæt hit wæs, oððe hwylc tell-PA.PPL what-NOM it-NOM be-3SG.PRET or which-NOM þæt medeme god wæs, the-NEUT.NOM.SG midmost-NEUT.NOM.SG good-NOM be-3SG.PRET hwylc þæt unmedeme. which-NOM the-NEUT.NOM.SG un.midmost-NEUT.NOM.SG (Boece XXXIV.318.2) ‘I think that there is now need first to tell you where the highest good is, now that I had previously told what it was, or which was the middle good and which was not the middle good.’ (163) Ut arbitror, haud multum tibi as arbitrate-1SG.PRES.MP not much thou-DAT haec in memoriam revocare laboraverim. this-NEUT.ACC.PL in memory-ACC recall-INF labour-1SG.PF.SUBJ (Boece IIp1.4) ‘As I judge, I would not be working hard to recall this to your memory.’ Ic wende þæt ic þe gio I-NOM ween-1SG.PRET that I-NOM thou-ACC formerly gelæred hæfde þæt þu hi teach-PA.PPL have-3SG.PRET that thou-NOM they-ACC oncnawan cuþest[…]. on.know-INF can-2SG.PRET (Boece XVII.252.14) ‘I thought that I had taught you long ago so that you could call them to mind.’ 223 The foregoing examples illustrate the relatively small number of pluperfects within the analysed sample which derive from a Latin original. Example (162), like (132) above, contains an Old English perfect in a context in which its Modern English equivalent would be less likely to be found, where the event in question does not seem to be marked as anterior to a point in the past, but seems rather to relate more closely to the present; the Latin original, which has been freely paraphrased in the process of translation, uses a perfect to refer to what is essentially the same event. Although the temporal reference is less clear, (208) below may fall into the same category. In the case of (163), Godden and Irvine (2009, II, 276) consider the pluperfect to result from the paraphrasing of a reading of the Latin passage in which the perfect subjunctive was misconstrued as a formally identical future perfect, with the less conditional meaning ‘I shall not have worked hard…’, from which an anterior event of teaching could be more readily inferred. (164) Ac ic wolde nu acsian hu þu and I-NOM will-3SG.PRET now ask-INF how thou-NOM ðis spell understanden hæfdest this-NEUT.ACC.SG spell-ACC understand-PA.PPL have-2SG.PRET (Boece XXXIV.322.118) ‘But I would now like to ask how you had understood this speech.’ (165) And he weld eallre gesceafta swa and he wield-3SG.PRES all-GEN.PL creature-GEN.PL so swa he æt frum getiohhad hæfde so he-NOM at first-DAT determine-PA.PPL have-3SG.PRET and get hæfð[…]. and yet have-3SG.PRES (Boece XXXIX.363.157) ‘And He governs all creatures just as He had determined at the beginning and has yet determined.’ 224 The two examples above illustrate the occurrence of the periphrastic pluperfect in entirely interpolated passages. Like (132), (162), and (208), (165) seems to be somewhat unusual in its temporal reference; the contrast between the two verbs seems not to be that between an event before a point in the past and an event before the present, but to be that between a past and a present state. In this respect it seems closer to the original stative meaning of the periphrastic constructions; however, the absence of any direct object seems to indicate a syntactic dissimilarity between this example and constructions in which have is merely a transitive lexical verb. Although these examples suggest that the semantic range of the periphrastic pluperfect may have been broader in some respects than its Modern English reflex, the evidence is sparse enough that it might be premature to offer any interpretation regarding the exact nature of this range. Within the sample of Boethius that has been analysed in the present study, only one preterite has been found that seems to fall within the semantic range of the pluperfect: (166) Þam wære mare þearf […] that-DAT.PL be-2SG.PRES.SUBJ more-NOM need-NOM þæt […] mon […] bæde þam mon that man-NOM bid-3SG.PRET.SUBJ that-DAT.PL man-NOM dyde swa micel wite swa hi þam do-3SG.PRET so much pain-ACC so they-NOM the-DAT.PL oðrum unscyldegum dydon other-DAT.PL innocent-DAT.PL do-3PL.PRET (Boece XXXVIII.357.232) ‘They would have more need that one should ask for them that as much harm should be done to them as they had done to other innocent people.’ This example occurs within a passage that bears so little resemblance to the Latin that it may be considered an interpolation. Considering the prolific use of 225 periphrastic construction in Boethius it is difficult to find any motivation for the use of the preterite in this instance; given the number of verbs in this passage, it might be thought that some explicit marking of the temporal relationships among the events denoted would have been even more desirable than in many contexts in which the periphrastic construction is used; conversely, it might be argued that the context is sufficiently informative that no such marking is needed for the avoidance of ambiguity. 4.3.5.5 Conclusion The preceding examples should serve to illustrate the range of contexts in which periphrastic constructions and semantically comparable preterites are found in the Dialogues, Bede, and Boethius. A comparison of these works shows that the differences of expression among them are not merely a statistical artifact; the variation among these works in their use of the periphrastic constructions is greater than can be explained by translation practices or by other factors, such as the syntactic environment, and that conversely, formally distinct expressions of similar semantic content can occur in apparently identical contexts. In such a case, it might seem that the variation described above in the use of the periphrastic constructions is not the result of a consistently operating factor or group of factors that produce a given output in a given environment; instead, a greater degree of freedom seems possible than such an explanation would predict. 226 4.4 Interpretation of the Data 4.4.1 The Proposed Hypothesis It may be useful to precede the theoretical proposals made in the present work with a recapitulation of the data that they are intended to address. An analysis was made of the distribution of the preterite and the periphrastic constructions as expressions of the perfect and pluperfect semantic domains, and it was found that in Old Saxon the data exhibit little variation and do not differ markedly from the state depicted in previous studies of the Old Saxon verbal system such as Arnett (1997) and Watts (2001). In Old English, however, there is considerable variation among contemporary texts in their use of the relevant grammatical forms; in the absence of any observable diachronic trend, the range of usage found among Old English texts of all periods may reasonably be considered to form part of a persistent pattern of synchronic variation. The differences among texts in their use of these forms cannot be attributed to variation in the grammatical status of the periphrastic constructions, nor to any other identifiable syntactic factors; although the influence of other languages upon Old English translations may be a factor in some cases, much of the observed variation cannot be ascribed to such a cause. Despite the absence of any obvious factor or set of factors motivating the observed variation, statistical analysis establishes that the differences among texts analysed in their use of the relevant grammatical forms are far greater than chance could be expected to produce. Accordingly, some explanation for this variability would seem desirable. It is proposed here that the preterite and the periphrastic perfect and pluperfect were available in Old English as means of expressing the semantic and 227 pragmatic content belonging to the perfect and pluperfect domains, but that these two formal categories differed in their perceived stylistic value and were used differentially on this basis. Such a stylistic value may have been related to register or to some other sociolinguistic variable; however, the ways in which these categories were perceived to differ by native speakers of Old English may no longer be recoverable. 4.4.2 Stylistic Variation in Old English The ascription of any phenomenon to stylistic or sociolinguistic variation within Old English involves certain difficulties, chief among which is the difficulty in evaluating Old English texts from this perspective. The most easily identifiable form of stylistic variation within Old English is that between poetry and prose. Old English poetry is recognizable not only by its metrical and alliterative structure but by a distinctive poetic vocabulary, which makes greater use of devices such as compounding and includes many distinctive lexical items, providing large numbers of synonyms for common lexical items such as man (see e.g. Godden 1992, 498). In addition to these lexical differences, the existence of certain syntactic differences between poetry and prose has been suggested, although the distinction between the two genres is not equally sharp in all cases; for example, poetic texts have been said to make more frequent use of SOV word order (see Mitchell 1985, II, 981–2) and to use determiners less than would be possible in prose (see Mitchell 1985, I, 135; Godden 1992, 504–6). Poetic texts have also been said to mix phonological and morphological features from different dialects more freely than is normally the case in prose (see Sisam 1953). As might be expected, the distinction between prose and poetry was not purely 228 binary, but because two distinct poles of usage are clearly recognizable it is possible to identify texts which combine stylistic features from both genres; one of the most notable examples of such stylistic hybridization can be found in those works of Ælfric’s in which he writes prose with some of the alliteration and metrical patterns characteristic of poetry, but far more loosely and with little use of poetic vocabulary (see e.g. Mitchell 1985, II, 995–9). The distinction between poetry and prose, however, is not a likely factor in the variation at issue here; both poetic and prose texts exist in which the periphrastic forms are used freely (see the figures for Genesis B above; further examples are cited by Mitchell 1985, I, 280– 99). Beyond the detection of elements associated with poetic style, however, it is difficult to attach specific meanings to the stylistic variation found among Old English prose texts; where such attempts have been made, interpretations often differ. For example, it has been suggested that Wulfstan’s style may have been more colloquial in some respects than that of contemporary authors such as Ælfric, on the basis of his frequent use of intensifiers and avoidance of markedly poetic vocabulary (Godden 1992, 532–3); however, the same elements of his style have elsewhere been attributed to the conscious application of a technique influenced by classical rhetoric (Bethurum 1957, 88–90). Similarly, the style of Boethius has been seen both as artificially Latinate (Potter 1939, 48–9) and as much freer and closer to the vernacular than many other Old English translations (Godden 1992, 525). Such judgements necessarily have a tendency towards the subjective, depending as they do on the perceived similarity or dissimilarity of an Old English text to other forms of speech. An additional difficulty in interpreting 229 variation among Old English texts relates to the homogeneity of the surviving prose, a homogeneity that has often been noted (e.g. Hogg 1988, 189); the prose texts of which we have knowledge are primarily works on theology, history, law, and other such learned subjects, composed in ecclesiastical and court settings, and as such they might be expected to have more similarities than differences when considered in light of the full range of linguistic variation that might be supposed to have existed. In spite of speculations as to the existence of distinct styles associated with the different topics treated (e.g. Sweet 1871, xl), it is seldom easy to differentiate usages representative of widely recognized styles or registers from those based on the personal preferences of individual authors (see further Godden 1992). Nevertheless, despite this textual homogeneity, variation among texts to the extent shown by the data presented here was evidently possible; given the existence of this variation, it seems plausible that the different formal means of expressing similar semantic content may not have been perceived in exactly the same way by native speakers. 