8.2.1 To-infinitives

From the OE evidence, the prehistory of the to-infinitive can be easily surmised. Its to-marker can be linked to the homonymous preposition, and the infinitive verb preserved what looks like a fossilized dative case-ending (-e) appropriate to a noun in a prepositional phrase. It has even been argued that in OE the to-infinitive still was a prepositional phrase. However, Los (Reference Los2005: 170), following a review of the evidence, concludes otherwise. Among other things, she shows that to-infinitives obligatorily followed the finite main verb, as in (9a), whereas prepositional phrases could also precede it, as in (9b).

1. 1. a ða  eaðmodan … weorðen geniedde hiera unðeawas   to     herianne

      the humble          be         forced     their  faults         to     praise (CP.41.302.13)

      ‘The humble … would be forced to praise their faults.’

   2. b Gif hwa      to hwæðrum   þissa       genied sie    on woh

      if  anyone to either        of-these  forced be    unjustly (LawAf.1(1–1))

      ‘If anyone is forced to either of these unjustly …’

She also shows that to was inseparable from the infinitive verb and that when two infinitive verbs were conjoined, as in (10), to was obligatorily repeated – both indications that to had cliticised to the infinitive verb. The preposition to, in contrast, was again not subject to those restrictions.

1. (10) He hæfde þa gleawnesse Godes bebodu            to   healdanne  &     to   læranne

   He had     the wisdom      God’s  commandments  to   obey            and    to   teach (Bede.3.14.206.10)

   ‘He had the wisdom to obey and to teach God’s commandments.’

Curiously, to decliticized again from the infinitive verb in ME, giving rise to conjoined infinitives with a single to, as in (11a), as well as to the infamous split infinitive, illustrated in (11b).

1. 1. a Bidde we nu þe holigost þat he … gife us hige and mihte to forleten and bireusen and beten ure sinnes.

      ‘Let us now ask the Holy Spirit that he give us the disposition and power to abandon and repent of and defeat our sins.’ (OEHom.Ch.XX)

   2. b I kepe nouȝt to hastiliche afferme wheþer it be in þat place oþer no.

      ‘I do not wish to hastily affirm whether it is in that place or not.’ (Polychron.VI,63.435)

The decliticization of to has attracted some attention in the grammaticalization literature, because it goes against the expected tendency for grammaticalizing elements to undergo progressive formal coalescence and reduction (Norde, Reference Norde2009: 190–99). Different explanations have been put forward. Fischer (Reference Fischer1997) links both the decliticization of to and the loss of reinforcing for (see the discussion that follows below) to renewed analogical attraction between infinitival to and the preposition to. Los (Reference Los2005: 211), following Pullum’s (Reference Pullum1982) more theory-driven analysis of PDE infinitival to, proposes that to decliticized when it was reanalysed as the non-finite counterpart of a modal verb.

Several other changes took place in the to-infinitive’s internal syntax. There was the appearance of formally marked passives. OE to-infinitives occasionally conveyed passive meaning, even though this was not marked by a passive auxiliary. A typical context was the construction in (12), which is sometimes called the ‘modal passive’. Its subject was the patient of the infinitive verb, and the construction as a whole conveyed root or deontic modality. In PDE, the modal passive survives in a few lexicalized expressions, such as This house is to let and You are to blame.

1. 1. a hit     nis       no     to     forseone

      it       is-not   not    to    despise (Bo.24.56.2)

      ‘It is not to be despised.’

   2. b Eac     is    ðeos   bisen         to     geðencenne …

      Also    is    this     example     to     think-of … (Bo.23.52.2)

      ‘This example can also be thought of …’

Formally passive infinitives began to appear in the beginning of ME, as in (13a). Their rise has been linked to word-order change and is discussed in some more detail in Chapter 9. As to the modal passive construction, it too was increasingly infiltrated by formally marked passives, as in (13b).

1. 1. a he till hiss Fader wass/Offredd forr uss o rode, / All als he wære an lamb to ben / Offredd

      ‘He was offered to his Father for us on the cross, just as if he was a lamb to be offered.’ (Orm.12644–47)

   2. b The whiche hevene … nys nat … to be wondryd upon

      ‘… which heaven is not to be marvelled at.’ (Chaucer,Boece,III.8)

Also in ME, the to-infinitive first began to appear with its own explicit subject, in various constructions. An example is given in (14), where (on semantic grounds) stoones must be the subject of be turned rather than the indirect object of axid. This development, too, has been linked to word-order change and is therefore further discussed in Chapter 9.

1. (14) Suche men semen to turne þe breed of pore men into stoones, and in þis þei ben more cruelar þan þe deuel þat axid stoones to be turned into bred.

   ‘Such men seem to turn the bread of the poor into stones, and in that respect, they are crueller than the devil who asked that stones should be turned into bread.’

   (Wycliffe,Lant.of Liӡt)

Finally, the infinitive marker to was sometimes reinforced, especially in contexts where the to-infinitive was to express its original purposive meaning. Perhaps under Scandinavian influence (Kytö and Danchev, Reference Kytö and Danchev2001), one reinforcement strategy was by adding for, as in (15). Except in some traditional dialects, reinforcing for again disappeared soon after ME. More recently, in order and so as were recruited to do a similar job, as illustrated in (15b–c).

1. 1. a And þe scherref aboute Gamelyn for to take.

      ‘The sheriff [went] in all directions in order to capture Gamelyn.’ (Chaucer,Gamelyn,550)

   2. b I shall next Week come down in order to take my Seat at the Board. (1711, OED)

   3. c The devil and his disciples are notable method-mongers, so as to deceive, if it were possible, the very elect. (1647, OED)

The introduction of reinforcement strategies may have to do with what happened to the external syntax of the to-infinitive. Given their reconstructed prehistory, as well as the evidence from Gothic (Los, Reference Los2005: 28–31), it is plausible that to-infinitives started out as adverbial clauses marking purpose. Already in OE, however, to-infinitives had expanded to a broader range of contexts, including uses as noun clauses, though they typically preserved some sense of purpose. For example, they could be used as theme-argument to intention verbs, as in (16) (Los, Reference Los2005: 169).

