Olga Fischer the perfect in Old English may also be partly a matter of the style and subject matter of the extant manuscripts. Zimmermann has noted that the Middle English (plu)perfect occurs more often in colloquial style; it is rare in purely narrative contexts and frequent in instructional texts. Thus, in the Early Middle English texts he has examined (Ancrene Wisse, the ' Katherine Group', La3amon's Brut and The Owl and the Nightingale), the perfect occurs almost exclusively in direct speech. This also explains why the perfect occurs mostly in the first and second person, while the preterite and pluperfect are rare in these cases. He also shows (pp. 155-8) that in the Old English Orosius, where the perfect is seldom used, it is found in places where the sentence structure is looser, especially in conclusions of chapters and in connecting passages. Thus it is possible that the greater frequency of the (plu)perfect in Middle English not only reflects a change in actual usage but is also related to the nature of the extant texts. However, another cause might be the general change in the English language from a morphological tense/mood (aspect) system to a grammaticalised auxiliary system. As such it shares a trend found in other Germanic languages. The (plu)perfect is not fully grammaticalised in Middle English: it freely alternates in almost all its functions with the preterite. The type of constructions, however, in which the (plu)perfect occurs already anticipates its later usage in present-day standard English. The primary function of the perfect is to indicate that an activity has started at a certain moment in the past but that it is still rele- vant/continuing at the moment of speaking: (128) And alle \>o )?at seyn for me a Pater noster I make hem parteneres & graunte hem part of all [\>e] gode pilgrymages & of all the gode dedes f>at I haue don, 3if ony ben to his plesance. {Mandev. (Tit) 210.36-211.4) The preterite is also commonly found in these constructions. When the relevance of the activity is related not so much to the moment of speaking but to the future or the generic present (which includes the future), the perfect is clearly favoured over the preterite. Zimmermann (1968: 110-12) has found no examples of the preterite here: (129) Ase ofte as T,C habbed /m/eawiht her on, greted j?e leafdi wiS an aue; for him )?et swonc her abuten. (Ancr. (Coxp-C) 222.12-14) 'As often as you have read anything in this [book], greet the Lady with an Ave for him who took pains over it.' 258 Syntax (130) ye hen hwen ha haued ileid, ne con bute cakelin. {Ancr. (Corp-C) 36.18) 'the hen when she has laid, knows no better than to cackle.' As in Present-Day English certain temporal adverbs favour a particular construction. Thus, now, here, today, etc. are usually found with the perfect, whereas (n)ever, pa, etc. are commonly combined with the preterite. But it is not unusual in Middle English (especially in poetry, so metrical considerations could play a role here) to find a perfect with a past-time adverbial: (131) I am youre doghter Custance / That whilom ye ban sent unto Surrye. 1 (CT II.1107-8 [3: 1107-8]) Just like the non-past, the perfect is found in narrative past-time contexts often in conjunction with the preterite. It is not clear how far the perfect has a special function, and how far the exigencies of rhyme and metre are responsible, since this phenomenon occurs mainly in poetry (see Mustanoja 1960: 506-7; Visser 1963-73: §§766, 772): (132) His brother, which that knew of his penaunce,/ Up caughte hym and to bedde he hath hym broght. (CT V.1082-3 [6: 374-5]) The pluperfect occurs exclusively in narrative passages and always refers to a completed action. In most cases it is used, as in Present-Day English, to indicate a past within the past. It is often accompanied by clarifying temporal adverbials such as (f>er)biuoren, ear 'ere', etc. and occurs especially in temporal subclauses. In Middle English, however, a preterite can also be used. 9 The first example shows a pluperfect, the second a preterite: (133) I schewed hym this tretys )?at I had made after informacioun of men )>at knewen of thinges )?at I had not seen my self, (Mandev. (Tit) 210: 1-3) (134) Moyses was blide / And ches 6o men god made wis. (Gcn.