THE FUTURE OF FOR TO

Elly van Gelderen, Arizona State University

In this paper, I compare the use of the for to infinitive in several stages of Middle English. I will show that both for and to expand their grammatical functions from prepositions to conjunctions and extend their meaning from location to future and non-tense as well. Processes where grammatical function increases are generally referred to as grammaticalization or desemantization (cf. Heine et al. 1991; Traugott 1991; Bybee et al. 1994). My aim is to trace what syntactic effects accompany these changes. I will argue that, by the end of the fourteenth century, for and to occupy different structural positions than they do in Old and Early Middle English. Thus, for to first immediately precedes the VP (to be made more precise); then, it occupies C; and finally, to is placed in a position separate of V and C while for remains in C. In Middle English, they come to introduce infinitival complements and adjuncts and, in these contexts, they mark future and non-tense. In this paper, I focus on changes connected with for to and for and phrase these changes in a Minimalist way (Chomsky 1992; 1995), namely, that grammaticalization changes the intrinsic features of for to from Case (related) to future. Once that change occurs, for to changes its position accordingly from P to C, with possibly an intermediate stage in AGRo.

The Middle English texts used in this paper are two versions of Layamon's Brut, Gawain and the Green Knight, The Alliterative Morte d'Arthur, Langland's Piers Plowman (the B-version), Chaucer's Canterbury Tales, Gower's Confessio Amantis, The Chancery Letters and The Paston Letters. The earlier version of Layamon is often dated as early to middle thirteenth century; the later version is said to be from around 1275. Gawain is probably from the middle of the fourteenth century; the Morte is from after 1360; and Piers is from the 1370s. Gower's Confessio and Chaucer's The Canterbury Tales are from the latter part of the fourteenth century; The Chancery Letters and The Paston Letters were written throughout the 15th century. The paper is organized as follows. In section 1, I start by outlining the modern English situation where for and to are concerned. In section 2, I present some data on for to in Old English, mainly on the basis of Beowulf (composed c700 but written down c1000), The Junius Manuscript, The Exeter Book and The Vercelli Book (all from around c1000). Then, in sections 3 to 6, I provide a description of Middle English environments in which for to appears as well as an analysis in terms of positions and features. In section 7, I add some observations about the decrease in use of for to by Early Modern English on the basis of Mulcaster, Shakespeare and Bacon and by eighteenth century English on the basis of Swift and Hume. It is hard to argue that one stage of Middle English changes into another stage. Except perhaps in the case of Layamon, the dialects and dates vary. The different stages can therefore also be examined synchronically, rather than diachronically, indicating possible variation. That is the reason that, in section 8, I end by providing examples of similar shifts of `for to' in other varieties and languages.

1. Modern `standard' English: for and to

In Modern English, for is a complementizer as in (1) and (2), even though in most variants it only introduces a tenseless clause as in (2). Sentence (3) indicates that once for is used, the perfect in the subordinate is future perfect. The verbs that use for-clauses are such as like and want, indicating a desire on the part of the speaker. These will be referred to as deontic verbs (cf. Lyons 1977: 823ff). Sentence (4) indicates that, when the for infinitive is in subject position, the tense of the infinitive is future:

1. He laughed for he felt happy.

2. I would like for you to read the following sentences.

3. I would like for you to have finished the book by tomorrow.

4. For him to go there is stupid.

For in (2) introduces an action/event that must be future with respect to the action/event in the main clause. I take this to mean that the tense of the infinitive is determined by for. The perfect in (3) is perfect with a point in the future as reference point, i.e. future perfect.

There are verbs, such as believe and know, that cannot be complemented by for and their tense properties are very different. These are called epistemic verbs and their complements indicate the belief or knowledge of the speaker:

5. I believe (*for) Zora to be nice.

6. I believe her to have lived a happy life.

In (5) and (6), the action of the infinitive must be at the same time as that of the verb in the main clause. This is expected if the infinitive can only have believe as a reference point. Enç (1987) calls this reference point an `anchor'. Thus, the tense of the infinitive in (5) needs to be understood and the only `anchor' is the tense of the verb in the main clause. In (2), on the other hand, there is a closer `anchor', namely for, and that is why the reference point for the embedded infinitive is future; in (4), the only `anchor' is for and the infinitive has future sense.