1. (16) Esau     ðin     broðor     þe    ðencð     to     ofsleane

   Esau     your   brother     you   intends   to     kill (Gen.27.42))

   ‘Your brother Esau intends to kill you.’

By PDE, to-infinitives can be used in a range of syntactic positions. Sometimes, the original purposive meaning is no longer very prominent, as in (17a) or (17e), possibly explaining why infinitival to may occasionally take a reinforcing element when purposive meaning needs highlighting.

1. 1. a [Subject]

      To deny it is therefore impossible. (BNC)

   2. b [Subject complement]

      Their plan was to feed on the forest as they went (BNC)

   3. c [Noun complement]

      The idea to promote an urban walkway/cycleway on the disused railway line from Easter Road to Seafield goes back to the Leith Local Plan reports of the 1970s. (BNC)

   4. d [Verb complement]

      They wanted to pass on this property to their children (BNC)

   5. e [Adjective complement]

      he began to wonder if Smallfry was right to fear she would poison him at the slightest opportunity. (BNC)

Some of the constructions in (17) have had complicated histories and deserve a closer look. To-infinitives used as verb complements have been on the increase since OE. To an important extent, this went at the expense of that-clauses (Manabe, Reference Manabe1989; Los, Reference Los2005). Thus, after verbs expressing an order, purpose or intention, the þæt-clause was the rule in OE, where the to-infinitive would now be much more common. Los (Reference Los2005: 179–85) demonstrates that this development was already underway in OE. She shows this by comparing two versions of the same text, Gregory’s Dialogues, one from the late ninth century, and a reworked version about a century older. A good number of þæt-clauses in the former have been substituted by to-infinitives in the latter, as illustrated in (18).

1. 1. a Dauid,    þe      gewunade,            þæt   he hæfde witedomes     gast     in     him

      David    who   was-accustomed  that  he had     of-prophecy spirit   in    him

      (GD.1(C)4.40.24)

   2. b Dauid,    þe     gewunode              to     hæbbenne  witedomes    gast   on him

      David     who   was-accustomed  to    have           of-prophecy spirit in  him

      (GD.1(H)4.40.22)

      ‘David, who was accustomed that he had / to have the spirit of prophecy in him.’

Next to to-infinitives and that-clauses, the domain of verb complementation also saw some variation between to-infinitives and bare infinitives. More so than to-infinitives, bare infinitives appear to have been used in contexts involving high levels of conceptual integration between the main clause event and the subordinate event.Footnote ¹ For example, the bare infinitive was favoured in cases where main verb and subordinate verb had a shared time domain and there was direct logical entailment (Fischer, Reference Fischer1995). In (19a), the bare infinitive stigan denotes a state of affairs that is understood to be co-temporal with and directly dependent on the main verb nydan. In (19b), by contrast, the to-infinitive to beten denotes a state of affairs in its own time domain, posterior to that of the main clause.

1. 1. a Đa    sona      he nydde  his leorningcnihtas on scyp stigan

      Then at-once he forced his disciples          on ship ascend (Mk(WSCp)6.45)

      ‘Then straightaway he forced his disciples to go on board’

   2. b Godde  we scullen bihaten,  ure sunnen to beten

      to-God  we must    promise  our sins     to atone-for (Laӡ.Brut(Clg)9180)

      ‘We must promise God to atone for our sins.’

Although bare infinitives lost ground since ME, this was not exclusively due to the rise of the to-infinitive. In some contexts they were replaced by present participles, as in (20).

1. (20) than he will come renne vpon vs (Merlin,XII)

   ‘then he will come running upon us’

In PDE, the bare infinitive has a rather restricted range of occurrence. Apart from its use with the modals (see Chapter 6), it is only common after verbs of perception and some verbs of causation, as in (21).

1. 1. a Then we saw the sky burst into red (BNC)

   2. b They made her feel very sophisticated (BNC)

To-infinitives functioning as adjective complements fall into a number of interrelated types. Quirk et al. (Reference Quirk, Greenbaum, Leech and Svartvik1985: 1226) distinguish no less than seven constructions, but for present purposes we can start from the two-way classification illustrated in (22a–b). In the so-called eager-to-please construction in (22a), the main clause subject functions also as agent-argument of the infinitive. In the easy-to-please construction, the main clause subject corresponds to the patient-argument of the infinitive. The easy-to-please construction is therefore somewhat reminiscent of the modal passive, discussed above (cf. (12)), expressing a passive-like meaning without passive marking.

1. 1. a [Eager-to-please construction]

      Andropulos was a bit reluctant to go ( BNC)

   2. b [Easy-to-please construction]

      He was lying in wait in the hallway, where he was impossible to overlook ( BNC)

Note that many of the adjectives in the easy-to-please construction also appear with extraposed to-infinitives, as in (23), where the infinitive verb again functions as a regular active, in form as well as in meaning.

1. (23) [Subject extraposition]

   it is impossible to overlook the signs of squalor and decay amidst the glamor.

   (2008, COCA)

Both the eager-to-please construction and the easy-to-please construction were already available in OE, as shown in (24) and (25), respectively. And so was subject extraposition, though the dummy subject (it) was still optional (Mitchell, Reference Mitchell1985: 18–19; see Chapter 7 and examples (4) and (5) there).

1. (24) ic   eom   gearo     to     gecyrrenne to    munuclicere drohtnunge

   I    am     ready     to     turn           to     monastic      way-of-life (ÆCHom.I,35 484.251)

   ‘I am ready to turn to a monastic way of life.’

1. (25) ðis     me         is     hefi        to     donne

   this    to-me     is     heavy     to     do (Mart.5(Kotzor)Se.16,A.14)

   ‘This is hard for me to do.’