& Ex.3671-2) 'Moses was glad and chose those men that God had made wise.' The description of the knight in Chaucer's 'General Prologue' (CT 1.47-63) shows that preterite and pluperfect forms could easily be interchanged. The pluperfect was used in hypothetical contexts to indicate what could or should have happened but did not (' counterfactuals'). This use 2 59 Olga Fischer is equivalent to that of the modal preterite, except the pluperfect refers to a past in the past. Like the modal preterite, the modal pluperfect was originally a subjunctive, but the loss of inflections caused the indicative and the subjunctive past forms of habben to become alike, with the exception of the second-person singular. This subjunctive form, however, was soon levelled out in favour of the indicative: (135) ' Haddestow be hende', quod I, ')?ow woldest haue asked leeue.' (PP/.B (Trin-C) xx, 188) (For the use of the inverted word order in this clause, see section 4.6.3.3.) The modal pluperfect is also found in main clauses where nowadays we prefer a modal verb: (136) And 3if here fader had not ben dronken he hadde not jleye with hem. (Mandev. (Tit) 68: 10-11) In addition to the (plu)perfect formed with the auxiliary habben/hauen, there was a form with ben. As in Old English, this was mainly restricted to the so-called mutative verbs (i.e. intransitive verbs involving a change of place or state). However, the distinction in Middle English becomes less clear-cut. Habben encroached upon the ^«-domain, while ben is found with some intransitive, non-mutative verbs. A number of verbs occur with both auxiliaries. Attempts to describe the difference between the two constructions have not been completely successful so far in that they always admit of exceptions. Some linguists have described the habben construction as denoting an action and the ben construction as denoting a state (see Bogholm 1944; Friden 1957; Visser 1963-73: § 1898). It is unlikely that such a clear dichotomy exists in view of the facts that the two constructions often seem to be interchangeable and that in the later development have ousts the be forms. Such a dichotomy would also mean that a structure like he is come always referred to state, never to completed action. Friden (1948) formulated the rule that mutative verbs take have when they are used transitively as in he has gone half a mile. Visser (1963—73: §1898) objects to this because, he says, there are many exceptions. However, a large number of these exceptions can be explained if one widens Friden's rule (see also Zimmermann 1973) so that with mutative verbs be is normally used when location (in time or place) or direction is emphasised, while have is used to refer more purely to the activity conveyed by the verb, as the following examples show: 260 Syntax (137)a. Be wel avysed on that ilke nyght/ That we ben entred into shippes bord,'/ That noon of us ne speke nat a word, (CT 1.3584-6 [1: 3578-80]) b. For ye ban entred into myn hous by violence, (C7" V11.1812 [10: 1812]) (138)a. The nexte houre of Mars folwynge this,/ Arcite unto the temple walked is I Of fierse Mars to doon his sacrifice, (CT 1.2367-9 [1: 2369-71]) b. 'Saw ye,' quod she, 'as ye ban walked wyde,/ Any of my sustren walke yow besyde ' (LCIV 978-9) For this reason it is not surprising to find adverbials of manner or degree, which highlight the activity of the verb, more often collocated with mutative verbs in have constructions, and adverbials of time and place with the same verbs in be constructions (as in the above examples). It also explains why the have construction prevails in hypothetical statements: there the emphasis is always on the activity, on what should have happened: (139)a. She wende nevere ban come in swich a trappe. (CT V.1341 16: 633]) b. He wende have cropen by his felawe John,/ And by the millere in he creep anon, (CT 1.4259-60 11: 4251-2]) Friden (1948: 43—57) gives a list of contexts in which have is preferred to be. They all agree with the above observation except for one category: 'Have is used if the sentence contains an adverb or adverbial phrase denoting the place of action' (p. 48; italics mine). However, in all the examples he quotes this adverbial can be interpreted as one of degree: ' and when you have gone as far as you can',' thy slander has gone through and through her heart''. Finally, it remains to consider the reason(s) why have ousted be in the formation of the perfect. Various factors are at work (see Zimmermann 1973): (a) the greater functional load of be (used as an auxiliary of the passive, progressive and perfect) compared to have (at this stage only perfect) and the ambiguity that could arise because of this (i.e. ^ + past participle of a transitive verb could be perfect as well as passive; the progressive in -ing is sometimes found written as -en, the same ending as the past participle of strong verbs) made have a more suitable candidate 261 Olga Fischer for the perfect; (b) be had become the auxiliary par excellence of the passive voice. One might have expected wurthen to play a larger role here as it did in other West Germanic