Above, I have been vague about what part of the embedded infinitive is `anchored', but now I assume to to be the relevant element. Thus, for and to play a role in tense marking. The reason behind this assumption is that when neither for nor to are present, the action expressed by the subordinate verb is not seen in terms of a separate action but must be seen as realized, as in (7) and (8):

7. I hear Zora leave.

8. I made Zora leave.

In (7) and (8), to is not present and the leaving cannot later be denied by the speaker whereas in (6) the having lived happily can be. So, to in sentences such as (5) and (6) contributes a sense of uncertainty. In fact, to indicates a separate action and this can be expressed through its (non)-tense features.

As to the position of the two elements, it is most often assumed that for occupies the C position because it precedes the subject and that to is in a position separate from the VP (cf. Akmajian, Steele & Wasow 1979). A structural representation is provided in (9), abstracting away from the AGRo position (Agreement with the object). The reasons for assuming that to is separate are that the VP can be deleted without to deleting, as in (10), and that the VP can be moved without taking to along, as in (11):

9. INSERT FIGURE

10. Zora tried to finish the book and Dora tried to also.

11. and [finish the book] I tried to.

In a Minimalist framework (Chomsky 1992; 1995), functional categories such as I and C contain (categorial) features which ensure that the Case and agreement features of lexical items such as nouns and verbs are checked (on heads through adjunction and on maximal projection through Specifier-Head Agreement). The evidence for the presence of functional categories is that elements move to certain positions which can be seen as I, C, Spec IP and Spec CP. On the basis of (2) and (3), I assume that for can be attached to interpretable, intrinsic future features which are relevant for the Interpretation of the sentence at LF (another use of for involves Case marking, and for is therefore also attached to intrinsic non-interpretable Case-features, i.e. when it is a Preposition). To in (5) and (6) is connected with non-tense in Modern English, but here I do not examine to in Middle English (cf. van Gelderen 1993; Fischer 1995).

I first present some Old English data on the use of for and the non-use of for to and then discuss this data in terms of grammaticalization.

In Beowulf, for is used 41 times; fore 4 times; forþan or forðon 8 times as in (14). The meaning related to for is location as in (12), purpose as in (13), and cause as in (14). Fore is used similarly. The meaning of forþan is purpose or cause as in (15):

12. Beowulf, 358

þæt he for eaxlum 3estod

that he before shoulder stepped

`He stood face to face'.

13. Beowulf, 457-8

for werefyhtum ...

ond for arstafum usic sohtest

for fighting ... and for support (you) us sought

`you sought us for the purpose of fighting and for support'.

14. Beowulf, 169

No he þone 3ifstol 3retan moste

maþðum for metode ...

not he the throne greet must

treasure for fate ..

15. Beowulf, 679

forþan ic hine sweorde swebban nelle

therefore I him sword kill not want

`Therefore I do not want to kill him with my sword'.

In Junius and Exeter, similar constructions occur, but the syntactic freedom of for increases in that it also moves along with a wh-element as in (16):

16. Junius, Genesis 873-4

for hwon secest ðu sceade sceomiende?

for why seek you shadow shaming

`why do you walk ashamed in darkness'?

There are also more forþons relative to fors (46:90 and 76:81 respectively) than in Beowulf (8:41). The use of fore increases dramatically in Exeter (and Vercelli) but not in Junius: 102 in Exeter, 30 in Vercelli, but only 3 in Junius and 4 in Beowulf. Vercelli also shows an increase in the use of forþons (34:54) but does not have for fronted as in (16). Taking Beowulf to be an older sample of Old English, one might argue that the syntactic functions of for widens: forþon `therefore' with a causal or purposive connotation introduces clauses and, in that sense, becomes a conjunction; for does not yet function that way.

Thus, in Old English, for is a preposition with multiple meanings: `in front of' as in (12), `for the purpose of' as in (13) and `because of' as in (14). Often the basic meaning of a preposition is seen as the locational one. If one adheres to a theory of grammaticalization as in, for instance, Heine et al. (1991), the meaning of for is generalized to future by metaphoric extension of the locational meaning, i.e. `in front of', to the temporal one, i.e. future and purpose. As Sweetser (1988: 393) argues, "[i]n addition to the mapping of space onto time ..., there is a mapping of spatial motion onto the domain of intentional actions". In this way, location, purpose and benefit are linked. With respect to the extension to cause marker, Heine et al. (1991: 166) say the following: "a sequence of events in time is used metaphorically to refer to a sequence of events in a causal relationship" (see also Traugott & König (1991: 194ff). The preposition for in Old English can be seen as a grammaticalized element because of its extended meaning. Yet, it is not a conjunction introducing clauses; forþon is used for that. The earliest instance in the Oxford English Dictionary of for as a conjunction is from 1200 and the second earliest is from Layamon (cf. (18) below).