4.4.3 Evidence for the Proposed Hypothesis The notion that the variation observed in the use of the periphrastic tenses may have been motivated by some stylistic or sociolinguistic differentiation among the relevant formal categories receives some support, if not absolute confirmation, from data in several areas. It was observed in Section 4.3.4 above that in the Pastoral Care the periphrastic forms seem to be preferred significantly in the main text, while the preterite is preferred in Biblical quotations and paraphrases; whatever the precise motivation for this distribution, such a pattern seems more compatible with the stylistic hypothesis proposed here than with an 230 explanation based on grammatical factors. Additional support is provided by data from languages other than Old English and Old Saxon (see the discussion of the latter in Section 4.2.2). Evidence from Middle High German suggests that the periphrastic perfect occurred more frequently in dialogue contexts than elsewhere; this distribution has been connected to the growing displacement of the preterite by the periphrastic form, a development that proceeded most quickly in colloquial registers (Zeman 2010). Such an association of the periphrastic forms with dialogue and informal-register contexts has also been found for Middle English, a period at which a more diverse range of texts are available than in Old English (Zimmermann 1968, 108–58; see also Fischer 1992, 256–8). Despite the existence of associations between periphrastic forms and such environments, it may be too simplistic to assume that the value of the periphrastic constructions in Old English was specifically colloquial, and that the works in which it was avoided would necessarily be seen as more formal than those in which it was used freely. Attempts to place specific sociolinguistic interpretations upon the variation found in Old English texts face a number of problems, as has been noted previously (e.g. Trousdale 2005). Any such attempt to view the differences described here in terms of register in a strictly sociolinguistic sense might lead one to conclude that the language of King Alfred, to whom the translation of the Pastoral Care is ascribed, had less prestige than that of Bishop Wærferth, who translated the Dialogues, or, if the reverse were the case, that the king and his court would take pains to disseminate a translation such as the latter despite its perceived lack of linguistic prestige (for a discussion of the societal setting in which such translations were produced, see e.g. Godden 2004). Moreover, the 231 fact that not one of the texts analysed in this study entirely avoids the use of the periphrastic constructions should be a further warning against the assumption of a style or register that was free from these forms. Nevertheless, despite the paucity of the evidence and the difficulty of reconstructing the exact value that these forms were perceived to have, the hypothesis that the variation observed in their use had some stylistic associations seems at present more compatible with the data than any known alternative. 4.4.4 Stylistic Variation and its Grammatical Prerequisites It should be noted that if the proposed hypothesis is correct in assuming that the preterite and the periphrastic forms were differentiated stylistically, the existence of such differentiation is dependent on the availability within the grammar of different formal means for expressing similar semantic content. Variability of this sort may not persist over time; it was noted in Section 3.3.2.1 that in Modern English the past tense and the present perfect have entered into a paradigmatic opposition, so that the use of either creates a presupposition that the other would be less appropriate pragmatically. The absence of such an opposition in Old English has long been noted (e.g. Hoffmann 1934; Mitchell 1985, I, 298); these modern observations are reinforced by what is virtually the only native- speaker evidence for the status of the periphrastic perfect tenses in Old English, the grammatical writings of Ælfric. As stated in Section 2.5.2.2, the Latin perfect had both a perfect sense and a perfective sense, so that a form such as steti could mean either ‘I have stood’ or ‘I stood’. Ælfric, who used this verb in his Latin grammar to illustrate the meaning of the Latin perfect tense, recognized the existence of this semantic duality and provided separate Old English translations 232 for both these senses, rendering the latter as ic stod ‘I stood’ and the former as ic stod fullice ‘I stood fully’; the corresponding Latin pluperfect steteram was translated by him as ic stod gefyrn ‘I stood long ago’ (ÆGr 123–4). His avoidance of the periphrastic constructions in this context, which has provoked considerable comment (see e.g. Hoffmann 1934; Mustanoja 1960, 498–9; Mitchell 1985, I, 295–6), would not have been possible in the presence of a paradigmatic opposition such as exists in Modern English; it is the absence of such an opposition that permits the variability seen in his work and in that of others. The absence of such a paradigmatic opposition is not restricted to Old English; it would appear from the data that Old Saxon, which uses preterites freely as an expression of perfect and pluperfect meaning, also lacked such an opposition. The primary difference between the two languages may therefore have been not in the grammatical status of the periphrastic constructions, nor in their paradigmatic relation to the rest of the verbal system, but in the extent to which the resulting possibility of variation was exploited for other purposes. It seems probable that the Old English constructions not only had the temporal and aspectual content shared with their Old Saxon equivalents but also an additional stylistic significance which differed from that of the cognate Old Saxon forms. Whatever the exact value placed on these constructions in Old English, it seems likely that it operated to retard the use of the new periphrastic forms available within the language, rather than to promote their use; the surviving Old Saxon texts make more consistent and proportionally greater use than their Old English 233 counterparts of the potential provided by these forms for the explicit marking of present relevance and anteriority. 234 5. Conclusions 5.1 Introduction The data provided by the current study present a picture noticeably different from that seen in much previous literature, as discussed in Chapter 1. It was said in Section 1.3.3 that the usual view of early Germanic languages assumes little synchronic variation in the use of the perfect and a steady, though not necessarily continuous, diachronic progress toward the states observed in the modern languages; in the case of Old English both these assumptions have been found to be incorrect. Moreover, the variation observed is of a nature some of whose aspects are previously unconsidered. The inclusion of data from Old English and Old Saxon allows cross-linguistic comparison that provides additional aid in the identification of relevant factors. These findings also suggest avenues for further research; in particular, the analysis of data from additional languages and time periods may be fruitful in shedding light on the phenomena described here. 5.2 Findings of the Present Study in Context One of the most substantial differences between the present study and previous work on the perfect in Old English and Old Saxon is methodological; rather than considering the periphrastic constructions in isolation, their distribution is compared with that of semantically comparable preterites. The discussion in Section 1.4 indicates how such a standard of comparison is essential to the obtaining of meaningful data regarding the distribution of the relevant 235 formal categories; without comparison of this sort, it is impossible to avoid conflating differences in the extent to which authors choose to express semantic content from the perfect and pluperfect domains with differences in the grammatical forms chosen as expressions of this meaning. In applying for the first time such a standard of comparison to the quantitative analysis of the selected textual data from Old English and Old Saxon, the present study not only reveals previously undescribed trends but can place prior findings on a firmer basis in those cases where the data analysed here concur with those from previous research. As discussed in Chapter 1, the existence of observable diachronic change in the use of the periphrastic constructions over the course of the Old English period has been proposed in much previous work, including that of Traugott (1972), Visser (1963–73), Denison (1993), and Carey (1994). The lack of any such trend in the data of the present study may therefore help to resolve an issue about which there has been a certain amount of variance. Although a negative finding of this sort is necessarily inconclusive, the use within the present study of a larger body of data, based on a broader corpus of texts, than has been employed in previous work may lend credence to the position taken here; the extent to which a limited corpus may have led to unwarranted conclusions in the case of Carey (1994) was noted in Section 1.3.3. From the data considered here, it seems safest to conclude that the constructions in question were largely stable diachronically over the period in question. In contrast to this diachronic stability, the present study reveals a far greater degree of synchronic variation in Old English than has generally been 236 acknowledged. The existence of variation among Old English texts in their choice of form to express perfect-type semantic content has been remarked casually (e.g. Zimmermann 1968, 155–7), but little material was available that would allow the extent of this variation to be assessed. Moreover, the possibility of such variation has often been overlooked altogether; recent quantitative studies such as Carey (1994) and Wischer (2002) amalgamate data from multiple texts and thereby obscure the effects of such textual variation. A greater awareness of the synchronic diversity established here can help to ensure that analyses of Old English data accurately reflect actual usage. The synchronic variation found in the present study is also unusual in certain respects. Other instances exist in which synchronic variation can be clearly related to syntactic developments in progress (e.g. van Kemenade 1997) or to the existence of largely discrete, independently attested sociolinguistic strata (e.g. Biber 1995). In the present instance the completion of the most substantial syntactic change relevant to the phenomena under study, that which led to the availability of the periphrastic constructions as a means of expressing perfect-type and pluperfect-type semantic content, is a prerequisite for the existence of such variation; as discussed in Section 4.4.2, the observed variation exists within a group of texts that would appear by most standards to be extremely homogeneous in its subject matter and origin. The findings of the present study demonstrate that existence of variation cannot necessarily be predicted with reference to a small set of well-understood factors. A further contribution of the