The eager-to-please construction has been essentially stable throughout the history of the language. It is interesting to note, however, that some adjectives in the construction developed into modal markers. A straightforward example is bound, originally meaning ‘under a binding obligation’ but now typically used to mark epistemic necessity. The development is illustrated by the examples in (26).

1. 1. a I requyre yow, that ye make the houses as ye be bound to repaire and kepe.

      ‘I ask you that you would mend the houses that you are bound to repair and maintain.’ (a1475, Reg.Godst.Nunnery,471)

   2. b Remember to keep all the topsoil you remove; you’re bound to find a use for it later! (BNC)

On these grounds, expressions like be bound to could be regarded as semi-modals – hence auxiliaries (see Chapter 6) – rather than main clause predicates. Other examples include be sure to or be certain to, as in (27).

1. 1. a If … they shoot at our searchlight they are pretty sure to miss us. (1909, COHA)

   2. b whatever it is one has planned to do is certain to be altered in the process (BNC)

The easy-to-please construction has undergone a number of changes. From around 1400, two slightly more complex variants of the construction are found. One has a stranded preposition in the subordinate clause, as in (28a).Footnote ² The other has a passive infinitive (28b).

1. 1. a þei  fond hit good and esy  to dele wiþ also

      ‘They found it good and easy to deal with as well.’ (Cursor(Trin-C)16557)

   2. b the excercise and vce [= ‘use’] of suche … visible signes […] is good and profitable to be had at certein whilis [= ‘times’] (Pecock,Represser,Ch.XX)

The development in (28b) is reminiscent of that in the modal passive (cf. (12) and (13b) earlier in this chapter), and some degree of mutual influence seems likely. Curiously, however, passive marking eventually became obligatory in the modal passive, but not so in the easy-to-please construction, where formal passives never became systematic and sometimes even disappeared again, as shown by the now-ungrammatical example in (29a). Fischer (Reference Fischer and Kastovsky1991: 175ff) suggests that formally passive infinitives tend to occur with easy-type adjectives when the relation between adjective and infinitive rather than between adjective and subject is stressed (cf. 29a). In such cases, an adverb rather than an adjective is also often found, as in (29b).

1. 1. a when once an act of dishonesty and shame has been deliberately committed, the will having been turned to evil, is difficult to be reclaimed (1839, COHA)

   2. b Jack Rapley is not easily to be knocked off his feet (1819, Fischer ibid.)

From this, one might speculate that passive forms failed to fully establish themselves in this context because the meaning the construction conveys is not always purely passive. As (30) illustrates, the subject of many easy-to-please constructions combines both patient-like and agent-like qualities. While the subject undergoes the action, its intrinsic qualities also contribute to how that action unfolds. The construction could therefore be analysed as marking middle voice.Footnote ³

1. (30) more experienced opponents … can sometimes be tricky to play against. (BNC)

8.2.2 Gerunds

Gerunds are noun clauses par excellence. They can be found occupying all the positions an ordinary NP can occupy – see the examples in (31) – and hardly any others. Arguably, the only important exception is its use as extraposed subject, as in (32).

1. 1. a [Subject]

      Finding the right person for a job can be hard work (BNC)

   2. b [Subject complement]

      Their principal duty is making sure the two little princesses are safe and well (BNC)

   3. c [Verb complement]

      Charles bitterly regretted having allowed the cameras in (BNC)

   4. d [Complement of a preposition]

      The difficulty lies in finding an acceptable implied limitation (BNC)

1. (32) [Extraposed subject]

   I knew it was pointless expecting him to change his mind. (BNC)

The gerund looks very similar to the present participle and it has been argued for PDE that the two are in fact just one clause type.Footnote ⁴ Their historical origins are clearly distinct, however. Originally, the present participle had a different ending in -end, and even following their formal collapse there has always been one syntactic feature clearly distinguishing the two: only the gerund can have its subject in the possessive form, as in (33).

1. (33) each day away from Cumbria increased the chance of his missing the conclusion of the investigation. (BNC)

The origins of the gerund lie in OE deverbal nouns in -ing/-ung, as in (34). Such nouns behaved exactly as nouns would. They could combine with adjectives and determiners, as in (34a), where the adjective gelomum modifies the deverbal noun scotungum. Any participants to the actions these nouns denoted would be expressed by a genitive, such as deofles in (34b), which functions as the agent of tyhttinge.

1. 1. a þa  wunda  þe   þa  wælhreowan hæþenan mid  gelomum scotungum on his

      the wounds that the  barbarous     heathens with frequent   shootings    on his

      lice macodon

      body made (ÆLS(Edmund)181)

      ‘the wounds that the barbarous heathens had made on his body with numerous shots’

   2. b Đa    wearð    an    þæra  twelfa   cristes     þegena   se     wæs Iudas gehaten

      Then became one of-the twelve of-Christ disciples who was Judas called

      þurh     deofles tyhttinge    beswicen.

      through devil’s persuading  deluded (ÆCHom.I,1 188.266))

      ‘Then one of Christ’s twelve disciples, the one who was called Judas, was deluded through the devil’s persuasive powers’

There have been many attempts to explain how and why nominal gerunds came to adopt verbal behaviour and ended up heading full-fledged non-finite clauses. The most comprehensive overview is provided by Jack (Reference Jack1988), who considers ten possible sources or causes. Eventually, he accepts four as plausible. First, -ing-derivation saw an increase in productivity during OE and early ME. This was a preliminary to the emergence of the verbal gerund. Second, French had a gerund – known as the gérondif – with clausal qualities. English may have copied the French construction using its forms in -ing. Third, the ME genitive sometimes coalesced with the common case, particularly in the plural (see also Chapter 5, especially Section 5.3.2). This might just explain how -ing-forms suddenly found themselves accompanied by common case verb arguments.Footnote ⁵ Fourth, verbal gerunds unlike infinitives could follow prepositions, which may account for their subsequent success in ME and ModE. After all, English could well do with a clause type that would pattern with prepositions.