languages like German and Dutch. But from earliest Old English, weorpan had been far less frequent in the passive, and it became very infrequent in Early Middle English, when the auxiliary system was undergoing great change. Why wurthen disappeared is still unclear; reasons for this development have been sought in the nature of the verb itself, while it has also been ascribed to foreign influence (for a discussion of the possible factors, see Mustanoja 1960: 616-19). When the (plu)perfect became frequent in Middle English, it was necessary to streamline the various constructions; and have, which already acted as the auxiliary of the perfect with most types of verbs and even occurred with mutatives in special constructions, slowly took over the functions of be. This development was possibly also facilitated by the fact that in co-ordinate constructions the auxiliary was usually not repeated so that have was often used where be was expected. Another factor that is frequently adduced is the influence of the use of the reduced form 's (especially in spoken language), but evidence for has reduced to 's (from spelling and metre) is mainly post-Middle English. 4.3.3.3 Modal auxiliaries In section 4.3.2.2, we have seen that by the end of the Middle English period periphrastic constructions far outweighed subjunctive forms. This development started in Late Old English when periphrastic constructions became increasingly common. What probably happened is this: on the one hand, the gradual erosion of verbal inflections made it necessary to replace the subjunctive by something more transparent; on the other, the use of periphrastic constructions at a fairly early stage was itself responsible for the disappearance of the subjunctive. The early use of the periphrastic construction may be due to a desire to be more emphatic and possibly to be more specific than was possible with the subjunctive form. Interesting in this respect is the use in Old English of periphrastic auxiliaries which are themselves in the subjunctive form. Together with the loss of the subjunctive came a grammaticalisation of the modal verbs, which in Old English in many ways still had the status of full verbs (but see Warner 1990). A full list of changes that the modal verbs underwent can be found in Plank (1984). This list shows that the development was gradual and not of a radical nature, as suggested in Lightfoot (1979). Some of the more important changes 262 Syntax taking place in Middle English but not completed in that period (and continuing for a long time after in certain dialects) are: (a) the modals lost the possibility of appearing in non-finite forms and of taking objects; in general they move towards an invariable form. This is related to the loss of the notional meaning of the modals. (b) Tense differences in modals no longer serve a temporal purpose, (c) The close relation between a modal and its infinitive is emphasised by the fact that the to- infinitive never replaces the bare infinitive as happened after most other full verbs, and also by the increasing unwillingness of modals to appear without an infinitive of another verb in series. In Middle English we still find the modal verbs used in some of their 'non-auxiliary' functions (for Old English see vol. I, section 4.3.2.3); (140) shows them as full verbs, (141) in non-finite forms. 10 (140)a. She koude muchel of wandrynge by the weye. (CT 1.467 [1: 469]) ' She knew a lot about travelling.' b. And by that feith I shal Priam of Troie. (Jroilus 111.791) 'And by the faith I owe Priamus of Troy.' c. And seyde he mostt unto Itayle, (HF 187) 'And said he had [to go] to Italy' (141)a. E>att mannkinn shollde mu^henn wel/ Upp cumenn inntill heoffne (Orm. 3944-5) 'that mankind should be able to go up to heaven' b. But Pandarus, if goodly hadde he myghtJWe. wolde han hyed hire to bedde fayn, (Jroilus 111.654-5) Next to the so-called 'core' modals {sbal, mil, may, mot, can) periphrastic constructions also expressed modality. Some of these occur in Old English (e.g. 