present study is the provision of additional data regarding the semantics of the preterite in Old English. In previous work 237 there has been some disagreement regarding the extent to which the preterite could be used as a genuine expression of perfect meaning (e.g. Mitchell 1985, I, 246–8) as opposed to an unmarked past tense which was merely not incompatible with such an interpretation (e.g. Denison 1993, 352–4). However, according to the criteria discussed in Sections 2.5.2 and 3.3.2.1, the data presented here indicate the existence of certain identifiable semantic distinctions within the preterite as a formal category; perfect-like preterites differ from other preterites in their semantic compatibility with temporal adverbs and with the Latin tenses that they are used to translate, as well as in the sequences of tenses in which they occur. Additional support is accordingly provided for the concept of the preterite as a polysemous formal category. The cross-linguistic comparison permitted by the inclusion of data from different languages makes it possible to distinguish the effects of language- specific causes from those stemming from causes common to both languages. The preterite and the periphrastic constructions in Old English and Old Saxon would seem to be morphosyntactically and semantically comparable, as evaluated by the criteria discussed in Section 3.3; the fact that the periphrastic forms are used with more uniform freedom in Old Saxon than in Old English provides further support for the notion that the variation in the latter language is not the product of grammatical factors. This cross-linguistic perspective is especially valuable given the apparent similarity of the Old English and Old Saxon verbal systems; in both languages the periphrastic forms are relatively recent innovations competing with an older preterite, and the preterites and periphrastic perfect systems of the two languages are semantically similar. Without the opportunity 238 for such cross-linguistic comparison, it would be necessary to consider the possibility that the Old English distributional pattern is related to features of the verb system which are also present in Old Saxon. 5.3 Directions for Further Research One way in which the findings of the present study could be pursued further is through additional analysis of the Old English texts considered. Although every effort was made in the present study to identify all relevant variables related to perfect constructions and the morphosyntactic, semantic, and pragmatic contexts in which they occur, the possibility exists that other linguistic variables, not necessarily related to the perfect or even to the verbal system, may vary among texts in a manner correlated with the differences in the use of the periphrastic forms. If correlations were found between the forms preferred for the expression of perfect meaning and other, syntactically and semantically unrelated variables, this would provide further support for the hypothesis of a stylistic motivation for the observed variability. Such an investigation would necessarily be very broad in its scope; it would be difficult to ascertain a priori that a particular linguistic variable could not take part in hypothetical stylistic variation. Accordingly, the identification of appropriate variables would be an important preliminary to the analysis of their distributional patterns. Another avenue for further research consists in the addition of data from other languages. As discussed in Section 1.3, the morphosyntactic development of the English perfect has proceeded along broadly the same lines as the development of similar forms in other Germanic languages; the possibility should 239 be considered of whether any other Germanic languages have passed through a stage of synchronic variability similar to that identified here for Old English. One possible candidate for analysis is Old High German, in which similar periphrastic constructions are found (e.g. Schrodt 2004, 10–18), although there is a relative paucity of material for Old High German in comparison with that for Old English (see e.g. Ebert 1978, 58–9). Periphrastic constructions of similar form and meaning are also found in Old Norse (e.g. Rauch 1982; Faarlund 2004, 130–1). The analysis of data from languages such as these along similar lines to that performed in the present study would allow the comparison of other Germanic languages, insofar as possible, to the patterns described here for Old English and Old Saxon, and might shed further light on the variables involved in the development of these patterns. A promising approach for future research can also be found in the expansion of the data analysed to include material from other periods, providing material for further diachronic analysis. While such a diachronic expansion would be desirable for Old Saxon, the chronological and linguistic discontinuities between Old Saxon and later varieties such as Middle Low German would complicate any endeavour of this sort (see e.g. Rauch 1992, 104). A more fruitful possibility would be the analysis of perfect constructions in Middle English. Although no diachronic trends were identified in the Old English data analysed here, it is possible that the perfect did undergo diachronic changes in Old English which were not recorded; if significant divergences were visible even in the earliest Middle English texts this might be a sign of changes that had begun at an earlier period. There are a number of linguistic developments visible in Middle 240 English whose history in Old English is obscured to a certain extent by the standardized and dialectally homogeneous nature of the surviving Old English texts (e.g. Burrow and Turville-Petre 1985, 4), and it is not impossible that the periphrastic perfect and pluperfect were among the forms whose development was obscured in this way, although the possibility of such unrecorded changes does not invalidate any analysis of texts apparently uninvolved in any such developments. At present, there is little evidence regarding the point at which Middle English began to diverge from the Old English pattern described above. The most substantial data on the Middle English perfect forms continues to be that provided by Mustanoja (1960, 480–504) and Zimmermann (1968). Mustanoja presents evidence that in Early Middle English periphrastic tenses of any sort, not only the perfect and pluperfect but the simple future, are far less frequent numerically than the present and past, all periphrastic categories together making up less than 15% of verb forms in the texts studied, and considers the perfect and pluperfect in Middle English to be semantically modern; he also remarks on the Middle English occurrence of perfects with definite past-time modifiers, and sees the replacement by perfects of semantically comparable preterites in later manuscripts of certain texts as showing the establishment of a nascent paradigmatic opposition between the two categories. As discussed in Section 4.4.3, Zimmermann examines the environments in which the periphrastic perfect occurred and the semantic purposes for which it was used, concluding that the periphrastic tenses in Middle English are associated with factors such as a more colloquial style and looser syntactic structures; it has been suggested that the growing frequency of the 241 periphrastic forms over the course of the Middle English period might be an indicator not only of grammatical changes but of stylistic changes in progress (Fischer 1992, 256–8). Carey (1994) also includes Middle English data in her study; although no measures of statistical significance are given, her sample of periphrastic forms from Middle English includes verbs from a broader range of semantic categories than those found within her Old English sample. At present the previous work on this subject provides little quantitative information regarding the distribution of the periphrastic forms in Middle English, and as with Old English, no attempt has been made to compare the distribution of the periphrastic forms with that of semantically comparable preterites; as a result it is difficult to use existing Middle English data to interpret the findings presented here for Old English. It is hoped that future research on Middle English may illuminate further changes in the formal means of expressing semantic content from the perfect domain in English. 5.4 Conclusion At a more general level, the present study illustrates the extent to which generalizations based on insufficient data can result in a picture of a language very different from that which emerges upon closer consideration. If research upon a language is based on data that do not reflect actual usage, but rather a statistical composite reflecting the practices of no individual speaker, the conclusions reached by such means necessarily have a less stable foundation than might be wished. Issues of this sort are especially relevant to a topic such as the perfect; even in modern languages such as German, which have been intensively studied 242 and for which native-speaker data are readily available, little consensus exists about the precise semantics of the periphrastic constructions and their relationship to the simple preterite (see e.g. Duden 2005, 513–20; Schaden 2009). Especially in historical research, by taking as close an account as possible of the data available for a particular language, it is possible to gain an understanding of the language which is more complex, but also more accurate. 243 Appendix A This appendix contains additional textual examples similar to those presented above, within the main text; examples in this appendix are cited in the text at the relevant point. Comparison of these examples with those occurring elsewhere should make it possible to see the extent to which the constructions in question are representative of a recurring type rather than being isolated occurrences. (167) Hi sint […] to manienne ðæt hi […] they-NOM be-3PL.PRES to admonish-INF.DAT that they-NOM ða […] synna gelæden beforan hira the-ACC.PL sin-ACC.PL lead-3PL.PRES.SUBJ before they-GEN modes eagan, & ðonne hi hi mood-GEN eye-DAT.PL and when they-NOM they-ACC gesewene hæbben, gedon ðæt hie ne see-PA.PPL-ACC.PL have-3PL.PRES.SUBJ do-INF that they-NOM NEG ðyrfen bion gesewene æt need-3PL.PRES.SUBJ be-INF see-PA.PPL-ACC.PL at ðæm nearwan dome. the-MASC.DAT.SG narrow-DAT.SG doom-DAT (CP LIII.413.14) ‘They are to be admonished that they should lead the sins before their mind’s eye and, when they have seen them, to act so that they need not be seen at the strict Judgement.’ (168) Þa se cyng þas word when the-MASC.NOM.SG king-NOM this-NEUT.ACC.PL word-ACC hæfde gehered, he wæs swiðlice have-3PL.PRET hear-PA.PPL he-NOM be-3SG.PRET severely abreged[…]. alarm-PA.PPL (GD MS C II.XIV.133.2) ‘When the king had heard these words, he was extremely alarmed.’ (169) Nequaquam hunc fuisse nor.any-FEM.ABL.SG this-MAS.ACC.SG be-INF.PF 244 cuiusquam discipulum audivi […]. any-MASC.GEN.SG disciple-ACC hear-1SG.PF (GD I.I.6, p. 78) ‘In no way have I heard him to have been anyone’s disciple.’ Ne gehyrde ic næfre þæt he NEG hear-1SG.PRET I-NOM never that he-NOM æniges mannes leorningman wære any-MASC.GEN.SG man-GEN learning.man-NOM be-3SG.PRET-SUBJ (GD MS C I.I.12.23) ‘I have never heard that he was any man’s disciple.’ (Passage missing in MS H) (170) Usus […] est, ut praeesse non audeat use-NOM be-3SG.PRES that before.be-INF NEG dare-3SG.PRES.SUBJ qui subesse non didicit […]. REL-MASC.NOM.SG under.be-INF NEG learn-3SG.PF (GD I.I.6, p. 78) ‘It is the custom that he who has not learned to be under someone should not dare to be over anyone.’ Hit soðlice gewuna is[…], þæt se ne it-NOM truly custom is that that-MASC.NOM.SG NEG durre beon wisdomes lareow dare-3SG.PRES.SUBJ be-INF wisdom-GEN teacher-NOM oðres mannes, se þe hine other-MASC.GEN.SG man-GEN that-MASC.NOM.SG REL he-ACC ær him sylfum nan ne geleornað 1 erer he-DAT self-DAT none-ACC NEG learn-3SG.PRET (GD MS C I.I.12.27) ‘Truly, it is the custom that he dare not be a teacher of wisdom who himself has not previously learned from anyone.’ (Passage missing in MS H) (171) Quia secretum ratio aperuit, nihil because secret-NEUT.ACC.SG reason-NOM open-3SG.PF nothing mihi dubietate remansit I-DAT doubt-GEN remain-3SG.PF (GD I.VIII.7, p. 110) 1 The inclusion of this example is based on the interpretation of geleornað as a scribal error for the preterite geleornade; Hecht (1900, 12) cites the variant geleornode from MS O. 245 ‘Because reason has opened the secret, nothing of doubt has remained for me.’ Nis me naht æniges tweon nu to NEG.be-3SG.PRES I-DAT nought any-GEN doubt-GEN now to lafe be þissere foresprecenan wisan, remainder-DAT by this-DAT.PL aforesaid-FEM.DAT.PL wise-DAT.PL forþon þæt rihtgescead me ontynde þa for that discernment I-DAT open-3SG.PRET the-FEM.ACC.SG deogolnesse þurh þine gesægene. secrecy-ACC through thy-MASC.ACC.SC dictum-ACC (GD MS C I.XXI.55.33) ‘There is now nothing of any doubt remaining to me from these aforesaid things, for reason has opened the secrecy through your words.’ Nis me nan twynung nu to NEG.be-3SG.PRES I-DAT no-FEM.NOM.SG doubt-GEN now to laue be þam þingum, þe þu nu remainder-DAT by the-DAT.PL thing-DAT.PL RELthou-NOM now sædest, forþam þe þæt rihtgescead me say-3SG.PRET for REL that discernment-NOM I-DAT geopenode þa digolnysse þurh open-3SG.PRET the-FEM.ACC.SG secrecy-ACC through þine gesægene. thy-MASC.ACC.SC dictum-ACC (GD MS H I.XXI.55.33) ‘There is now no doubt remaining to me from the things that you have now said to me, for reason has opened the secrecy through your words.’ (172) Ecce posui verba mea in lo put-3SG.PF word-NOM.PL my-NEUT.NOM.PL in ore tuo. mouth-ABL thy-NEUT.ABL.SG (GD I.IV.8, p. 92) ‘Behold, I have put my words in your mouth.’ Geseoh nu, þæt ic sette min see-2SG.IMP now that I-NOM set-1SG.PRET my-NEUT.ACC.PL word in þinum muðe. word-ACC.PL in thy-MASC. DAT.SG mouth-DAT (GD MS C I.X.32.22) ‘See now that I have placed my words in your mouth.’ 246 Efne nu ic asette mine even now I-NOM set-1SG.PRET my-NEUT.ACC.PL word on þinum muðe word-ACC.PL on thy-MASC. DAT.SG mouth-DAT (GD MS H I.X.32.15) ‘I have just now placed my words in your mouth.’ (173) Os adhuc ad laudem Dei non aperui, mouth-ACC hereto to praise-ACC God-GEN NEG open-1SG.PF et ille cum simia veniens and that-MASC.NOM.SG with ape-ABL come-PRES.PPL cymbala percussit cymbal-ACC.PL strike-3SG.PF (GD I.IX.8, p. 114) ‘I have not yet opened my mouth in praise of God, and he, coming with an ape, struck cymbals.’ Nu gyt ic na minne muþ to Godes now yet I-NOM never my-MASC.ACC.SG mouth-ACC to God-GEN herenisse ne ontynde, & he veneration-DAT NEG open-3SG.PRET and he-NOM com mid apan & sloh cymbalan. come-3SG.PRET with ape-DAT and strike-3SG.PRET cymbal-ACC.PL (GD MS C I.XXIII.62.20) ‘Now I have not yet opened my mouth in praise of God, and he came with an ape and struck cymbals.’ Ic nu gyta ne geopenode minne muþ I-NOM now yet NEG open-3SG.PRET my-MASC.ACC.SG mouth-ACC to Godes lofe, & he com to God-GEN praise-DAT and he-NOM come-3SG.PRET mid apan & slyhð cimbalan. with ape-DAT and strike-3SG.PRES cymbal-ACC.PL (GD MS H I.XXIII.62.20) ‘Now I have not yet opened my mouth in praise of God, and he came with an ape and strikes cymbals.’ (174) Ad episcopum respiciens, dixit: “O quid to bishop-ACC respect-PRES.PPL say-3SG.PF O what-ACC fecisti? O quid fecisti?” Cui episcopus do-2SG.PF O what-ACC do-2SG.PF what-DAT bishop-NOM 247 respondit, dicens: “Quid feci?” respond-3SG.PF say-PRES.PPL what-ACC do-1SG.PF (GD I.X.18, p. 128) ‘Looking at the bishop, he said, “O, what have you done? O, what have you done?” To which the bishop responded, saying, “What have I done?”’ Locigende to þam biscope he þus look-PRES.PPL to the-MASC.DAT.SG bishop-DAT he thus cwæð: eala, hwæt dydest þu? speak-3SG.PRET O what-ACC do-2SG.PRET thou-NOM Eala, hwæt dydest þu? Þa þam O what-ACC do-2SG.PRET thou-NOM then that-NEUT.DAT.SG se biscop andswarode þus the-MASC.NOM.SG bishop-NOM answer-3SG.PRET thus cweðende: hwæt dyde ic? speak-PRES.PPL what do-1SG.PRET I-NOM (GD MS C I.XXXIII.85.6) ‘Looking at the bishop, he spoke thus: “O, what have you done? O, what have you done?” Then the bishop answered this, speaking thus: “What have I done?”’ Lokiende to þam bisceope he þus look-PRES.PPL to the-MASC.DAT.SG bishop-DAT he thus cwæð, eala, hwæt dydest þu? speak-3SG.PRET O what-ACC do-2SG.PRET thou-NOM Eala, hwæt dydest þu? Him O what-ACC do-2SG.PRET thou-NOM he-DAT se bisceop andswarode & the-MASC.NOM.SG bishop-NOM answer-3SG.PRET and cwæð, hwæt dyde ic? speak-3SG.PRET what do-1SG.PRET I-NOM (GD MS H I.XXXIII.85.10) ‘Looking at the bishop, he spoke thus: “O, what have you done? O, what have you done?” The bishop answered him, saying, “What have I done?”’ (175) Reducite eum, quia Fortunatus episcopus in back.lead-2PL.IMP he-ACC because Fortunatus-NOM bishop-NOM in domum illius venit. house-ACC that-MASC.GEN.SG come-3SG.PF (GD I.X.18, p. 130) ‘Take him back, because Bishop Fortunatus has come into his house.’ 248 Gelædað hine eft to þam lichaman, lead-2PL.IMP he-ACC back to the-MASC.DAT.SG body forþon þe Furtunatus se biscop for that Fortunatus-NOM the-MASC.NOM.SG bishop-NOM com on his hus & hine come-3SG.PRET on he-GEN house-ACC and he-ACC to him gecigde. to he-DAT call-3SG.PRET (GD MS C I.XXXIII.85.20) ‘Lead him back to the body, for Bishop Fortunatus has come into his house and called him to him.’ Gelædað hine eft to þam lichaman, lead-2PL.IMP he-ACC back to the-MASC.DAT.SG body forþon þe Furtunatus bisceop for that Fortunatus-NOM bishop-NOM becom in to his huse & hine come-3SG.PRET in to he-GEN house-DAT and he-ACC to him clypode. to he-DAT call-3SG.PRET (GD MS H I.XXXIII.85.19) ‘Lead him back to the body, for Bishop Fortunatus has come into his house and called him to him.’ (176) Ea quae mihi loqueris, that-NEUT.ACC.PL REL-NEUT.ACC.PL I-DAT speak-2SG.PRES.MP ego quoque mecum ipse pertracto. I-NOM also I-ABL.with self-MASC.NOM.SG feel-1SG.PRES (GD I.IV.8, p. 92) ‘That which you say to me, I myself also consider on my own.’ Þa wisan, þe þu nu to me the-FEM.NOM.PL wise-NOM.PL REL thou-NOM now to I-DAT sprecst, þa ic me sylf ær speak-2SG.PRES that- FEM.ACC.PL I-NOM I-DAT self ere swiþe georne gemunde. very eagerly remember-1SG.PRET (GD MS C I.X.32.15) ‘I have myself very clearly remembered those things that you are now saying to me.’ 249 Þa þing, þe þu nu to me the-NEUT.NOM.PL thing-NOM.PL REL thou-NOM now to I-DAT sprycst, þa ic ær on minum speak-2SG.PRES that-NEUT.ACC.PL I-NOM ere on my- MASC.DAT.SG mode swiðe georne me sylf smeade. mood-DAT very eagerly I-ACC self think-3SG.PRET (GD MS H I.X.32.14) ‘Those things that you are now saying to me I have quite often thought myself in my own mind.’ (177) Crebro ad me nunc usque veniens, dum frequently to I-ACC now even come-PRES.PPL while facta mihi veterum narrat, deed-ACC.PL I-DAT old-MASC.GEN.PL narrate-3SG.PRES nova refectione me satiat. new-FEM.ABL.SG refection-ABL me-ACC satiate-3SG.PRES (GD I.X.20, p. 130) ‘Coming to me frequently even now, as he tells me deeds of the elders, he satisfies me with new refreshment.’ Se com nu full oft gelomlice that-MASC.ACC.SG come-3SG.PRET now full often habitually to me, & þonne he sæde gehwæt be to I-DAT and when he-NOM say-3SG.PRET something by ealdra manna dædum, symble he me old-MASC.GEN.PL man-GEN.PL deed-DAT.PL always he-NOM I-ACC gereordode mid niwre gereordnysse. sustain-3SG.PRET with new-FEM.DAT.SG sustenance-DAT (GD MS C I.XXXIV.86.20) ‘He has now come to me customarily quite often, and when he has said something about old men’s deeds, he has always sustained me with new refreshment.’ Se nu gita gelomlice com to that-MASC.ACC.SG now yet habitually come-3SG.PRET to me, & þonne he me hwæt rehte be I-DAT and when he-NOM I-DAT something tell-3SG.PRET by ealdra manna dædum, simle he me old-MASC.GEN.PL man-GEN.PL deed-DAT.PL always he-NOM I-ACC 250 gefyllde mid niwre gereordnysse. fill-3SG.PRET with new-FEM.DAT.SG sustenance-DAT (GD MS H I.XXXIV.86.19) ‘He has yet now come to me customarily, and when he has told me something about old men’s deeds, he has always filled me with new refreshment.’ (178) Cum vir Dei, oratione facta, when man-NOM God-GEN prayer-ABL do-PA.PPL-FEM.ABL.SG eius oculis signum crucis inprimeret, […] he-GEN eye-DAT.PL sign-ACC cross-GEN impress-3SG.IMPF.SUBJ nox caecitatis abscessit. night-NOM blindness-GEN leave-3SG.PF (GD I.X.8, p. 124) ‘When the man of God, his prayers completed, made the sign of the cross upon his eyes, his night of blindness departed.’ Þa þa se Godes wer hæfde then when the-MASC.NOM.SG God-GEN man-NOM have-3SG.PRET gedon & gefylled his gebedu, he do-PA.PPL and fill-PA.PPL he-GEN bead-ACC.PL he-NOM asett & awrat Cristes rodtacen set-3SG.PRET and write-3SG.PRET Christ-GEN rood.token-ACC ofer þæ s mannes eagan […] & þa sona over the-MASC.GEN.SG man-GEN eye-ACC.PL and then soon seo niht þære blindnysse the-FEM.NOM.SG night-NOM the-FEM.GEN.SG blindness-GEN gewat[…]. leave-3SG.PRET (GD MS C I.XXXI.77.24) ‘Then when the man of God had performed and completed his prayers, he set and wrote the sign of the cross on the man’s eyes and then immediately the night of blindness departed.’ Þa þa se Drihtnes wer hæfde then when the-MASC.NOM.SG Lord-GEN man-NOM have-3SG.PRET his gebedu geendod, þa awrat he he-GEN bead-ACC.PL end-PA.PPL then write-3SG.PRET he-NOM Cristes rodtacen on þæs Christ-GEN rood.token-ACC on the-MASC.GEN.SG blindan mannes eagum & þær […] blind-MASC.GEN.SG man-GEN eye-DAT.PL and there 251 seo niht þære blindnysse the-FEM.NOM.SG night-NOM the-FEM.GEN.SG blindness-GEN gewat[…]. leave-3SG.PRET (GD MS H I.XXXI.77.24) ‘Then when the man of God had finished his prayers, he wrote the sign of the cross on the man’s eyes and then the night of blindness departed.’ (179) Cumque pro utilitate monasterii ad constitutionem when.and for utility-ABL monastery-GEN to constitution-ACC causae egressus fuisset[…]. matter-GEN exit-PA.PPL be-3SG.PLPF.SUBJ (GD I.II.11, p. 84) ‘And when he had gone out for the arrangement of business for the benefit of the monastery…..’ & þa þa he for þæs mynstres and then when he for the-NEUT.GEN.SG minster-GEN þearfe, swa swa he ær gecweden hæfde, need-DAT so so he-NOM ere bespeak-PA.PPL have-3SG.PRET wæs utfarende[…]. be-3SG.PRET out.fare-PA.PPL (GD MS C I.IV.22.11) ‘And then when he was going out for the needs of the monastery, just as he had said previously…..’ þa þa he for þæs mynstres then when he for the-NEUT.GEN.SG minster-GEN þearfe utferde, swa swa he ær need-DAT out.fare-3SG.PRET so so he-NOM ere gecweden hæfde […]. bespeak-PA.PPL have-3SG.PRET (GD MS H I.V.22.10) ‘Then when he went out for the needs of the monastery, just as he had said previously…..’ (180) Coepere […] in Honorato venerari abstinentiam, begin-3PL.PF in Honoratus-ABL venerate-INF.MP abstinence-ACC quam ante deridebat[…]. REL-FEM.ACC.SG before deride-3PL.IMPF (GD I.I.2, p. 76) ‘They began to venerate the abstinence of Honoratus, which they previously derided.’ 252 Hi ealle ongunnon wurðian & herigan they-NOM all-NOM.PL begin-3PL.PRET venerate-INF and honour-INF þa forhæfdnesse & þæt fæsten the-FEM.ACC.SG abstinence-ACC and the-NEUT.ACC.SG fast-ACC on Honorate, þe hi ær bysmerodon. on Honoratus-DAT REL they-NOM ere deride-3PL.PRET (GD MS C I.XV.44.23) ‘They began to venerate and honour the abstinence and fasting of Honoratus, which they previously derided.’ (Passage missing from MS H) (181) Iter quod coeperat peregit. journey-ACC REL.NEUT.ACC.SG begin-3SG.PLPF complete-3SG.PF (GD I.II.6, p. 82) ‘He continued on the journey that he had begun.’ He ferde him forð on his wege, he-NOM fare-3SG.PRET he-DAT forth on he-GEN way-DAT þe he ær ongan. REL he-NOM ere begin-3SG.PRET (GD MS C I.IV.18.29) ‘He fared forth on the journey that he had previously begun.’ He […] ferde he forð on his weg, he-NOM fare-3SG.PRET he-DAT forth on he-GEN way-DAT þe he ær begann. REL he-NOM ere begin-3SG.PRET (GD MS H I.IV.18.29) ‘He fared forth on the journey that he had previously begun.’ (182) Pro iniuria quam ingesserat for injury-ABL REL-FEM.ACC.SG inflict-3SG.PLPF recedere eum velle ex monasterio putabat[…]. recede-INF he-ACC want-INF out monastery-ABL think-3SG.IMPF (GD I.II.9, p. 84) ‘He believed him to want to leave the monastery because of the injury that he had inflicted.’ Þa wende he, þæt for ðan then ween-3SG.PRET he-NOM that for the-MASC.DAT.SG teonan, þe he him þy harm-DAT REL he-NOM he-DAT the-MASC.INST.SG ærran dæge gedyde mid ealle earlier-MASC.INST.SG day-INST do-3SG.PRET with all-NEUT.DAT.SG 253 of þam mynstre gewitan wolde. of the-NEUT.DAT.SG minster-DAT depart-INF will-3SG.PRET (GD MS C I.V.21.10) ‘Then he thought that because of the injury that he had done him the previous day, he wanted to leave with all the monastery.’ Se abbud […] wende þa, þæt he the-MASC.NOM.SG abbot-NOM ween-3SG.PRET then that he-NOM mid ealle of þam mynstre with all-NEUT.DAT.SG of the-NEUT.DAT.SG minster-DAT gewitan wolde for þam teonan, depart-INF will-3SG.PRET for the-MASC.DAT.SG harm-DAT þe he him ær gedyde. REL he-NOM he-DAT ere do-3SG.PRET (GD MS H I.V.21.7) ‘The abbot then thought that he wanted to leave with all the monastery because of the injury that he had previously done him.’ (183) Se reum esse testatus self-ACC guilty-MASC.ACC.SG be-INF witness-PA.PPL-MASC.NOM.SG est, qui […] facere tam crudelem be-3SG.PRES REL-MASC.NOM.SG do-INF so cruel-FEM.ACC.SG contumeliam praesumpsisset. contumely-ACC presume-3SG.PLPF.SUBJ (GD I.II.10, p. 84) ‘He bore witness that he was guilty who had presumed to inflict such cruel injuries.’ He hine cyðde, þæt he scyldig he-NOM he-DAT make.known-3SG.PRET that he-NOM guilty-NOM wære, forþon þe he geþrystlæhte, þæt be-3SG.PRET.SUBJ for REL he-NOM presume-3SG.PRET that he […] swylce wælhreownesse fraceþa he-NOM such-FEM.ACC.SG bloodthirstiness-ACC outrage-GEN.PL gefremede. frame-3SG.PRET (GD MS C I.V.21.25) ‘He made it known to him that he was guilty, because he had presumed to inflict such outrageous bloodthirstiness.’ He […] cwæð, þæt he wið hine he-NOM speak-3SG.PRET that he-NOM with he-ACC agylt hæfde & wið hine scyldi guilt-ACC have-3SG.PRET.SUBJ and with he-ACC guilty-NOM 254 wære, forþam þe he geþristlæhte, þæt be-3SG.PRET.SUBJ for REL he-NOM presume-3SG.PRET that he […] swa wælhreowne teonan gedyde he-NOM so bloodthirsty-MASC.ACC.SG harm-ACC do-3SG.PRET (GD MS H I.V.21.21) ‘He said that he had guilt towards him and was culpable towards him, because he had presumed to do such bloodthirsty injury.’ (184) Libertinus […] suae culpae[…] fuisse Libertinus-NOM own-FEM.DAT.SG fault-DAT be-INF.PF referebat quod pertulerat. refer-3SG.IMPF what-ACC undergo-3SG.PLPF (GD I.II.10, p. 84) ‘Libertinus was ascribing what he had undergone to his own fault.’ Libertinus […] sæde, þæt hit his sylfes Libertinus-NOM say-3SG.PRET that it-NOM he-GEN self-GEN gylt wære[…], þæt broc, guilt-NOM be-3SG.PRET.SUBJ the-NEUT.NOM.SG injury-NOM þæt he þær aræfnode. REL-NEUT.ACC.SG he-NOM there endure-3SG.PRET (GD MS C I.V.21.34) ‘Libertinus said that the injury that he had endured there was his own fault.’ Libertinus […] cwæð, þæt hit wære Libertinus-NOM say-3SG.PRET that it-NOM be-3SG.PRET.SUBJ for his agene gylt þæt for he-GEN own-MASC.DAT.SG guilt-DAT the-NEUT.NOM.SG broc, þæt he þolode[…]. injury-NOM REL-NEUT.ACC.SG he-NOM endure-3SG.PRET (GD MS H I.V.21.32) ‘Libertinus said that it was for his own guilt, the injury that he had endured.’ (185) Accedere quispiam monachorum in approach-INF any-MASC.NOM.SG monk-GEN.PL in congregationem virginum minime audebat; congregation-ACC virgin-GEN.PL least dare-3SG.IMPF quanto minus ille how.much-NEUT.ABL.SG less that-MASC.NOM.SG 255 qui novus advenerat[…]. REL-MASC.NOM.SG new-MASC.NOM.SG to.come-3SG.PLPF (GD I.IV.4, p. 90) ‘Each of the monks scarcely dared to approach the women’s quarters, least of all for him who had newly arrived to do so.’ Ac þa ne dorste nan þæra muneca […] But then NEG dare-3SG.PRET none-NOM the-GEN.PL monk-GEN.PL gan in þa gesamnunge þara go-INF in the-FEM.ACC.SG gathering-ACC the-GEN.PL fæmnena, & mycle læs þam wæs maiden-GEN.PL and much less that-MASC.DAT.SG be-3SG.PRET alyfed, þam þe niwan com[…]. grant-PA.PPL that-MASC.DAT.SG REL newly come-3SG.PRET (GD MS C I.VIII.28.34) ‘But then none of the monks dared go into the women’s quarters, and this was permitted much less to him who had newly come.’ Þa ne dorste nan þæra muneca […] then NEG dare-3SG.PRET none-NOM the-GEN.PL monk-GEN.PL gangan inn to þa fæmnena gesomnunge go-INF in to the-GEN-PL maiden-GEN.PL gathering-ACC & micele læs se þe niwan com[…]. and much less that-MASC.NOM.SG REL newly come-3SG.PRET (GD MS H I.VIII.28.33) ‘Then none of the monks dared go into the women’s quarters, much less for him who had newly come to do so.’ (186) Eum […] restituit saluti, ut reveniens he-ACC restitute-3SG.PF health-DAT that return-PRES.PPL pater ea hora filum father-NOM that-FEM.ABL.SG hour-ABL son-ACC restitutum vitae cognosceret, restitute-PA.PPL-MASC.ACC.SG life-DAT discover-3SG.IMPF.SUBJ qua vitam illius ex ore REL-FEM.ABL.SG life-ACC that-MASC.GEN.SG out mouth-ABL veritatis audisset. truth-GEN hear-3SG.PLPL.SUBJ (GD I.IV.6, p. 86) ‘He restored him to health so that the father, returning, might know his son to have been restored to life at the hour at which he had heard life from the mouth of Truth.’ 256 He […] hine gehælde, & sona swa he-NOM him-ACC heal-3SG.PRET and soon so þæ cnihtes fæder him fram the-MASC.GEN.SG knight-GEN father he-DAT from cyrde, on þa ylcan tide turn-3SG.PRET on the-FEM.ACC.SG same-FEM.ACC.SG tide-ACC he oncneow, þæt him wæs eft he-NOM know-3SG.PRET that he-DAT be-3SG.PRET back lif seald, þe he ær gehyrde of life-ACC sell-PA.PPL REL he-NOM ere hear-PA.PPL of þæs hælendes sylfes muðe, þæt the-MASC.GEN.SG healing-GEN.SG self-GEN mouth-DAT that him lif gehaten wæs he-DAT life-NOM promise-PA.PPL be-3SG.PRET (GD MS C I.VIII.30.7) ‘He healed him, and as soon as the boy’s father turned from Him, he knew that life was given back to him at the same time that he had previously heard from the Saviour’s own mouth that life was promised him.’ He […] hine gehælde, swa þæt þæs he-NOM him-ACC heal-3SG.PRET so that the-MASC.GEN.SG cnihtes fæder ham cyrrende, oncneow þæt knight-GEN father home turn-PRES.PPL know-3SG.PRET that him wæs lif forgifen on þære he-DAT be-3SG.PRET life-ACC forgive-PA.PPL on the-FEM.DAT.SG ilcan tide þe he ær gehyrde, same-FEM.DAT.SG tide-DAT REL he-NOM ere hear-PA.PPL þæt him lif behaten wæs of that he-DAT life-NOM promise-PA.PPL be-3SG.PRET of þæs hælendes sylfes muðe. the-MASC.GEN.SG healing-GEN.SG self-GEN mouth-DAT (GD MS H I.VIII.30.5) ‘He healed him so that the boy’s father, returning home, knew that life was given back to him at the same time that he had previously heard that life was promised him from the Saviour’s own mouth.’ (187) Eiusdem Iulianui animum same-MASC.GEN.SG Julianus-GEN spirit-ACC intolerabilis pavor invasit, ita ut[…] intolerable-MASC.NOM.SG fear-NOM invade-3SG.PF so that 257 ad insinuandum hoc to insinuate-GDV-NEUT.ACC.SG this-NEUT.ACC.SG ipsum quod venerat vix self-NEUT.ACC.SG REL-NEUT.ACC.SG come-3SG.PLPF scarcely sufficere lingua potuisset suffice-INF tongue-NOM be.able-PLPF.SUBJ (GD I.IV.14, p. 94) ‘An intolerable fear assailed the spirit of this same Julianus, so that his tongue could barely manage to suggest that for which he had come.’ On þæs ylcan Iulianus mod on the-MASC.GEN.SG same-MASC.GEN.SG Julianus-GEN mood-ACC gefor unaræfendlicu fyrhtu, swa þæt[…] fare-3SG.PRET intolerable-FEM.NOM.SG fright-NOM so that uneaðe him mihte his tunge genihtsumian uneasily he-DAT may-3SG.PRET he-GEN tongue-NOM suffice-INF to þon þæt he sæde his to that-NEUT.INST.SG that he say-3SG.PRET he-GEN ærende, þe he þider fore com. errand-ACC REL he-NOM thither for come-3SG.PRET (GD MS C I.XII.37.22) ‘Into the mind of this same Julianus came an intolerable fright, so that his tongue could scarcely suffice for this, that he might tell his errand, for which he had come there’ Þa gefor on þæs ylcan then fare-3SG.PRET on the-MASC.GEN.SG same-MASC.GEN.SG Iulianes mod unacumendlic forhtnys, swa Julianus-GEN mood-ACC intolerable-FEM.NOM.SG fright-NOM so þæt[…] he earfoðlice hæfde his tungan that he-NOM laboriously have-3SG.PRET he-GEN tongue-GEN geweald, þæt he mihte abeodan power-ACC that he may-3SG.PRET announce-INF þæt ærende, þe he þyder fore the-NEUT.NOM.SG errand-ACC REL he-NOM thither for com. come-3SG.PRET (GD MS H I.XII.37.23) ‘Then came an intolerable fright into the mind of this same Julianus, so that only with difficulty did he get control of his tongue, that he might announce the errand for which he had come there.’ 258 (188) Libertinum existimo ista Libertinus-ACC consider-1SG.PRES that-NEUT.ACC.PL potuisse, quia […] didicerat de magistri […] be.able-INF.PF because learn-3SG.PLPF of master-GEN virtute confidere. virtue-ABL trust-INF (GD I.II.7, p. 82) ‘I consider Libertinus to have been able to do these things because he had learned to trust in his master’s virtue.’ Ic wene, þæt Libertinus mihte I-NOM ween-1SG.PRES that Libertinus may-3SG.PRET ðis gedon forþon he geleornode, þæt this-NEUT.ACC.SG do-INF for he-NOM learn-3SG.PRET that he getreowde […] be his lareowes magne[…]. he-NOM trust-3SG.PRET.SUBJ by he-NOM teacher-GEN main-DAT (GD MS C I.IV.19.7) ‘I think that Libertinus could do this because he had learned that he should trust in his teacher’s virtue.’ Ic wene, þæt Libertinus mihte I-NOM ween-1SG.PRES that Libertinus may-3SG.PRET þis gedon forþam þe he getruwode this-NEUT.ACC.SG do-INF for REL he-NOM learn-3SG.PRET be his lareowes mægene[…]. by he-GEN teacher-GEN main-DAT (GD MS H I.IV.19.6) ‘I think that Libertinus could do this because he trusted in his teacher’s virtue.’ (189) Putamus hic tam egregius think-1PL.PRES this-MASC.NOM.SG so egregious-MASC.NOM.SG vir, ut post magister discipulorum man-NOM that after master-NOM disciple-GEN.PL fieret, prius habuit magistrum? become-3SG.IMPF.SUBJ earlier have-3SG.PF master-ACC (GD I.I.5, p. 78) ‘Do we think that such an outstanding man as this first had a master, that he should have become a master of disciples?’ Wenað we, hwæþer þes ween-3PL.PRES we-NOM whether this-MASC.NOM.SG æðele wer ær ænigne lareow noble-MASC.NOM.SG man-NOM ere any-MASC.ACC.SG teacher-ACC 259 hæfde, se þe æfter þan have-3SG.PRET that-MASC.NOM.SG REL after that-NEUT.DAT.SG þus manigra manna lareow gewearð? thus many-GEN.PL man-GEN.PL teacher-NOM become-3SG.PRET (GD MS C I.I.12.21) ‘Do we think that this noble man (had) previously had any teacher, he who afterwards thus became a teacher of many men?’ (Passage missing from MS H) (190) Non tamen hoc facere possum[…], NEG yet this-NEUT.ACC.SG do-INF be.able-1SG.PRES cum ille mihi nil mali fecerit, when that-MASC.NOM.SG me-DAT nothing ill-GEN do-3SG.PF.SUBJ nil adhuc inimicitiarum intulerit. nothing hereto enmity-GEN.PL in.bear-3SG.PF.SUBJ (Bede II.12, I, p. 272) ‘Yet I cannot do this when he has done me no wrong, shown me no enmity hereto.’ Hwæðre ne mæg ic þæt don […] whether NEG may-1SG.PRES I-NOM that-NEUT.ACC.SG do-INF mid þy he me noht with that-NEUT.INST.SG he-NOM me-DAT nought-ACC yfeles dyde ne laðes æteawde. evil-GEN do-3SG.PRET nor loathing-GEN show-3SG.PRET (Bede II.9.128.3) ‘Yet I cannot do that given that he has done me no evil nor shown me any hatred.’ (191) ista omnia quae vidisti that-NEUT.NOM.PL all-NEUT.NOM.PL REL-NEUT.ACC.PL see.2SG.PF (Bede V.12, II, p. 262) ‘all these things that you have seen’ ðas þing ealle […] ðe þu this-NEUT.NOM.PL thing.PL all-NEUT.NOM.PL REL thou.NOM sceawadest & gesawa show-2SG.PRET and see-2SG.PRET (Bede V.13.430.29) ‘all these things that you have observed and seen’ (192) Frigora ego vidi. colder-NEUT.ACC.PL I-NOM see-1SG.PF (Bede V.12, II, p. 268) ‘I have seen colder things.’ 