Subsequent work has refined Jack’s conclusions, but mostly in the details. Kranich (Reference Kranich, Johnston, von Mengden and Thim2006) doubts the role of French. De Smet (Reference De Smet2008: 294–96), by contrast, believes French influence is plausible but points not only to the French gérondif, but also to French infinitives, which like the gerund could follow prepositions.Footnote ⁶ Additionally, Fanego (Reference Fanego2004) and De Smet (Reference De Smet2009) nuance the idea that gerunds filled a syntactic gap. It is true that the English infinitive could not combine with prepositions, but this gap in the system could also have been filled by the nominal gerund, as shown in (35). Indeed, the nominal gerund saw a spectacular increase in frequency, exactly in prepositional contexts. This happened just prior to the rise of the verbal gerund. One might therefore speculate that English first filled the gap in its system of noun clauses by exploiting nominal gerunds and then gradually switched to a better alternative in the form of the verbal gerund – the advantage of the latter being its greater syntactic flexibility.

1. (35) For þis principle of love moten men suppose, whanne þer ben two þingis put in a mannis chois, and he mut nede leve þe toon for takinge of þe toþer. (Wyclif, Bible,LX,p.182)

   ‘Because men must bear in mind this principle of love, when a man is to choose between two things, and he must abandon the one in return for taking the other.’

As verbal gerunds came to be used more and more, they also acquired more clausal features. Common-case direct objects and adverbs appeared first. Later negation appeared, as well as other clausal constituents such as indirect objects and subject complements. Finally, the gerund began to allow passive and perfect auxiliaries (Tajima, Reference Tajima1985). The examples in (36) illustrate some of the steps in this gradual adoption of clausal syntax.

1. 1. a [Gerund with direct object]

      And thus he slougħ hem bothe in doinge this orible synne. (14^thc.,Kn.of LaTour-Landry,LXII,82)

      ‘And in this way he killed them both (while they were) in (the act of) committing this terrible sin.’

   2. b [Gerund with negation]

      Al our pes … is raþer to be sette in meke suffryng þan in not feling contrarieties. (c1430, OED)

      ‘All our peace of heart is rather in being put in meek suffering than  in not feeling any opposition.’

   3. c [Gerund with perfect auxiliary]

      … whose pillow she kissed a thousand times for having borne the print of that beloved head. (1580–1, Tajima, Reference Tajima, Embleton, Joseph and Niederehe1999: 269)

In the meantime, nominal gerunds continued to be used, and there also arose various types of mixed nominal-verbal gerunds. For example, in (37) putting is preceded by a determiner the, like a noun, but takes an object the prisoners like a verb. This mixing of nominal and verbal properties has a PDE remnant in the verbal gerunds with possessive subjects illustrated in (33). Another remnant is the occasional gerund with the noun quantifier no, as in (38).

1. (37) Assisting to the sayd Commissioners for the putting the prisoners … to suche tortours as they shall think expedient. (1551, OED)

1. (38) He spoke softly, but there was no denying the impatience in his voice. (BNC)

On the whole, however, mixing between the two clause types declined. The type in (37) disappeared again, possessive subjects in the verbal gerund are being increasingly replaced by common-case subjects, and the type in (38) remains marginal. The result is that the formal distinction between nominal and verbal gerunds has grown sharper.

This formal polarization came with a growing functional differentiation. For example, Fonteyn et al. (Reference Fonteyn, De Smet and Heyvaert2015) find that during ModE verbal gerunds became less likely targets for anaphoric reference. That is, examples like (39), where the demonstrative that refers back to the verbal gerund describing the measures of his duty, became less common. In contrast, nominal gerunds became more likely targets.

1. (39) So that we cannot discourse of the man’s right, without describing the measures of his duty; that therefore follows next. (1762, Fonteyn et al., Reference Fonteyn, De Smet and Heyvaert2015: 42)

Being a potential target for anaphoric reference is a typical discourse feature of nominal referents. Therefore, this finding indicates that verbal gerunds, once established, continued to become more clause-like, not so much in their form but in how they are employed in discourse. Nominal gerunds developed in the opposite direction, reverting back to the discourse uses of ordinary NPs. Fonteyn (Reference Fonteyn2016) further shows that nominal gerunds closely match the referential subtypes found in ordinary NPs, showing functional distinctions such as generic/specific/non-specific and indefinite/definite, whereas verbal gerunds came to be increasingly dissociated from the system of nominal reference.

Although the gerund was originally most common following prepositions, its rise also had a significant impact on the English system of verb complementation, as gerunds came to be used as complements to transitive verbs as well. This development involved a slow process of lexical diffusion that continued throughout ModE and into PDE, as described in detail by De Smet (Reference De Smet2013a). The first verbs to select for gerundial complements were typically verbs that collocated with deverbal nouns. For instance, (40a) is an early instance of love with a gerund, but the gerund is still part of a series of deverbal nouns (and is therefore probably itself still a noun). From these beginnings, gerundial complement clauses developed and spread to other verbs, initially favouring some clusters of semantically related main verbs, including emotive verbs, such as love or hate, or negative implicative verbs, such as forbear, omit or avoid in (40b). As gerundial complement clauses became more productive, they eventually spread beyond these clusters. In PDE they appear with most transitive verbs that are semantically compatible with clausal complementation, such as imagine, remember, try or consider, as in (40c), with two important groups of exceptions. First, because gerunds themselves convey no temporal or modal information about the state of affairs they denote, they did not spread to main verbs that convey no such information either (e.g. think, know, say). Second, gerunds tend to be absent still with many verbs that already took to-infinitival complements before the rise of gerund complements (e.g. want, desire, wish). With such verbs, the rise of the gerund was often blocked or at least delayed by the availability of the pre-existing pattern.