The first instance of forto as a preposition that the Oxford English Dictionary lists is from 1200. However, under the entry of for, there is an earlier instance introducing an infinitive, namely from 1175 as in (17). Visser (1966: 1001) mentions 2 earlier, late Old English, ones:

17. Cotton Homilies 221

Forte don him understonden

`In order to make him understand'.

Example (17) is Early Middle English and the next entries list Layamon as a source. I therefore assume the early part of the 13th century to be the period where for to is introduced as a preposition as well as a conjunction. This period is examined in detail in the next section.

In the Caligula version, the metaphorical extension is evident in that for to introduces purposive infinitives, as in (18) to (21). This occurs seven times. No element intervenes between for to and the verb (hence, (22) is unattested). For is never separate from to (as indicated in (23)) even though for frequently occurs by itself introducing adverbial (mainly causal) finite clauses and as a preposition as in (24) and (25) respectively. As mentioned above, these are among the earliest instances of this use. For is not attested as introducing a bare infinitive, which is indicated in (26). Forto is encountered once as one word in (27), where it introduces an infinitival complement:

18. Cal. 1113-4

Locrin 7 Camber to þon scipen comen.

for to habben al þa æhte,

Locrin and Camber to the ship came for to have all the property

`Locrin and Camber came to the ship in order to have all the property'.

19. Cal. 4074-5

one man [...]

þat was þider icome. for to se þi(s) cnihtes game,

that was there come. for to see this knight's game

`that had come in order to see this knight's game'.

20. Cal. 6724

Þe king me bi-tahte þis ard for to beon his stiward,

the king me granted this land for to be his steward

`The king granted me this land for the purpose of being his steward'.

21. Cal. 7803

þat he ilad weore. limen for to leosen,

that he led was. limbs for to lose

`that he was led, with the purpose of suffering the loss of limbs'.

22. $for to hire see.

23. $for hire to see.

24. Cal. 76

for he nefde nenne sune.

because he not-had no son

`because he didn't have a son'.

25. Cal. 32-5

þat he þeos soðfeste word. segge to-sumne.

for his fader saule,

`that he said those truthful words for the benefit of his father's soul'.

26. $for see hire.

27. Cal. 12767

agan ich forto slepe,

`started I to sleep'.

For to is not used to introduce an infinitival subject, but then, as has often been discussed, the to-infinitive as subject is absent from Old English and rare in Middle English (Visser 1966: 952). If, as I argue in this paper, the (for)to connected with the infinitive in Old and early Middle English is a preposition, its absence in subject position is due to the failure of PPs to serve as subjects (for reasons I do not go into here).

As to the structural position of for to, sentence (21) where the infinitival object precedes the for to and the absence of (22) indicate that for to is closely connected to the VP. It has been assumed that infinitives are nouns in Old and Middle English, for instance, Visser (1966) and Lightfoot (1979: 191ff). The latter argues that infinitives are NPs because they have Case endings (-enne) and can be preceded by prepositions. The data in van Gelderen (1993: 104-6) indicate that all infinitives following forto in Caligula have an -en(e) ending. Jack's data (1991: 313) confirm this. He shows that infinitives with an ending are always preceded by forto or to in the extensive Early Middle English corpus he examines. I assume that in Caligula, the infinitive is a noun and that for to assigns Case to it as in (28):

28. INSERT FIGURE

Sentence (21) is an exception in that the object limen `limbs' precedes for to. It is the only example and may be derived through topicalization of the object.

The meaning related to for to is to express that the action of the infinitive is in the future and that the action is separate from that of the main clause. Thus, for to has these features independent of a functional category. I assume they are intrinsic to for to at that point in Middle English. As for for in (24), this is a preposition as well (as e.g. Lightfoot 1979: 195 argues it is throughout) indicating purpose and cause. The evidence for this is (a) that there are many for ifs as in (29) where if is a complementizer and (b) that for introduces a finite but not an infinitival clause or a nominal as in (28). This is typical of prepositions:

29. Cal. 483

For 3if we hit 3eorneð to wonien her mid Gricken,

`Because if we it yearn to live here amidst the Greeks'.

In conclusion, in the Early Middle English Caligula version, for is a preposition and a conjunction indicating purpose and causation; for to has a similar meaning but is still prepositional (and connected with Case) as in (28).