'to be to', 'to have to'). These, together with the Middle English innovation borrowed from Old French ' to be able to', remain sporadic until they come to fill a systemic gap left by the grammaticalisation of the core modals, which, as we have seen, lost all but their non-finite forms (and to a great extent they even lost their finite past forms). For other examples of periphrastic constructions conveying modality, see Plank (1984: 321-2). A rather special development is the use of the modals, especially shal 263 Olga Fischer and wil (but in some cases also mot), as markers of the future. Since, however, these constructions remain modally marked for the greater part of the period (see Mustanoja 1960: 490-1), they are discussed here and not under tense. Already in Old English * sculan / willan are used with predictive meaning, but in these cases *sculan usually expresses obligation or necessity as well, and willan volition. (Traugott (in vol. I, section 4.3.2.3) states that there are no examples in Old English where * sculan or willan has pure future reference. Warner (1990), however, shows that these verbs must be mere futurity markers when they occur in impersonal constructions in Old English (see also Mitchell 1985: §§1023fF).) This situation continues in Middle English: (142) And rightful folk shul gon, after they dye,/ To hevene; (PF 55-6) In the above example .r^w/expresses future as well as 'ordained event'. Thus, shal is more frequent in prophesies, in contexts in which a sense of obligation is present, in commands and instructions. Because of this modal function, shal'is particularly common in the third person. Wil, on the other hand, occurs far more often in the first person, since modally it is connected to the desire of the speaker/subject: (143) we wulled folhi ]>e, we wulled don alswa, leauen al as ]>u dudest (Ancr. (Corp-C) 87.6-7) 'we will follow you, we will do likewise, [we will] leave everything [behind] as you did' Wil occurs especially in promises, wishes and resolutions. Because shal is not related to the will of the subject, it develops into a pure future marker earlier than wil (see Mustanoja 1960: 490). The more frequent use of predictive shal in Early Middle English may, however, also be due to other factors. It could be partly a matter of style. In biblical writings shal is preferred to wil {wil is reserved for the translation of Latin velle). Wil seems to be a product of a more popular style. The pure future use of wil may have developed out of its use in generic and habitual contexts, (144) He is a fool that wol foryete hymselve (Treilus V.98) Examples that indicate that wil is used without any modal colouring are those that have inanimate subjects (although note the personified nature of the subject in this particular instance): (145) And I, book, wole be brent but Iesus rise to lyue (PPl.B (Trin-Q xviii. 2553 264 Syntax Other verbs employed in periphrastic future expressions were wurthen (see Mustanoja 1960: 495), in Late Middle English to be about to (Mustanoja 1960: 354) and right at the end of the period to be going to (Mustanoja 1960: 592). The expanded form be + ing was not used for future reference in Middle English (except with the verb to come which is inherently futural) even though it was fairly frequent in Old English particularly as a translation of Latin esse + -urus (see Visser 1963-73: §1830). 4.3.3.4 The periphrasis with gan The verb ginnen is used in Middle English, just like beginnen, to refer to the beginning of an action: (146) Se)?)?en f»at ich here regni gan/ Y no fond neuer so fole-hardi man (Or/eo (Auch) 425-6) 'From the time I began to rule here I have never found anyone so foolhardy.' However, there are many contexts in which such an inchoative meaning does not fit: (147) ]?us )>e bataile_£w« leste long/ Til }>e time of euesong, (Bevis (Auch) 797-8) 'thus the battle went on a long time, till evensong' (148) A knaue he. gan imete. {Horn (Cmb) 940) ' He met a lad' In (147) the inchoative function of gan is incompatible with the adverbial adjunct long, which expresses duration. In (148) the inherently punctual (non-durative) verb mete cannot be combined with ingressive^/?, which only collocates with durative verbs. It is clear that in these examples gan has a different function. The very beginnings of this new function of gan are found in Old English with the related verb onginnan, as shown by Funke(1922: 8-9). As to what this new function of gan entails, there seem to be two schools of thought. One maintains that gan is a mere stopgap and is exclusively used as a metrical device (Visser 1963—73: §§ 1477ff.; Smyser 1967; Terasawa 1974; Tajima 1975). The other believes that gan has a particular descriptive function, that it is used as a stylistic device, which later also becomes, or could become, a mere line-filler (Funke 1922; Homann 1954; Mustanoja 1960, 1983; Kerkhof 1982; Brinton 1983; cf. 265 Olga Fischer Brinton 1988: §3.8). The evidence available strongly supports the 'mere stopgap' theory: the£tf«-periphrasis occurs almost exclusively in poetry. In Chaucer, for instance, nearly 700 instances have been found in his poetry, and only three in his prose (all in Melibee) of which two are probably a direct translation of OF commence Likewise, in his study of the Gawain-poet, Tajima notes that con/can (the northern form