1. 1. a Þe luueden tening & stale, hordom & drunken

      ‘[You] that loved doing harm and theft and whoring and drunkenness’ (early ME, PoemaMorale,253)

   2. b If … I could have avoided meddling with him, I should not desirously have begun with a Gentleman … of so … turbulent a Disposition. (1635, OED)

   3. c He considered asking for the feather back but knew the request would sound foolish. (BNC)

8.2.3 Participles

Like to-infinitives and gerunds, participles can take up a range of different positions in the clause. Participial subordinate clauses were historical sources for the perfect, passive and progressive constructions illustrated in (41a–c) respectively. In these constructions, an original clause-combining structure has become a mono-clausal one, with the participle now providing the main verb (for details, see Chapter 6). Participles also figure as noun-clause-like complements to perception verbs, as in (42) – the form has its roots in OE (cf. Mitchell, Reference Mitchell1985: 895) but the construction’s underlying syntax has probably changed subtly since.Footnote ⁷

1. 1. a Nails wasn’t quite sure what had hit him. (BNC)

   2. b Michael Sweeney was hit in the shoulder. (BNC)

   3. c I’m hitting my head repeatedly against something that isn’t there and I suppose that that could be a definition of insanity. (BNC)

1. 1. a He could see her deciding whether to agree or disagree, he could see her weigh the advantages and the possible disadvantages. (BNC)

   2. b to see her shamed like that was terrible. (BNC)

In this section, we focus on adverbial past and present participles, as illustrated in (43). As the examples show, such adverbial participle clauses can come with or without a linking element to mark their relation to the main clause, and they can come with or without an explicit subject. When their subject is unexpressed, it typically corresponds to an element in the main clause, mostly the main clause subject. Adverbial participle clauses without subject are sometimes referred to as ‘free relatives’, the ones with explicit subject as ‘absolutes’. Absolutes are said to have been borrowed from Latin, which is probably partly true, as discussed in Chapter 4 (see especially, Section 4.3.1).

1. 1. a When stopped by the car park attendant, the disturbed man shoved the attendant aside, causing him to fall and break his wrist. (BNC)

   2. b June Roberts went and touched the old silk fabric, tears filling her eyes (BNC)

In both the free relative and absolute constructions, participles alternate with other predicate types, particularly adjectives and prepositional phrases. For instance, the head of the free relative is an adjective in (44a), while the absolute in (44b) has as its head a prepositional phrase.

1. 1. a Though logical, these arguments are inaccurate (BNC)

   2. b The lassitude still on her, she came slowly to her feet (BNC)

Even though they are mainly a feature of the written language, free relatives and absolutes are remarkably common in PDE when compared to similar constructions in Dutch and German. This is particularly true of the constructions with present participles and may reflect a more general fondness of English for backgrounding and stativizing grammatical devices as well as the influence of the gerund (Van de Pol and Petré, Reference Van de Pol and Petré2015; cf. Section 6.4.1 in this volume).

Historical developments within free relatives and absolutes have received relatively little scholarly attention. Clearly, however, their use has not been stable. Focusing on free relatives with a present participle, Killie and Swan (Reference Killie and Swan2009) show that the semantic relation between main clause and free relative was more coordinate-like in ME than in PDE. In ME, the free relative just elaborated on the main clause or specified an accompanying circumstance, as in (45a). In eModE, free relatives increasingly came to mark more specific adverbial relations such as manner, time or cause, as in (45b). As the examples show, the semantic change was accompanied by a formal change. From occurring almost exclusively in clause-final position, free relatives began to occupy clause-initial position as well. Killie and Swan take these changes to reflect a tighter syntactic integration between main clause and subordinate clause.

1. 1. a Jahel … took a neyl of the tabernacle, takynge there-with an hamer

      ‘Iahel took a nail from the tent and took a hammer with it.’ (ME, Wyclif,Bible,Cap.IV)

   2. b Having no alternative, Fiver accompanied Hazel and Bigwig to the burrow where Hazel had spent the previous night. (BNC)

The history of absolutes shows no such shift. Van de Pol (Reference Van de Pol2016) shows that OE absolutes already covered the full range of adverbial relations and, if anything, came to be used increasingly with the more coordinate-like meanings of elaboration and accompanying circumstance. This suggests that with time free relatives and absolutes grew more alike.

Another area of change is the linking elements free relatives and absolutes combine with. Free relatives tend to allow adverbial subordinators, such as when, while, if and the like, as in (43a) and (44a) above, but just when and how those combinations arose is not clear. Absolutes, by contrast, typically take with as a linking element, as in (46), and they have done so increasingly over time (van de Pol and Cuyckens, Reference Van de Pol, Cuyckens, Ramat, Mauri and Molinelli2013). The source of this use probably lies in prepositional phrases like (47), where the relation between a nominal head and its postmodifying dependent can be reinterpreted as one between a subject and a predicate (see also Section 4.4.1).Footnote ⁸

1. (46) I stand up and brush my coat down just in case it got dirty with me sitting on the ground. (BNC)

1. (47) Heo … feng to þonki þus godd wið honden up aheuene.

   ‘She thus fell to thanking God, with her hands lifted up.’ (early ME, St.Juliana,60)

Linking elements other than with occasionally occurred, too, as in (48).

1. (48) Wherupon the duke sent him a lettre of defiaunce, and called Paulmer, who after denial made of his declaracion was let goe (1550–2,Edward VI,Diary)

Perhaps these point to interference from gerund clauses. Although marginally attested in OE, the bulk of examples occurred in eModE, when the gerund was on the rise (Visser, Reference Visser1963: 1158, 1271–78). Recall that gerunds also combined with prepositions, could also have subjects, and just like an important subset of absolute constructions they would have a form in -ing (Section 8.3.2). The line between gerunds and present participles is particularly thin in non-finite clauses introduced by what with, whose convoluted history is discussed by Trousdale (Reference Trousdale2012).

8.3.1 That-clauses

PDE that-clauses can occupy a range of positions, as shown in (50), functioning as straightforward noun clauses most of the time, though (50c) and (50e) are significant exceptions. Their distribution is similar to that of to-infinitives (compare Section 8.2.1).