In the Otho version, there is an increase in the number of times for to introduces infinitives: from 7 times in Caligula to 22 times and even though most are purposive adjuncts as in (30), (31) is an example of a complement infinitive (albeit only one) with an independent tense. Five of the 22 are separated from the infinitive by an object. They are listed in (32) to (36). However, in 2 instances an object precedes a for to infinitive, as in (37). For occurs without to not only to introduce adverbial finite clauses as in (38) and as a preposition as in (39) but also to introduce bare (purposive) infinitivals as in (40). Forte as in (41) occurs mainly to introduce finite clauses, has a temporal meaning and can be compared with the modern English preposition until which introduces finite clauses:

30. Otho, 5873

woche faire wodes. ine for to wonie.

`which fair woods to live in'.

31. Otho, 5523

moche he lofde echn(e) cniht. þat lofde for to segg(e) riht,

`Much he loved every knight who loved to say the truth'.

32. Otho, 5496

for to him reade,

`in order to counsel him'.

33. Otho, 6419

for to heom kepe,

`in order to keep him'.

34. Otho, 6915

fo[r] to londes seche,

`in order to seek countries'.

35. Otho, 8488-90

wide his men sende. for to hine finde,

`Wide his men (he) sent in order to find him'.

36. Otho, 8570

for to worch makie,

`in order to make work'.

37. Otho, 3123

And lawes he sette stronge. his folk for to stewe,

`and he made strong laws to constrain his people'.

38. Otho, 249

Wide he sende ouer al þat lond. for he was leoden king,

`Wide he sent over all the country because he was the people's king'.

39. Otho, 33

and bidde for þe saule. þat hine to manne strende,

`and pray for the soul that him into man procreated' (i.e. his father)

40. Otho, 713

Corineus was to wode ivare. for hunti deor wilde,

`C. had gone to the wood, in order to hunt wild animals'.

41. Otho 3770

Alle dai was þat fiht; forte hit were dorcke niþt.

`All day lasted that fight until it was dark night'.

42. $For hire (to) hunt.

It has been argued in van Gelderen (1996) that Case marking is changed from theta-related as in Old English to structural in Otho. Assuming as in Chomsky (1995) that all structural Case is checked in an AGR-projection (either AGRs or AGRo), AGRoP is triggered in the language of Otho. AGRoP is introduced in Kayne (1989) to account for past participle agreement and is taken by Chomsky (1992) to be responsible for the Case of the object. A second change one can argue for is that infinitives are becoming verbal and that for to ceases to assign Case. For instance, the infinitival endings have disappeared in Otho when the infinitive is preceded by for to. On the basis of these changes, I suggest a structure as in (43). The changes are that a PP is no longer triggered but that an AGRoP is, and that for to is reanalyzed as being in C. The object of the infinitive checks its Case in Spec AGRoP and the infinitive is no longer Case marked but assigns Case to the object in Spec AGRoP:

43. INSERT TREE

For to is in C in (32) to (36) and (30) and (31). Sentences such as (37) are rare in Otho and may be `misanalyses', where for to is still seen as a Case marker (no longer of the infinitive but of the object of the infinitive). A reason for this may be the following. When the infinitive is nominal, (for)to assigns Case but as the infinitive (for reasons unknown) becomes verbal, for to is used to `assign' Case to the infinitival object until the infinitive is completely reanalysed as verbal and moves to AGRo itself. Thus, (37) represents an intermediate stage where for to is in AGRo as in (10) and through Spec Head Agreement checks the Case of the object:

44. INSERT TREE

For as in (38) is used much less frequently in Otho than in Caligula, but it is used with a bare infinitive as in (40) to introduce adjuncts. There is some evidence that for is a preposition when used to introduce a finite clause (e.g. because for 3if-clauses as in (45) occur) but there is also evidence that for is moving towards becoming a conjunction because it introduces infinitival clauses as in (40). Forte in (41) is mainly a preposition because it is followed by þat as in (46) and mostly introduces finite clauses, as shown in (41):

45. Otho, 483

For 3if we here 3erneþ. wonie mid Greckes,

`because if we yearn to live among the Greek'.

46. Otho 5746

forte þat he com. to Maximian to Rome,

`until that he came to M to R'.

As to the features associated with for to, these are future in (31) because segge `say' is future from the point of view of the event in the main clause, i.e. lofde `loved'. In Modern English, love has a complement where the action of the infinitive is future with respect to the form of love, as (47) indicates:

47. I would love for him to be elected.

This means that the tense of the infinitive in (47) is anchored in C and not in the main verb (cf. Enç 1987 and also Stowell 1982). However, in (31), the action of the infinitive may be simultaneous with that of the action in the main clause, like forte in (27). There are too few instances in Layamon to decide whether or not this is the normal situation and hence whether or not it differs from Modern English use. However, the purposive use increases and therefore it is possible to say that for (as in Modern English) is associated with future features and that it is then reanalysed as a complementizer, as shown in (43).