of gari) appears only very sporadically in the unrhymed alliterative lines, but is frequent in rhymed lines. Ninety-five per cent of the examples put the infinitive in rhyme position. In Chaucer the equivalent figure is 73 per cent, according to Smyser. Another interesting feature is that the construction occurs only in the past tense (with the exception of Pearl, where some present-tense forms are found). Smyser (1967: 74) explains this as follows: the preterite, especially of weak verbs, is very difficult to rhyme; for that reason, the infinitive is preferred in rhyming position. His evidence supports this: verbs that have the same form in preterite and infinitive {hente, sette, sterte, stente) only occur twice in Chaucer in the £«»-periphrasis (compared to eighty-two times in the simple form). This should be contrasted with e.g. cried/gan crye: cried occurs six times in rhyme, gan crye twenty-one times. The evidence provided by the second school of thought for their hypothesis is based on the presumption that a great poet like Chaucer, who uses the construction frequently, would not have stooped to the use of stopgaps (Homann 1954; Brinton 1983). Funke (1922) suggests that there must have been an intermediate stage between inchoative function and pleonastic use. He suggests that gan was used as a signal to introduce a new event, that it has a descriptive, intensive function. Although such an intermediate stage is likely he.iote.gan was semantically reduced to zero, we are quite in the dark as to the meaning of gan at that stage. This is clearly shown by the many different interpretations that have been offered for this descriptive function. It is also difficult to prove conclusively whether this descriptive function continued to exist side by side with stopgap gan at a later stage. That different linguists give widely different interpretations of the meaning of the descriptive function of gan is a serious weakness of this theory. Homann believes that Chaucer 'utilized "gan" to add vitality to dramatic scenes, intensity to emotional situations, and an inner meaning and depth to his characters' (1954: 398). Brinton (1983: 244) very tentatively suggests that gan may convey a notion of contingency. The problem with all these suggestions, especially Homann's, is that they can be read into the context, the danger of' hineininterpretieren' looms 266 Syntax large. On the other hand one can see that if gan was used as a stopgap, it could be easily turned to stylistic use. In the following example from Chaucer, for instance: (149) For with that oon encresede ay my fere/ And with that other gan myn herte bolde;/ That oon me hette, that other dide me colde; (PF 143-5) it is likely that Chaucer used gan not just to put bolde in rhyming position, but also to put it in a symmetrical position to fere with which it forms a contrast. Notice also the completely symmetrical ordering of all the other clause structures in these lines. Just as it is virtually impossible to decide whether gan has a descriptive function along with its line-filler function, and, if so, which of the two it is in each individual case, it is also almost impossible to distinguish between the ingressive function oigan and its other uses. Brinton (1983) gives criteria to do just that (criteria that have been used by Smyser before). She discusses the collocation possibilities between aspectual ginnen and the main verb (be)ginnen. What needs to be realised, however, (and this clearly diminishes the overall usefulness of these criteria) is that, if the collocation indeed allows of an ingressive interpretation, that does not necessarily mean that gan could not in that very example also be a stopgap. Once gan has developed that function one can expect to find it everywhere. 4.3.3.5 The verb do The Middle English period is a time of rapid expansion in periphrastic constructions involving the rise of an auxiliary system, including perfect have/be, progressive be and the modals. The development of a periphrastic verb do in this period is of considerable importance on account of the later establishment of this do as an empty syntactic marker in constructions in which the simple verb no longer suffices (e.g. in negative and interrogative clauses). As we will see, do begins to fill a gap that results from the development of the auxiliary system described above. Before we look at Middle English developments, we will consider the way the verb do was employed in Old English. OE don was used (a) as a full lexical verb ('notional' or Visser's 'factitive do'): (150) Uton don hyne on )>one ealdan pytt (Gen. 37.19) 'Let us do [= put] him in this old well' 267 [...]