1. 1. a [Subject ]

      That he came at all was surprising; that he came without a chainsaw was amazing. (Google)

   2. b [Subject complement]

      The problem is that they clog up. (BNC)

   3. c [Noun complement]

      And then there were hats for every occasion, sunglasses, sportswear and even the questionable piece of advice that a firm corset would prevent seasickness. (BNC)

   4. d [Verb complement]

      I didn’t realize that he was boss. (BNC)

   5. e [Adjective complement]

      Only when we are sure that all possible test factors have been controlled can we feel confident that we understand the causal process at work. (BNC)

That-clauses, typically with a subjunctive verb, were a very common feature of OE, too. But, while their distribution was more or less similar to that of PDE that-clauses, there were some differences, largely purporting to OE þæt-clauses being a little less ‘noun-clause-like’ than the PDE that-clause. First, the construction type in (50a) did not yet occur in OE – its appearance is discussed below. Second, OE þæt was also frequently used to introduce adverbial purpose clauses, as in (51), making it sometimes hard to decide whether a given þæt-clause was an adverbial clause or a noun clause. Þæt was also frequently part of complex conjunctive phrases introducing adverbial clauses (see Section 8.3.2).

1. (51) he wæs mid wacum cildclaðum         bewæfed, Ϸæt he us forgeafe ϸa

   he was with poor    swaddling-cloths covered   that he us bestow   the

   undeadlican tunecan þe  we forluron on ϸæs  frumsceapenan mannes

   immortal      coat      that we lost       on of-the first-created man’s

   forgægednesse

   transgression (ÆCHom.I,2 193.88)

   ‘He was wrapped in poor pieces of cloth so that he might return to us the immortal coat that we had lost upon the transgression of the first man.’

When it comes to the origin of that-clauses, it is hard to ignore the homonymy between the OE conjunction þæt and the demonstrative pronoun – or, for that matter, the relative pronoun (see Section 5.5.3). It is reasonable to assume that the conjunction somehow developed from the demonstrative pronoun (Traugott, Reference Traugott1992: 237), but how exactly that happened is unclear. The most common account states that the conjunction must have been reanalysed from an anticipatory demonstrative in a paratactic structure ([we heard that]_S1 [the king was dead]_S2 > [we heard [that the king was dead]]_S) (Heine and Kuteva, Reference Heine and Kuteva2007: 241). The best evidence to support this account is that OE þæt could indeed function as an anticipatory demonstrative pronoun, as seen in (52).

1. (52) Wæs    þæt    eac   gedefen,   þætte   þæt     swefn   gefylled     wære, þætte …

   was     that    also   proper      that     that     dream   fulfilled     were   that … (Bede.4.24.336.28)

   ‘It was also proper that that dream was fulfilled that…’

Even so, the reanalysis-based account is not entirely convincing. In (52), the pronoun þæt is not actually in a position that would allow reanalysis as a conjunction, and in fact such contexts are unattested. Moreover, the account leaves various issues unresolved. One puzzle is the role of the OE subordinating particle þe, which was often used to mark subordinate clauses and is sometimes found accompanying þæt, mostly contracted to þætte, as in (53). Did the combination þæt þe only arise after þæt had developed into a conjunction, or had it been an ancestor to the conjunction þæt? Even though Mitchell (Reference Mitchell1985: 433) inclines towards the former, the latter is not completely implausible. For what it is worth, while Gothic did not use its demonstrative pronoun þata as a conjunction, it did use þatei – a contraction of þata and the subordinating particle ei (van der Horst, Reference Van der Horst2008: 281–82).

1. (53) Ic wene,  cwæð Orosius, þæt nan wis  mon  ne sie, buton  he genoh    geare   wite

   I believe said   Orosius that no  wise man not is  but     he enough clearly knows

   þætte God     þone  æresetan  monn    ryhtne     godne    gesceop

   that   God   the     first        man     just [and] good     created (Or.2,1.35.28)

   ‘I believe, Orosius said, that there isn’t a wise man whodoes not clearly know that God created the first man just and good.’

Another puzzle is the role of adverbial þæt-clauses and the various other uses of þæt where it did not introduce a noun clause. If the conjunction developed directly from the demonstrative pronoun, its first use would have been as a marker of noun clauses, so adverbial and other uses would have been derived. That is what Mitchell (Reference Mitchell1985: 432–33) suggests, and it is consistent with how Gothic þatei was used. However, the opposite direction of change – from adverbial clause to noun clause – seems much better-attested.Footnote ⁹ In all, there is much about the prehistory of þæt and, consequently, about its distribution in OE that is still poorly understood.

Less speculative is what happened to that-clauses during and after OE. The decline of the subjunctive had consequences for that-clauses, too. Sometimes the modals substituted for the old subjunctive forms, or subjunctives were replaced by indicatives (see Chapter 6). But in other cases, the that-clause lost ground to the to-infinitive (see Section 8.2.1).

In one respect, the distribution of that-clauses expanded. That-clauses functioning as a subject in preverbal position, as in (50a), first appeared in late ME. The reasons for their non-occurrence in OE and early ME are not clear but may have to do with the fact that subject clauses in initial position are difficult to process. Even today, they are largely restricted to writing, while the preferred option, especially in the spoken language, is to use the dummy subject it with the subordinate clause in extraposed position, as in (54) (Biber et al., Reference Biber, Johansson, Leech, Conrad and Finegan1999: 676).

1. (54) It was significant that it was the Consul General who spoke first. (BNC)

This option also existed in OE, mainly in impersonal and passive constructions, as in (55a). The alternative was a construction without it and the þæt-clause in final position, as in (55b). There is some scholarly debate as to whether the þæt-clauses in such examples were subjects or not (see Denison, Reference Denison1993: 61–102, for arguments for and against).

1. 1. a Eac hit awriten is, ðæt  sunne aþystrað ær     worulde ende

      also it   written is that  sun    darkens  before world’s end (WHom.3.41)

      ‘It is also written that the sun will be eclipsed before the world’s end.’