260 Caldran ic geseah. colder-ACC.PL I-NOM see-1SG.PRET (Bede V.13.436.12) ‘I have seen colder.’ (193) Universos quos in necem universal-MASC.ACC.PL REL-MASC.ACC.PL in murder-ACC suam conspirasse didicerat, aut occidit, own-FEM.ACC.SG conspire-INF.PF learn-3SG.PLPF or slay-3SG.PF aut in deditionem recepit. or in surrender-ACC receive-3SG.PF (Bede II.9, I, p. 252) ‘All those whom he had discovered to have conspired to his murder he either slew or took prisoner.’ Wæron him ealle his fynd be-3SG.PRET he-DAT all-NOM.PL he-GEN fiend-NOM.PL gecyðede, þa þa ær ymb his identify-PA.PPL-ACC.PL that-NOM.PL REL ere about he-GEN feorh syredon. & he þa sume life-ACC plot-3PL.PRET and he-NOM then some-ACC ofslog, sume on onweald onfeng[…]. slay-3SG.PRET some-ACC on power-ACC on.take-3SG.PRET (Bede II.8.124.9) ‘There were made known to him all his enemies, those who had previously plotted against his life, and he then slew some, took some prisoner.’ (194) Suidberct accepto episcopatu, de Swidbert-NOM accept-PA.PPL-MASC.ABL.SG episcopate-ABL of Brittania regressus, non multo post ad Britain-ABL return-PA.PPL-MASC.NOM.SG not much after to gentem Boructuarorum secessit[…]. people-ACC Boructuar-GEN.PL depart-3SG.PF (Bede V.9, II, p. 248) ‘The episcopate having been received, Swidbert, having returned from Britain, not long afterward departed to the people of the Boructuars.’ Ða Swiðberht hefde bisscophade onfongen, when Swidbert-NOM have-3SG.PRET episcopate-ACC receive-PA.PPL þa ferde he eft of Breotene & efter then fare-3SG.PRET he-NOM back of Britain-DAT and after 261 medmicelum fece he gewat mid.much-NEUT.DAT.SG interval-DAT he-NOM depart-3SG.PRET to ðere þeode Boruchtuarorum[…]. to the-FEM.DAT.SG people-DAT Boructuar-GEN.PL (Bede V.12.420.15) ‘When Swidbert had received the episcopate, then he travelled back from Britain and after a brief interval he departed to the people of the Boructuars.’ (195) Pro parvulis Christi, quos mihi for little-MASC.ABL.PL Christ-GEN REL-MASC.ACC.PL I-DAT in indicium suae dilectionis in token-ACC own-FEM.GEN.SG delight-GEN commendaverat vincula[…] pertuli commend-3SG.PLPF bond-ACC.PL suffer-1SG.PF (Bede II.6, I, p. 232) ‘For Christ’s little ones, which He had commended to me as a token of His love, I suffered bonds.’ Ic fore Cristes cneohtum, þa he me I-NOM for Christ-GEN knight-DAT.PL REL-ACC.PL he-NOM I-DAT in tacnunge his lufan bebead, in tokening-DAT he-GEN love-GEN entrust-3SG.PRET bende […] þrowade[…]. bond-ACC.PL suffer-3SG.PRET (Bede II.6.114.21) ‘I suffered bonds for Christ’s children, whom he entrusted to me as a token of his love.’ (196) Memoriam maeror hebetavit. memory-ACC sorrow-NOM blunt-3SG.PF (Boece Ip6.10) ‘Sorrow has blunted my memory.’ Me hæfð þeos gnornung me-ACC have-3SG.PRET this-FEM.NOM.SG sorrow-NOM þære gemynde benumen. the-FEM.GEN.SG mind-GEN rob-PA.PPL (Boece V.249.64) ‘This sorrow has deprived me of memory.’ (197) Deprehendisti caeci numinis seize-2SG.PF blind-NEUT.GEN.SG will-GEN 262 ambiguos vultus ambiguous-MASC.ACC.PL countenance-ACC.PL (Boece IIp1.11) ‘You have grasped the ambiguous face of blind desire.’ Nu þu hæfst ongyten Now thou-NOM have-2SG.PRES understand-PA.PPL þa wonclan treowa the-FEM.ACC.PL unstable-FEM.ACC.PL truce-ACC.PL þæs blindan lustes the-MASC.GEN.SG blind-MASC.GEN.SG lust-GEN (Boece VII.252.29) ‘Now you have understood the unstable faith of blind pleasure.’ (198) Sed hoc est quod but this-NEUT.NOM.SG be-3SG.PRES what-NOM recolentem vehementius contemplate-PRES.PPL-MASC.ACC.SG vehemently.COMP coquit[…]. cook-3SG.PRES (Boece IIp4.2) ‘But this is what disturbs me more vehemently in reflecting.’ Ac þæt me hæfð eallra But that-NEUT.NOM.SG I-ACC have-3SG.PRES all-GEN.PL swiðost gedrefed[…]. strongest disturbed-PA.PPL (Boece X.258.5) ‘But that has disturbed me most severely of all.’ (199) Nunc stuporem meum deus rector now stupor-ACC my-MASC.ACC.SG God-ACC rector-ACC exaggerat. exaggerate-3SG.PRES (Boece IVp5.5) ‘Now God, the Governor, increases my astonishment.’ Ac se ælmihtiga God but the-MASC.NOM.SG almighty-MASC.NOM.SG God-NOM hæfð geeced minne ege[…]. have-3SG.PRES eke-PA.PPL my-MASC.ACC.SG awe-ACC (Boece XXXIX.359.36) ‘But God Almighty has increased my awe.’ 263 (200) Iam scio inquit morbi tui already know-1SG.PRES say-3SG illness-GEN thy-MASC.GEN.SG aliam vel maximam causam[…]. other-FEM.ACC.SG or greatest-FEM.ACC.SG cause-ACC (Boece Ip6.17) ‘“Now I know,” she said, “another or the greatest cause of your illness.”’ Þa cwæð se wisdom. then speak-3SG.PRET the-NOM.MASC.SG wisdom-NOM nu Ic hæbbe ongiten now I-NOM have-1SG.PRES understand-PA.PPL þine ormodnesse[…]. thy-FEM.ACC.SG despair-ACC (Boece V.249.76) ‘Then Wisdom said, “Now I have understood your despair.”’ (201) Cuncta […] firmissimis nexa whole-NEUT.NOM.PL firmest-ABL.PL bind-PA.PPL-NEUT.NOM.PL rationibus constant. reason-ABL.PL stand -3PL.PRES (Boece IIIp11.1) ‘Everything stands bound with the strongest reasoning.’ Ðu hit hæfst geseðed mid thou-NOM it-ACC have-2SG.PRES prove-PA.PPL with gesceadwislicre race. rational-FEM.DAT.SG argument-DAT (Boece XXXIV.323.150) ‘You have proved it with rational argument.’ (202) Similiter ratiocinari de honoribus, gloria, similarly ratiocinate-INF.MP of honour-ABL.PL glory-ABL voluptatibus licet pleasure-ABL.PL be.permitted-3SG.PRES (Boece IIIp9.21) ‘It is permissible to deliberate similarly upon honours, glory, and pleasures.’ Nu hæbbe we gereht by welan now have-1SG.PRET we-NOM tell-PA.PPL by wealth-DAT and be anwealde, and þæt ilce and by power-DAT and the-NEUT.ACC.SG same-ACC 264 we magon reccan be þam þrim we-NOM may-3PL.PRES tell-INF by the-DAT.PL three-DAT þe we unareht habbað, þæt REL we-NOM untold have-3PL.PRES that-NEUT.NOM.SG is weorðscipe and formærnes and be-3SG.PRES worship-NOM and fame-NOM and willa. desire-NOM (Boece XXXIII.313.89) ‘Now we have spoken of wealth and power, and we may say the same about the three of which we have not spoken; that is, honour and fame and desire.’ (203) Þu cwist þæt we habban þe thou-NOM say-2SG.PRES that we have-3PL.PRES thou-ACC beswicenne, ac we magan cweþan betray-PA.PPL-MASC.ACC.SG but we-NOM may-3PL.PRES say-INF ma þæt þu hæbbe us more that thou-NOM have-2SG.PRES.SUBJ we-ACC beswicene[…]. betray-PA.PPL-ACC.PL (Boece VII.256.135) ‘You say that we have betrayed you, but we may say rather that you have betrayed us.’ (204) Quid est igitur, o homo, quod what-NOM be-3SG.PRES therefore O man-VOC REL-NEUT.NOM.SG te in maestitiam luctumque deiecit? thou-ACC in sorrow-ACC lamentation-ACC.and down.cast-3SG.PF (Boece IIp1.9) ‘What then is it, O man, that has cast you into sorrow and lamentation?’ Eala mod, hwæt bewearp þe on O mood-NOM what-NOM cast-3SG.PRET thou-ACC on þas care and on þas this-FEM.ACC.SG care-ACC and on this-FEM.ACC.SG gnornunga? lamentation-ACC (Boece VII.249.64) ‘O Mind, what has cast you into this sorrow and lamentation?’ (205) Non habes ius querelae tamquam NEG have-2SG.PRES right-ACC quarrel-GEN just.as 265 prorsus tua perdideris. forwards thy-NEUT.ACC.PL lose-2SG.PF.SUBJ (Boece IIp2.5) ‘You have no right to complain as if you had actually lost your own things.’ Ne miht ðu no gereccean NEG may-2SG.PRES thou-NOM never tell-INF þæt þu þines auht forlure. that thou-NOM thy-NEUT.GEN.SG aught lose-2SG.PRET.SUBJ (Boece VII.254.75) ‘You cannot claim that you have lost anything of your own.’ (206) Neque […] sapientum quisquam exsul […] nor discerning-GEN.PL any-NOM.MASC.SG exile-NOM esse malit[…]. be-INF prefer-3SG.PRES.SUBJ (Boece IVp5.2) ‘Nor would any of the wise prefer to be an exile.’ Ic næfre ne geseah ne gehyrde I-NOM never NEG see-3SG.PRET nor hear-3SG.PRET nænna wisne mon ðe no-MASC.ACC.SG wise-MASC.ACC.SG man-ACC REL more wolde bion wrecca[…]. will-3SG.PRET.SUBJ be-INF wretch-NOM (Boece XXXIX.358.20) ‘I have never seen or heard of any wise man that would rather be an exile.’ (207) Þa se wisdom þa þis when the-MASC.NOM.SG wisdom-NOM then this-NEUT.ACC.SG leoð asungen hæfde, þa ongan song sing-PA.PPL have-3SG.PRET then begin-3SG.PRET he eft spellian and þus cwæð. he-NOM again recount-INF and thus speak-3SG.PRET (Boece XXXIII.310.1) ‘When Wisdom had sung this song, then he began to talk again and spoke thus.’ (208) Quoniam tu idem es cui because thou-NOM same-NOM be-2SG.PRES REL-MASC.DAT.SG persuasum […] permultis persuade-PA.PPL-NEUT.NOM.SG through.many-FEM.ABL.PL 266 demonstrationibus scio mentes demonstration-ABL.PL know-1SG.PRES mind-ACC.PL hominum nullo modo esse man-GEN.PL no-MASC.ABL.SG manner-ABL be-INF mortales[…] mortal-FEM.ACC.PL (Boece IIp4.28) ‘Because you are the same one that was persuaded by very many demonstrations, I know the minds of men to be in no way mortal.’ Ic wene nu þæt ic hæfde ær I-NOM ween-1SG.PRES now that I-NOM have-1SG.PRET ere genog sweotole gereht be manegum tacnum enough clearly tell-PA.PPL by many-DAT.PL token-DAT.PL þætte monna sawula sint undeaþlice and that man-GEN.PL soul-NOM.PL be-3PL.PRES immortal and ece[…]. eternal (Boece XI.263.82) ‘I think that I had previously said clearly enough, through many tokens, that men’s souls are immortal and eternal.’ 267 Appendix B It has been found desirable to make the data on which the above analysis is based fully available. This can be done most conveniently in electronic form. The attached CD-ROM contains the data in the following file formats:  Plain text (comma-separated value format, Unicode little-endian encoding) o OEP.csv (Old English perfect data) o OEPP.csv (Old English pluperfect data) o OSP.csv (Old Saxon perfect data) o OSPP.csv (Old Saxon pluperfect data)  IBM® SPSS® format o OEP.sav o OEPP.sav o OSP.sav o OSPP.sav  Microsoft® Excel® 2007 format o Data.xlsx (contains four worksheets: OEP, OEPP, OSP, and OSPP)  OpenDocument spreadsheet format o Data.ods (contains four worksheets: OEP, OEPP, OSP, and OSPP) In each case, the files contain the following data fields:  Verb Form This field includes the relevant verb form. Tokens representing 268 periphrastic constructions include both the auxiliary and the participle, in the original order, separated by a space; any intervening words are omitted.  Text This field includes the text from which the reference is drawn, using abbreviations of the same form as elsewhere in this work.  Reference This field contains a reference to the location of the verb form within its text, in the format used in the Helsinki Corpus and subject to the restrictions discussed in Section 3.2.5.  Periphrastic This field contains either 1, for a periphrastic construction, or 0, for a preterite.  Person This field contains the values 1, 2, or 3, corresponding to the person of the verb. The statistical analysis presented in Chapter 4 makes no reference to this field, which is included here for informational purposes.  Discourse This field contains the discourse context, as described in Section 3.4.1. The codes used are as follows: 0. Indirect Discourse 1. Narrative 2. Exposition 3. Personal Discourse 269 4. Direct Discourse  Auxiliary (Old Saxon only) This field contains 1 if the verb is one of those that forms periphrastic constructions with the auxiliary uuesan; otherwise it contains 0.  Period (Old English only) This field assigns a text to one of the periods defined in Section 3.2.4, numbered sequentially from 1 to 4.  Translation (Old English only) This field contains the category of the Latin grammatical form translated by the Old English verb, in accordance with the discussion in Section 3.4.2. If the verb does not occur in a passage of translation, this field is left blank. Otherwise, one of the following codes is used: 0. Perfect 1. Pluperfect 2. Perfect Participle 3. Imperfect 4. Present 10. Other 20. Interpolation 21. Expansion 22. Recast 23. Interchange The data included in these files are the basis of the statistical analysis presented in Chapter 4. It is hoped that their inclusion will facilitate the replication of these 270 results, should this be desired by anyone, and will make possible a greater understanding of how the results of the present study were obtained. 