... ( 157 ) so ded sir Galahad delyver all the maydyns oute of the woofull Ca (Malory Wks (Add .59 678) 8 92. 13-14) Denison (1985b) (and see also Mitchell 19 85: §666) has shown convincingly that all the earlier examples attested by Visser are suspect in a number of ways The majority of Visser's Old English examples are of the form shown in ( 151 ), where the infinitive {habban) depends on uton and not on don Other... a pure maiden - to the creator of every creature Normally, in Present-Day English as well as in Middle English, the subject of the co-ordinate clause is omitted if it is co-referential with the subject of the main clause However, the omission of the subject was more extensive in Middle English (as it was in Old English) This may be a left-over from the time when the inflection on the verb made a pronominal... are: (a) there may be more than one modal in the sequence and the modal can still occur as V, either by itself or in the form of a past or present participle (see the examples in (140) and (141)); (b) whereas in Present- 2 77 Olga Fischer Day English only an adverb can be placed between the finite verb and the rest of the verb phrase, other sentence elements could still occur there in Middle English, ... {St.juliana (Roy) 25 .24 4) 'allow me that I may thee truly love.' Concerning the position of the infinitival verb with respect to the auxiliary, in Early Middle English (as in Old English) the infinitive was still often positioned before the auxiliary in subclauses, a feature typical of SOV languages In the later periods it can still be preposed but only as a marked construction Sanders (19 15: llff.) shows... for Present-Day English as well, this remark is more relevant for the older stages of the language, for Old English even more than for Middle English At that time the written language often presented ideas paratactically where written Present-Day English would use subordination (hypotaxis) In Old and 28 7 Olga Fischer Middle English the written language was still closer to the spoken language, which... participle rather than the expected infinitive (like the examples quoted above) as a mental process whereby the speaker has shifted his mind from the giving of the order {hatan + infinitive) to the accomplishment of it (hatan + past participle) 27 4 Syntax What makes the notion of a stage in which do x was perfective so attractive is the use Denison makes of it in connection with the regulation of periphrastic... Present-Day English, the wh-word is itself the subject Although do is found (albeit rarely) in questions in Middle English, such constructions should probably not be interpreted as containing empty interrogative do; rather, they are the questioned counterpart of a clause already containing do The first attested example is from Chaucer: (174) Fader, why do ye wepe? (CT VI1 .24 32 [10: 27 28]) 27 8 Syntax Only in the. .. instead of / think that it is not true) are not yet very frequent in Middle English but some examples can be found: (20 0) He thenkith nought that evere he shall/ Into ony syknesse fall (Rose 5 621 -2) The following instance looks like a case of negative concord rather than negative raising (see vol I, section 4 .5. 10): (20 1) amonges thise thynges / ne trowe not that the pris and the grace of the peple... discussion of the causative nature of the Old English verb hatan, he writes, " the verb of causing predicates the accomplishment of an act that has been brought about by the exercise of an influence of some one or of some thing upon some person or some object The causative verb affirms accomplished action; it is a perfective verb." He explains an Old English example where hatan is followed by a past participle... heavier use of parataxis than of hypotaxis (see Phillipps 1966a; Leith 1983: 1 12) It is only at the end of the Middle English period, with the development of a written standard, that the written language begins to make more extensive use of complex structures, under the influence of both French and Latin prose styles (see Fisher 1977) Here follow some instances of such paratactic structures: (21 0) and . respect is the use in Old English of periphrastic auxiliaries which are themselves in the subjunctive form. Together with the loss of the subjunctive came a grammaticalisation of the modal verbs,. John,/ And by the millere in he creep anon, (CT 1. 4 25 9-60 11: 4 25 1 -2] ) Friden (1948: 43 57 ) gives a list of contexts in which have is preferred to be. They all agree with the above observation. Fischer the perfect in Old English may also be partly a matter of the style and subject matter of the extant manuscripts. Zimmermann has noted that the Middle English (plu)perfect occurs more often