   2. b Næs      nanum men forgifen þæt he moste habban … his agen fulluht

      ‘not-was no      man granted  that he might have           his own  baptism

      buton Iohanne anum

      but   John     alone (ÆCHom.II,3 25.206)

      ‘It was granted to no one to perform his own baptism except to John’

The most remarkable change in the that-clause is the increased rate of omission of the conjunction that. Finite noun clauses without that have always existed but had been uncommon until eModE. A rare OE example is given in (56).

1. (56) and cwæð he wolde wiðsacan his Criste

   and said   he wanted forsake   his Christ (ÆLS (Basil) 371)

   ‘… and said (that) he wanted to forsake Christ.’

It is perhaps no coincidence that the subordinate clause in (56) represents reported speech (as do similar examples in Mitchell, Reference Mitchell1985: 30–31). As Otsu (Reference Otsu2002) points out, the origins of that-omission could well be due to influence from direct speech or thought representation, where that has always been virtually absent, as in (57). Confusion between indirect and direct reported speech was probably further promoted by the loss of word-order differences between main and subordinate clauses (Fischer, Reference Fischer2007: 304).

1. (57) Eft he cuæð be ðæm ilcan: Đonne ic wæs mid Iudeum ic wæs suelc hie

   ‘Again he said in the same [manner]: when I was with [the] Jews I was like them’ (CP.16.101. 5)

From eModE onwards, the incidence of that-omission increased dramatically. The rise of that-omission appears to have proceeded faster with main clauses that had a first person subject or an epistemic main verb, such as I think or I believe. Thompson and Mulac (Reference Thompson and Mulac1991) link this to a change in the status of the main clause itself. They argue that that-omission occurs so frequently with specific main clauses because those main clauses were grammaticalizing into modal particles marking speakers’ epistemic stance. As a result, the dependent noun clauses became de facto main clauses and no longer needed subordinate clause marking. Thompson and Mulac corroborate their argument with the positional flexibility of I think and similar main clause chunks, as illustrated in (58). Once main clauses like I think lost their main clause status, they argue, there no longer was any reason for them to occur only in sentence-initial position.

1. (58) Many would regard him, I think, as an eligible bachelor. (BNC)

Another indication that some main clauses began to be treated as modal particles is the increasing incidence of so-called negative raising. In negative raising, a negator is formally part of the main clause, but semantically has scope over the subordinate clause. In (59), for instance, I don’t think we should does not mean ‘it is not my opinion that we should’ but conveys the stronger meaning ‘it is my opinion that we should not’.

1. (59) ‘Dominic – why don’t you take Lee to the pub again, then come and see me later?’ ‘Too risky, Thea. You’re the branch secretary. Everyone knows you here.’ ‘I meant really late. Midnight.’ ‘I don’t think we should. Not in this place.’ (BNC)

In earlier stages, negative raising was uncommon (Denison, Reference Denison and Romaine1998). If a subordinate clause contained a negative proposition, it also contained the negative marker, as in (60).

1. (60) I think there is not a more necessary and useful thing to be thought on. (1665, Thomas Cock, Hygieine)

Negative raising is reminiscent of the behaviour of negation with some of the modals, which are similarly ‘transparent’ to sentential negation. For example, in (61) not does not negate the modal force of must (‘there is no obligation for letter post items to exceed 610mm in length’) but negates the modalized proposition (‘there is an obligation for letter post items not to exceed 610mm in length’).

1. (61) Letter post items must not exceed 610mm in length (BNC)

As such, negative raising supports the idea that some main clauses have been developing into modal particles.

All this makes Thompson and Mulac’s (Reference Thompson and Mulac1991) account of that-omission quite attractive. Nevertheless, there is a complication. Thompson and Mulac imply that positionally mobile parenthetical chunks, as in (58), arose as a result of that-omission, but that is implausible, because the parentheticals appeared well before that-omission became common (Brinton, Reference Brinton1996: 212) – witness the ME example in (59). Brinton (Reference Brinton1996) and Fischer (Reference Fischer2007: 305) therefore suggest alternative sources for the parenthetical uses.

1. (62) Of twenty yeer of age he was, I gesse. (Chaucer,Gen.Pr..82))

   ‘He was twenty years old, I guess.’

It seems then that the increase in that-omission, the rise of negative raising and the rise of positionally mobile parentheticals were developments that started out more or less independently but subsequently converged.

8.3.2 Adverbial Clauses

Finite adverbial clauses can be marked by a variety of subordinating conjunctions and it is with respect to these that the main changes in this area of the grammar occurred.Footnote ¹⁰ The most striking development is probably the expansion of the inventory of subordinators over time, described by Kortmann (Reference Kortmann1997). There is in PDE a core of high-frequency monomorphemic subordinators – as, when, if, where, because, while, before, since, after, until and so on (cf. Kortmann, Reference Kortmann1997: 131) – which, from a historical point of view, is both old and relatively stable. The subordinators that make up this core already functioned as subordinators in OE (as, if, while, since) or else came to do so no later than ME (when, where, because, before, after, until). Remarkably, they also tend to be the subordinators that are first acquired by children. Outside this core, however, fluctuations in the subordinator inventory are more pronounced. Throughout the history of English, the trend has been for the inventory of subordinators to grow. PDE has almost twice as many subordinators as OE, though eModE had even more (Kortmann, Reference Kortmann1997: 294). At the same time, many new additions – especially, it appears, the ones that entered the grammar in eModE – again disappeared, witness the subordinators in (63), all of which are now obsolete.

1. 1. a The Lyons … brake all their bones in pieces or euer [= ‘well before’] they came at the bottome of the den. (1611, OED)

   2. b Such of them as … had a desire to stay in Spain … were suffered to do so … conditioned, that [= ‘provided that’] they would be Christened. (1622–62, OED)

   3. c The Parts of Musick are in all but four, howsoever [= ‘even though’] some skilful Musicians have composed songs of twenty […] parts. (1674, OED)

English also saw changes in the formal make-up of its subordinators. Apart from its monomorphemic subordinators, OE made frequent use of complex subordinators built on the basis of a case-marked demonstrative, often introduced by a preposition and followed by þæt or the subordinating particle þe, as in (64a), or an adverb followed by þe, as in (64b). In some cases, these complex subordinators alternated with a reduced form, probably reflecting ongoing grammaticalization (e.g. ær þæm þe ‘before’ alternated with ær). This construction was more or less lost in the course of ME.