271 References Abusch, Dorit. 1997. ‘Sequence of Tense and Temporal De Re’, Linguistics and Philosophy, 20: 1–50 Andersen, Henning. 2008. ‘Grammaticalization in a Speaker-Oriented Theory of Change’, in Grammatical Change and Linguistic Theory: The Rosendal Papers, ed. by Thórhallur Eythórsson (Amsterdam: Benjamins), pp. 11–44 Aranovich, Raul (ed.). 2007a. Split Auxiliary Systems: A Cross-Linguistic Perspective (Amsterdam: Benjamins) ——. 2007b. ‘Split Auxiliary Selection from a Cross-Linguistic Perspective’, in Aranovich (2007a), pp. 1–23. Arnett, Carlee. 1997. ‘Perfect Auxiliary Selection in the Old Saxon Heliand’, American Journal of Germanic Linguistics and Literatures, 9: 23–72 Bately, Janet. 1978. ‘The Compilation of the Anglo-Saxon Chronicle, 60 BC to AD 890: Vocabulary as Evidence’, Proceedings of the British Academy, 64: 93–129 —— (ed.). 1980. The Old English Orosius, Early English Text Society SS, 6 (London: Oxford University Press) —— (ed.). 1986. MS A, The Anglo-Saxon Chronicle: A Collaborative Edition, 3 (Cambridge: Brewer) ——. 2009. ‘Did King Alfred Actually Translate Anything? The Integrity of the Alfredian Canon Revisited’, Medium Ævum, 78: 189–215 Behaghel, Otto, and Burkhard Taeger (eds). 1996. Heliand und Genesis, 10th edn (Tübingen: Niemeyer) Bennett, Jonathan. 1985. ‘Adverb-Dropping Inferences and the Lemmon Criterion’, in LePore and McLaughlin (1985), pp. 193–206 ——. 1988. Events and their Names (Oxford: Clarendon) Bethurum, Dorothy. 1957. The Homilies of Wulfstan (Oxford: Clarendon) Biber, Douglas. 1995. Dimensions of Register Variation: A Cross-Linguistic Comparison (Cambridge: Cambridge University Press) Braune, Wilhelm. 2004. Gotische Grammatik, 20th edn, rev. by Frank Heidermanns (Tübingen: Niemeyer) 272 Brinton, Laurel J. 1988. The Development of English Aspectual Systems: Aspectualizers and Post-Verbal Particles (Cambridge: Cambridge University Press) Burrow, J. A., and Thorlac Turville-Petre. 1996. A Book of Middle English, 2nd edn (Oxford: Blackwell) Burzio, Luigi. 1986. Italian Syntax: A Government-Binding Approach (Dordrecht: Reidel) Bybee, Joan, Revere Perkins, and William Pagliuca. 1994. The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World (Chicago: University of Chicago) Carey, Kathleen. 1994. ‘The Grammaticalization of the Perfect in Old English: An Account Based on Pragmatics and Metaphor’, in Perspectives on Grammaticalization, ed. by William Pagliuca (Amsterdam: Benjamins), pp. 103–117 Clackson, James. 2007. Indo-European Linguistics: An Introduction (Cambridge: Cambridge University Press) Clark Hall, J. R. 1960. A Concise Anglo-Saxon Dictionary, 4th edn, rev. by Herbert D. Merritt (Cambridge: Cambridge University Press, repr. Toronto: University of Toronto Press, [1984]) Clemoes, Peter. 1966. ‘The Composition of the Old English Text’, in The Old English Illustrated Hexateuch, ed. by C. R. Dodwell and Peter Clemoes (Copenhagen: Rosenkilde and Bagger), pp. 42–53 Comrie, Bernard. 1976. Aspect: An Introduction to the Study of Verbal Aspect and Related Problems (Cambridge: Cambridge University Press) ——. 1985. Tense (Cambridge: Cambridge University Press) Crawford, S. J. 1922. The Old English Version of the Heptateuch, Early English Text Society, 160 (London: Oxford University Press) Curat, Hervé. 1991. Morphologie verbale et r f rence temporelle en franc ais moderne : ssai de s manti ue grammaticale (Geneva: Droz) Dahl, Östen. 1985. Tense and Aspect Systems (Oxford: Blackwell) —— (ed.). 2000. Tense and Aspect in the Languages of Europe (Berlin: de Gruyter) —— and Eva Hedin. 2000. ‘Current Relevance and Event Reference’, in Dahl (2000), pp. 385–401 273 Davidson, Donald. 1967. ‘The Logical Form of Action Sentences’, in Davidson (1980), pp. 105–48 ——. 1969. ‘The Individuation of Events’, in Davidson (1980), pp. 163–80 ——. 1970. ‘Events as Particulars’, in Davidson (1980), pp. 181–7 ——. 1980. Essays on Actions and Events (Oxford: Clarendon) ——. 1985. ‘Reply to Quine on Events’, in LePore and McLaughlin (1985), pp. 172–6. de Acosta, Diego A. 2006. ‘ in Romance and English: Synchrony and Diachrony’ (doctoral dissertation, Cornell University) Denison, David. 1993. English Historical Syntax: Verbal Constructions (London: Longman) Derolez, R. 1995. ‘Genesis: Old Saxon and Old English’, English Studies, 76: 409–423 Diewald, Gabriele. 2002. ‘A Model for Relevant Types of Context in Grammaticalization’, in Wischer and Diewald (2002), pp. 103–20 Dixon, R. M. W. 1994. Ergativity (Cambridge: Cambridge University Press) Doane, A. N. 1991. The Saxon Genesis: An dition of the West Saxon ‘Genesis B’ and the Old Saxon Vatican ‘Genesis’ (Madison, WI: University of Wisconsin Press) Drinka, Bridget. 2003. ‘Areal Factors in the Development of the Indo-European Periphrastic Perfect’, Word, 54: 1–38 ——. 2007. ‘The Development of the HAVE perfect: Mutual Influences of Greek and Latin’, in Aranovich (2007a), pp. 101–21. Duden. 2005. Die Grammatik, 7th edn (Mannheim: Dudenverlag) Ebert, Robert Peter. 1978. Historische Syntax des Deutschen (Stuttgart: Metzler) Faarlund, Jan T. 2004. The Syntax of Old Norse: With a Survey of the Inflectional Morphology and a Complete Bibliography (Oxford: Oxford University Press) Fleischman, Suzanne. 1990. Tense and Narrativity: From Medieval Performance to Modern Fiction (London: Routledge) Fischer, Olga. 1992. ‘Syntax’, in The Cambridge History of the English Language, vol. II, ed. by Norman Blake (Cambridge: Cambridge University Press), pp. 207–408 274 Fortson, Benjamin W. IV. 2004. Indo-European Language and Culture: An Introduction (Malden, MA: Blackwell) Fuss, Eric, and Carola Trips. 2002. ‘Variation and Change in Old and Middle English: The Validity of the Double Base Hypothesis’, Journal of Comparative Germanic Linguistics, 4: 171–224 Gildersleeve, Basil L., and Gonzalez Lodge. 1895. Latin Grammar, 3rd edn (London: Macmillan; repr. Walton on Thames: Nelson, 1992) Giorgi, Alessandra. 2009. ‘Toward a Syntax of the Subjunctive Mood’, Lingua, 119: 1837–58 Godden, Malcolm (ed.). 1979. Ælfric’s Catholic Homilies: The Second Series, Early English Text Society SS, 5 (Oxford: Oxford University Press) ——. 1992. ‘Literary Language’, in The Cambridge History of the English Language, vol. I, ed. by Richard M. Hogg (Cambridge: Cambridge University Press), pp. 490–535 —— (ed.). 2000. Ælfric’s Catholic Homilies: Introduction, Commentary and Glossary, Early English Text Society SS, 18 (Oxford: Oxford University Press) ——. 2004. The Translations of Alfred and His Circle, and the Misappropriation of the Past, H.M. Chadwick Memorial Lectures, 14 (Cambridge: Department of Anglo-Saxon, Norse, and Celtic, University of Cambridge) ——. 2007. ‘Did King Alfred Write Anything?’, Medium Ævum, 76: 1–23 —— and Susan Irvine (eds and trans.). 2009. The Old English Boethius, 2 vols. (Oxford: Oxford University Press) Goodwin, W. 1894. A Greek Grammar, 3rd edn (London: Macmillan; repr. London: St. Martin’s, 1977) Gretsch, Mechthild. 1999. The Intellectual Foundations of the English Benedictine Reform, Cambridge Studies in Anglo-Saxon England, 25 (Cambridge: Cambridge University Press) Grünberg, M. 1967. The West-Saxon Gospels: A Study of the Gospel of St. Matthew with Text of the Four Gospels (Amsterdam: Scheltema & Holkema) Guillaume, Gustave. 1929. Temps et verbe (Paris: Champion) Harbert, Wayne. 2007. The Germanic Languages (Cambridge: Cambridge University Press) 275 Harris, Martin. 1982. ‘The “Past Simple” and the “Present Perfect” in Romance’, in Vincent and Harris (1982), pp. 42–70 Hecht, Hans (ed.). 1900. Bischof Wærferths von Worcester Übersetzung der Dialoge Gregors des Grossen (Leipzig: Wigand; repr. Darmstadt: Wissenschaftliche Buchgesellschaft, 1965) Heine, Bernd. 1993. Auxiliaries: Cognitive Forces and Grammaticalization (Oxford: Oxford Universiy Press) ——. 2002. ‘On the Role of Context in Grammaticalization’, in Wischer and Diewald (2002), pp. 83–101 Higginbotham, James, Fabio Pianesi, and Achille C. Varzi (eds). 2000. Speaking of Events (Oxford: Oxford University Press) Hill, Eugen. 2010. ‘A Case Study in Grammaticalized Inflectional Morphology: Origin and Development of the Germanic Weak Preterite’, Diachronica, 27: 411–58 Hoffmann, Gerhard. 1934. ‘Die Entwicklung des umschriebenen Perfektums im Altenglischen und Frühmittelenglischen’ (doctoral dissertation, Breslau) Hogg, Richard M. 1988. ‘On the Impossibility of Old English Dialectology’, in Luick Revisited, ed. by Dieter Kastovsky and Gero Bauer (Tübingen: Narr), pp. 183–203 Hopper, Paul J. 1991. ‘On Some Principles of Grammaticization’, in Traugott and Heine (1991b), pp. 13–35 —— and Elizabeth Closs Traugott. 1993. Grammaticalization (Cambridge: Cambridge University Press) Huddleston, Rodney. 2002. ‘The Verb’, in The Cambridge Grammar of the English Language, ed. by Rodney Huddleston, Geoffrey K. Pullum et al. (Cambridge: Cambridge University Press), pp. 71–212 Hundt, Marianne, and Nicholas Smith. 2009. ‘The Present Perfect in British and American English: Has There Been Any Change, Recently?’, ICAME Journal, 33: 45–63 Iatridou, Sabine, Elena Anagnostopolou, and Roumanya Izvorski. 2001. ‘Observations about the Form and Meaning of the Perfect’, in Ken Hale: A Life in Language, ed. by Michael Kenstowicz (Cambridge, MA: MIT Press), pp. 189–238 Irvine, Susan (ed.). 2004. MS E, The Anglo-Saxon Chronicle: A Collaborative Edition, 7 (Cambridge: Brewer) 276 Jacob, Daniel. 1995. ‘A propos de la périphrase habeo + participe parfait passif’, in Latin vulgaire — Latin tardif IV : actes du 4e Colloque International sur le Latin Vulgaire et Tardif, ed. by Louis Callebat (Hildesheim: Olms- Weidmann), pp. 367–381 Jakobson, Roman. 1957. ‘Shifters, Verbal Categories, and the Russian Verb’, in Selected Writings, vol. II (The Hague: de Gruyter, 1971), pp. 130–47 Joseph, Brian D., and Richard D. Janda. 1988. ‘The How and Why of Diachronic Morphologization and De-morphologization’, in Theoretical Morphology, ed. by M. Hammond and M. Noonan (New York: Academic Press), pp. 193–210 Kennedy, Charles W. (trans.). 1916. The Cædmon Poems (London: Routledge) Kenny, Anthony J. P. 1963. Action, Emotion and Will (London: Routledge) Kim, Jaegwon. 1966. ‘On the Psycho-Physical Identity Theory’, American Philosophical Quarterly, 3: 277–85 ——. 1973. ‘Causation, Nomic Subsumption, and the Concept of an Event’, Journal of Philosophy, 70: 217–36 Klein, Wolfgang. 2010. ‘On Times and Arguments’, Linguistics, 48: 1221–53 Kortmann, Bernd. 1991. ‘The Triad “Tense-Aspect-Aktionsart”: Problems and Possible Solutions’, Belgian Journal of Linguistics, 6: 9–30 Labourt, Jérôme (ed.). 1953. Lettres : Saint Jérôme, vol. III (Paris: Belles Lettres) Latham, R. E., and D. R. Howlett (eds). 1997. Dictionary of Medieval Latin from British Sources, Volume I (London: Oxford University Press) LePore, Ernest, and Brian McLaughlin (eds). 1985. Actions and Events: Perspectives on the Philosophy of Donald Davidson (Oxford: Blackwell) Lewis, C. T. and C. Short (eds). 1879. A Latin Dictionary (Oxford: Clarendon) Lindemann, J. W. R. 1970. Old nglish Preverbal ‘Ge-’: Its Meaning (Charlottesville, VA: University Press of Virginia) Lindstedt, Jouko. 2000. ‘The Perfect — Aspectual, Temporal and Evidential’, in Dahl (2000), pp. 365–83 Liuzza, R. M. 1994–2000. The Old English Version of the Gospels, 2 vols, Early English Text Society 304, 314 (Oxford: Oxford University Press) Lucas, Peter J. 1988. ‘Some Aspects of “Genesis B” as Old English Verse’, Proceedings of the Royal Irish Academy, 88C: 143–178 277 Lussky, George F. 1921. Die mit dem Partizip des Präteritums umschreibenen Tempora im Altsächsischen (Leipzig: Borna) Macleod, Morgan. 2008. ‘The Development of Middle Morphology in Selected Indo-European Languages’ (M.Phil. dissertation, University of Cambridge) McCoard, Robert W. 1978. The English Perfect: Tense-Choice and Pragmatic Principles (Amsterdam: North-Holland) McFadden, Thomas, and Artemis Alexiadou. 2010. ‘Perfects, Resultatives, and Auxiliaries in Earlier English’, Linguistic Inquiry, 41: 389–425 Marsden, Richard. 1995. The Text of the Old Testament in Anglo-Saxon England, Cambridge Studies in Anglo-Saxon England, 15 (Cambridge: Cambridge University Press) —— (ed.). 2008. The Old English Heptateuch, vol. I, Early English Text Society, 330 (Oxford: Oxford University Press) Michaelis, Laura. 1994. ‘The Ambiguity of the English Present Perfect’, Journal of Linguistics, 30: 111–57 ——. 1998. Aspectual Grammar and Past Time Reference (London: Routledge) Miller, Thomas (ed. and trans.). 1890. The Old English Version of Bede’s ‘ cclesiastical History of the nglish People’, Early English Text Society, 95–6 (London: Trübner) Minkoff, Harvey. 1976. ‘Some Stylistic Consequences of Ælfric’s Theory of Translation’, Studies in Philology, 73: 29–41 Mitchell, Bruce. 1975. ‘Linguistic Facts and the Interpretation of Old English Poetry’, in On Old English (Oxford: Blackwell, 1988), pp. 152–71 ——. 1985. Old English Syntax, 2 vols (Oxford: Clarendon) Mittwoch, Anita. 2008. ‘The English Resultative Perfect and its Relationship to the Experiential Perfect and the Simple Past Tense’, Linguistics and Philosophy, 31: 323–51 Moens, Marc, and Mark Steedman. 1988. ‘Temporal Ontology and Temporal Reference’, Computational Linguistics, 14: 15–28 Murphy, G. Ronald (trans.). 1992. The Heliand: The Saxon Gospel (Oxford: Oxford University Press) Mustanoja, Tauno F. 1960. A Middle English Syntax, Mémoires de la Société Néophilologique de Helsinki, 23 (Helsinki: Société Néophilologique) 278 Ogihara, Toshiyuki. 1995. ‘The Semantics of Tense in Embedded Clauses’, Linguistic Inquiry, 26: 663–79 Perlmutter, David. 1978. ‘Impersonal Passives and the Unaccusative Hypothesis’, Berkeley Linguistic Society, 4: 157–89 Pianesi, Fabio, and Achille C. Varzi. 2000. ‘Events and Event Talk: An Introduction’, in Higginbotham et al. (2000), pp. 3–47 Pintzuk, Susan, and Ann Taylor. 2006. ‘The Loss of OV Order in the History of English’, in The Handbook of the History of English, ed. by Ans van Kemenade and Bettelou Los (Oxford: Blackwell), pp. 249–78 Plummer, Charles. 1899. ‘Introduction’, in Plummer and Earle (1892–9), II, pp. xvii–cxxvii —— and John Earle (eds). 1892–9. Two of the Saxon Chronicles Parallel, 2 vols (Oxford: Clarendon; repr. 1965) Pope, J. C. (ed.). 1968. Homilies of Ælfric: A Supplementary Collection, Early English Text Society, 259–60 (Oxford: Oxford University Press) Portner, Paul. 2003. ‘The (Temporal) Semantics and (Modal) Pragmatics of the Perfect’, Linguistics and Philosophy, 26: 459–510 Potter, Simeon. 1939. ‘The Old English Orosius’, Transactions of the Philological Society, 38: 44–53 Quine, W. V. O. 1985. ‘Events and Reification’, in LePore and McLaughlin (1985), pp. 162–71 Rauch, Irmengard. 1982. ‘Use of the Germanic Past Perfect in Epic Backgrounding’, Journal of Indo-European Studies, 10: 301–14 ——. 1992. The Old Saxon Language: Grammar, Epic Narrative, Linguistic Interference (New York: Lang) Reichenbach, H. 1947. Elements of Symbolic Logic (New York: Macmillan) Ringe, Don. 2006. From Proto-Indo-European to Proto-Germanic, A Linguistic History of English, 1 (Oxford: Oxford University Press) Rissanen, Matti et al. (eds.). 1996. The Helsinki Corpus of English Texts (Helsinki: University of Finland), electronic, ICAME [accessed 10 February 2009] Roberts, Ian, and Anna Roussou. 2003. Syntactic Change: A Minimalist Approach to Grammaticalization (Cambridge: Cambridge University Press) 279 Rydén, Mats, and Sverker Brorström. 1987. The ‘Be’/‘Have’ Variation with Intransitives in English (Stockholm: Almqvist & Wiksell) Sasse, Hans-Jürgen. 1991. ‘Aspect and Aktionsart: A Reconciliation’, Belgian Journal of Linguistics, 6: 31–45 Schaden, Gerhard. 2009. ‘Present Perfects Compete’, Linguistics and Philosophy, 32: 115–41 Scherer, Philip. 1958. ‘Aspect in the Old English of the Corpus Christi MS’, Language, 34: 245–51 Schrodt, Richard. 2004. Althochdeutsche Grammatik II: Syntax (Tübingen: Niemeyer) Sedgefield, Walter J. (ed.). 1899. King Alfred’s Old nglish Version of Boethius ‘De Consolatione Philosophiae’ (Oxford: Clarendon) Sehrt, Edward H. 1966. Vollständiges Wörterbuch zum ‘Heliand’ und zur altsächsischen ‘Genesis’, rev. edn. (Göttingen: Vandenhoeck & Ruprecht) Shannon, Thomas F. 1995. ‘Toward a Cognitive Explanation of Perfect Auxiliary Variation: Some Modal and Aspectual Effects in the History of Germanic’, American Journal of Germanic Linguistics and Literatures, 7: 129–63 Siegel, Sidney. 1956. Nonparametric Statistics for the Behavioral Sciences (New York: McGraw-Hill) Sihler, Andrew. 1995. New Comparative Grammar of Greek and Latin (Oxford: Blackwell) Sisam, Kenneth. 1953. ‘Dialect Origins of the Earlier Old English Verse’, in Studies in the History of Old English Literature (Oxford: Clarendon), pp. 119–39 Skeat, Walter W. (ed.). 1871–1887. The Holy Gospels in Anglo-Saxon, Northumbrian, and Old Mercian Versions (Cambridge: Cambridge University Press) ——. (ed.) 1881–1900. Aelfric’s Lives of Saints, trans. by Skeat, Gunning, and Wilkinson, Early English Text Society, 76, 82, 94, 114, 2 vols (London: Trübner) Sorace, Antonella. 2000. ‘Gradients in Auxiliary Selection with Intransitive Verbs’, Language, 76: 859–90 Stotz, Peter. 1998. Handbuch zur lateinische Sprache des Mittelalters: Band 4 (Munich: Beck) 280 Sweet, Henry (ed.). 1871. King Alfred’s West-Saxon Version of Gregory’s Pastoral Care, Early English Text Society, 45, 50 (London: Trübner) —— (ed.). 1883. King Alfred’s Orosius, Early English Text Society, 79 (London: Trübner) Szemerényi, Oswald J. L. 1996. Introduction to Indo-European Linguistics, trans. by David and Irene Morgan Jones (Oxford: Oxford University Press) Taylor, Ann, et al. (eds). 2003. The York-Toronto-Helsinki Parsed Corpus of Old English Prose (York: University of York, 2003), electronic [accessed 28 January 2010] Thorpe, Benjamin (ed. and trans.). 1846. The Homilies of the Anglo-Saxon Church, 2 vols (London: Ælfric Society; repr. [Whitefish, MT]: Kessinger, [2007]) —— (ed. and trans.). 1853. King Ælfred’s Anglo-Saxon Version of the History of Paulus Orosius, in The Life of Alfred the Great (London: Bohn), pp. 238– 582 Timmer, B. J. 1948. The Later Genesis (Oxford: Scrivener) Tops, Guy A. J. 1974. The Origin of the Germanic Dental Preterite: A Critical Research History since 1912 (Leiden: Brill) Traugott, Elizabeth Closs. 1972. A History of English Syntax (New York: Holt, Rinehart, and Winston) —— and Bernd Heine (eds). 1991a. Approaches to Grammaticalization, 2 vols (Amsterdam: Benjamins) —— ——. 1991b. ‘Introduction’, in Traugott and Heine (1991a), I, pp. 1–14 Trousdale, Graeme. 2005. ‘The Social Context of Kentish Raising: Issues in Old English Sociolinguistics’, International Journal of English Studies, 5: 59– 76 Trudgill, Peter, Terttu Nevalainen, and Ilse Wischer. 2002. ‘Dynamic Have in North American and British Isles English’, English Language and Linguistics, 6: 1–15 van Kemenade, Ans. 1997. ‘V2 and Embedded Topicalization in Old and Middle English’, in Parameters of Morphosyntactic Change, ed. by Ans van Kemenade and Nigel Vincent (Cambridge: Cambridge University Press), pp. 326–352 Vincent, Nigel. 1982. ‘The Development of the Auxiliaries Habere and Esse in Romance’, in Vincent and Harris (1982), pp. 71–96 281 —— and Martin Harris (eds.). 1982. Studies in the Romance Verb: Essays Offered to Joe Cremona on the Occasion of his 60th Birthday (London: Croom Helm) Visser, F. T. 1963–73. An Historical Syntax of the English Language, 3 vols (Leiden: Brill) Watts, Sheila. 2001. ‘How to Become an Auxiliary: Progressive and Perfect in Old Saxon’, in Zur Verbmorphologie germanischer Sprachen, ed. by Sheila Watts, Jonathan West, and Hans-Joachim Solms, Linguistische Arbeiten, 446 (Tübingen: Niemeyer), pp. 117–135 Whitelock, Dorothy. 1962. ‘The Old English Bede’, Proceedings of the British Academy, 48: 57–90 ——. 1965. ‘Introduction’, in Whitelock, Douglas, and Tucker (1965), pp. xi– xxix ——, David C. Douglas, and Susie I. Tucker (eds and trans). 1965. The Anglo- Saxon Chronicle: A Revised Translation, rev. edn (London: Eyre and Spottiswoode) Wischer, Ilse. 2002. ‘The HAVE-“perfect” in Old English’, in New Perspectives on English Historical Linguistics: Selected Papers from 12 ICEHL, Glasgow, 21–26 August 2002, ed. by Christian Kay et al., 2 vols (Amsterdam: Benjamins, 2004), I, pp. 243–55 —— and Gabriele Diewald, eds. 2002. New Reflections on Grammaticalization, Typological Studies in Language, 49 (Amsterdam: Benjamins) Woods, Anthony, Paul Fletcher, and Arthur Hughes. 1986. Statistics in Language Studies (Cambridge: Cambridge University Press) Yerkes, David. 1982. Syntax and Style in Old English: A Comparison of the Two Versions of Wærferth’s Translation of Gregory’s Dialogues, Medieval & Renaissance Texts and Studies, 5 (Binghamton, NY: Center for Medieval & Early Renaissance Studies) ——. 1986. ‘The Translation of Gregory’s Dialogues and its Revision: Textual History, Provenance, Authorship’, in Studies in Earlier Old English Prose, ed. by Paul E. Szarmach (Albany: SUNY Press), pp. 335–43 Zeman, Sonja. 2010. Tempus und „Mundlichkeit“ im Mittelhochdeutschen (Berlin: de Gruyter) Zimmerman, Odo John (trans.). 1959. Dialogues of Saint Gregory the Great (New York: Fathers of the Church) Zimmermann, Rüdiger. 1968. ‘Untersuchungen zum frühmittelenglischen Tempussystem’ (doctoral dissertation, Kiel) 282 Original Texts (Latin) Bede Baedae Opera Historica, ed. by J. E. King, 2 vols, Loeb Classical Library (London: Heinemann, 1930) Boece The Consolation of Philosophy (from Godden and Irvine 2009, II) CP Règle Pastorale, ed. by Bruno Judic and Floribert Rommel, trans. by Charles Morel, 2 vols, Sources Chrétiennes, 381–2 (Paris: Cerf, 1992) GD Dialoghi, ed. by Adalbert de Vogüé, Opere di Gregorio Magno (Rome: Città Nuova, 2000) Oros King Alfred’s Orosius (Sweet 1883) Vulg Biblia Sacra, juxta Vulgatam Editionem Sixti V et Clementis VIII (Turin: Societa Editrice Internazionale, 1932) Wulf The Homilies of Wulfstan (Bethurum 1957) Original Texts (Old English) ÆCHom Aelfric’s Catholic Homilies (II), in The Helsinki Corpus of English Texts (Rissanen et al. 1996), HELSINKI/COAELHOM ÆGr Aelfric’s Grammatik und Glossar, ed. by Julius Zupitza (Berlin: Weidmann, 1880) ÆLS Aelfric’s Lives of Saints, in The Helsinki Corpus of English Texts (Rissanen et al. 1996), HELSINKI/COAELIVE Bede The Old nglish Version of Bede’s ‘ cclesiastical History of the nglish People’ (Miller 1890) Boece The Old English Boethius (Godden and Irvine 2009) ChronA1 Chronicle MS A Early (O2), in The Helsinki Corpus of English Texts (Rissanen et al. 1996), HELSINKI/COCHROA2 ChronA2 Chronicle MS A Late (O3), in The Helsinki Corpus of English Texts (Rissanen et al. 1996), HELSINKI/COCHROA3 ChronE Chronicle MS E, in The Helsinki Corpus of English Texts (Rissanen et al. 1996), HELSINKI/COCHROE4 CP King Alfred’s West-Saxon Version of Gregory’s Pastoral Care (Sweet 1871) GD Bischof Wærferths von Worcester Übersetzung der Dialoge Gregors des Grossen (Hecht 1900) 283 GenB Genesis B (Doane 1991) Oros King Alfred’s Orosius (Sweet 1883) OED The Oxford English Dictionary, ed. by John Simpson et al. (Oxford: Oxford University Press), electronic, OT The Old Testament, in The Helsinki Corpus of English Texts (Rissanen et al. 1996), HELSINKI/COOTEST WS West-Saxon Gospels, in The Helsinki Corpus of English Texts (Rissanen et al. 1996), HELSINKI/COWSGOSP Wulf The Homilies of Wulfstan (Bethurum 1957) Original Texts (Old Saxon) Genesis Genesis (Doane 1991) Heliand Heliand (Behaghel and Taeger 1996)