1. 1. a Mid  þæm þe  þa ærendracan to Rome comon,  þa   com   eac  mid  him…

      with that that the messengers to Rome came,   then came also with them(Or.3,5.58.8)

      ‘When the messengers came to Rome, there came also with them…’

   2. b God    wolde     geswutelian   þurh      þæt   syllice        tacn  þæt    his   sawl

      God   wanted   show          through that  wonderful sign  that   his  soul

      leofode  þeah þe    se      lichama  wære     ofslagen

      lived     though     the     body      were      killed (ÆLS(Denis)306)

      ‘Through that wonderful sign God wanted to show that his soul lived even if his body was destroyed.’

Another typically OE strategy was the use of correlative adverbs, where main and subordinate clause would each be introduced by one of a pair of identical or related adverbs, as in (65). In such cases, which clause was the main clause and which was the subordinate one would often be clear only from word-order differences. In principle, the main clause was VSO (due to the application of verb-second following the introductory adverb) and the subordinate clause SOV, as in (65a) (cf. Chapter 9). But this was no hard-and-fast rule, as shown by (65b), which has VS-order twice.Footnote ¹¹ This correlative pattern, too, largely disappeared in the course of ME.

1. 1. a Þa     ða     Ioseph  þis  smeade:     þa     com     him to godes engel.

      Then  then  Joseph   this considered  then   came   him to God’s angel (ÆCHom.I,13,284.89)

      ‘When Joseph was considering this, God’s angel came to him.’

   2. b Nu hæbbe we awriten þære Asian suþdæl, nu wille we fon to hire norðdæle

      Now have we written of-the of-Asia south-part, now will we take to her north-part

      (Or.1,1.11.25)

      ‘Now that we have described the southern part of Asia, we will continue with the northern part.’

In contrast, from ME onwards, adverbial subordinators began to be increasingly formed on the basis of verbs, such as save in (66a), and nouns introduced by a preposition, such as by cause in (66b) (Kortmann, Reference Kortmann1997: 305). Other examples include considering, excepting, provided, suppose or for fear, in case, on condition.

1. 1. a Te deum was oure song, and nothyng elles, / Save that to crist I seyde an orison, / Thankynge hym of his revelacion. (Chaucer,Sum,T.1866–8)

      ‘Te Deum was our song and nothing else, except that I said a prayer to Christ, thanking him for his revelation.’

   2. b For though that absolon be wood or wrooth, By cause that he fer was from hire sight, This nye nicholas stood in his light. (Chaucer,Mil.T.3394–6))

      ‘For however insane or angry this Absolon might be, because he was far from her sight, the nearby Nicholas stood in his light (i.e. outshone  him).’

Having developed from a lexical head with a dependent that-clause, these new subordinators typically continued to combine with that, causing an increase in the use of that as a marker of subordination. For a while, that even spread to subordinators it had not combined with before, such as if in (67).

1. (67) He myghte wel, if that he bar hym lowe, Lyue in atthenes everemoore unknowe. (Chaucer,Kn.T.1405–6)

   ‘He could well live in Athens virtually unknown, provided he behaved humbly.’

Perhaps this extension of that was partly in response to the loss of distinctive word order in subordinate clauses. Nevertheless, in ModE and PDE that is again increasingly omitted, echoing the behaviour of that in noun clauses (see Section 8.3.1), but probably also reflecting the progressive grammaticalization of the new subordinators.

Apart from the changing composition of the subordinator inventory, specific subordinators saw change too. As subordinators grammaticalize, they often undergo formal reduction and coalescence (e.g. OE eal(l)swa > ME as or ME by cause (that) > ModE because). At the same time, the semantic relation they establish between main and subordinate clause may change, sometimes along more or less predictable pathways. A well-known example is the conjunction while (Traugott and König, Reference Traugott and König1991: 199–201). Originally, while marked simultaneity, as in (68a). This often came with the implication that there was something remarkable about two situations occurring simultaneously, as in (68b) (it is unusual for an intruder to get in just as the gatekeepers are talking about him). Because of this association with unexpectedness, while developed into a concessive subordinator, meaning ‘although, whereas’, as in (68c).

1. 1. a & ðat  lastede þa xix wintre wile Stephne was king. (ChronE(Irvine)1137.32) )

      ‘and that lasted then nineteen winters during-the-time-when Stephen was king.’

   2. b ‘By my othe,’ said a nother, ‘it myghte well be mawgys, that is soo dysguysed for to dysceyve vs’ / ‘It is not soo,’ sayd thother / ‘mawgys is not a live’ / And while that they devysed thus togyder, mawgys cam nere to the wycket of the gate, and founde the meanes that he gate in anone (Caxton,Sons-of-Aymon,XXI,462)

      “I swear,’ said the one, ‘it might well be Mawgys, who is disguised in this way to deceive us.’ ‘It is not so,’ said the other, ‘Mawgys is not alive.’ And while they were conversing like this, Mawgys approached the opening in the gate and found the way to get in at once.’

   3. c While Adorno confines this category to ‘serious’ music, Paddison points out that there seems no reason why it could not include a good deal of avant-garde jazz … (BNC).

Another kind of change that sometimes affects adverbial subordinators is that they come to introduce complement clauses, including noun clauses (López-Couso and Méndez-Naya, Reference López-Couso and Méndez-Naya2015). A recent example is in case – originally an adverbial subordinator introducing conditional clauses, but now also marginally used as a ‘complementizer’ introducing complement clauses with some verbal and adjectival predicates. In (69), in case does not introduce the condition under which the subject would have been worried, but the possible state of affairs the subject was worried about.

1. (69) Oh good. We were worried in case you couldn’t get the band together. (BNC)