Thus, there are several differences between Caligula and Otho indicative of further change of for and to. As to the position, for to in Otho no longer exclusively precedes the VP but it is also in C. This is to be expected if, as outlined above, the change in position is `caused' by the change in the infinitive from nominal to verbal: there is no PP for for to to be placed in; yet it has features interpretationally relevant and is reanalyzed as being situated in C. The use of for by itself is less straightforwardly only prepositional in Otho. In terms of features, it can be said that tense features (connected with the infinitive) are in P in Caligula, but in C in Otho. The features connected with for are more often future. I suggest these features are present in purposive adverbials and in complements to deontic verbs such as (2) above introduced later than Layamon.

In this section, I will examine some other Middle English texts from a variety of dialect areas and times. The conclusions are tentative. Gawain and the Green Knight is from the middle of the fourteenth century and written in a North Midlands dialect. There are 33 instances of for to as in (48):

48. Gawain, 1945

And þat is ful pore for to pay for suche prys þinges

`And that is little to pay for such a costly thing'.

An ending occurs twice on the infinitive. The early loss of endings is a Northern characteristic. Seven objects immediately precede the for to infinitive, the function of which is that of a purposive or future adjunct. This situation is problematic because, even though there is generally no infinitival ending, for to is still positionally close to the infinitive. It is possible that the infinitive is verbal as the endings indicate and that for to is seen as related to the Case of the infinitival object and therefore placed in AGRo as in (44). The object then optionally moves to Spec AGRoP.

The Alliterative Morte d'Arthur was written down by a Northern scribe from a middle to late fourteenth century Midlands original (cf. Krishna 1976) and has a number of Northern characteristics. For instance, again, the endings are lost `early' compared to Southern texts. There are 25 infinitives introduced by for to and 2 of these end in -n. This might indicate a verbal infinitive and a for to that is no longer a P. There is one object (as compared to the 7 in Gawain) that precedes for to (l. 1220) which is problematic for for to being in C as in (43) and may indicate that the structure is (44). The reason for this might be the same as that in the Otho version of Brut and Gawain, i.e. a `misanalysis' of for to as Case assigner. The other objects as in (49) follow the infinitive and are compatible with for to being in C. Most infinitival clauses are purposive adjuncts; some complement verbs such as command, let and wene `think, hope' all with future sense as in (50). Thus, except for one exception, for to is in C and the infinitive is verbal and triggers AGRoP:

49. Morte, 695-6

50. Morte, 2779

Thow wenes for to flay vs, floke-mowthede schrewe,

`You intend to frighten us, flat-mouthed shrew'.

In Piers Plowman, a text in from 1377 (cf. Skeat 1954) from the South Midlands, there are 33 fortos introducing an infinitive; only 4 of these have endings. As in Gawain, there are a number of instances (eight) where an object precedes forto. In addition, there are 60 instances of for to introducing an infinitive, as in (51). Five of these have an ending and again, a number (about 13) of these for tos are preceded by the infinitival object:

51. Piers, P 114-5

and clerkes he made For to conseille the kyng

`and he made clerks counsel the king'.

In Piers, 16 instances of for occur in C as in (52). Three of these have an ending. This use of for is introduced in the late fourteenth century and Piers's use is perhaps the earliest (Visser 1966 mentions mainly first instances from Wyclif and Chaucer). Its introduction seems related to the extension of the Accusative-with-Infinitive (ACI) construction from its Old English use to a use with verbs of mental perception, volition and informing as in (53) (cf. van Gelderen 1993 for a discussion of how this occurred at the end of the fourteenth century). The reason for this expansion is that to is reanalyzed as an element on a par with modals and do, i.e. in I as in (9) above, and the same is true for to in (52). The difference between the ACI and (52) is for: it Case marks the subject of the infinitive whereas the ACI is dependent on a higher verb, and for is connected to purpose/future and is therefore used adverbially or as complement to certain verbs as outlined above:

52. Piers, 12. 015

It is but murth as for me to amende my soule,

`It is but mirth for me to amend my soul'.

53. Piers, 12.229

And Kynde kenned the pecok to cauken in swich a kynde,

And nature taught the peacock to breed in such a way.

(for (53), cf. Visser 1973: 2320)

In sum, one might argue that these three texts present some evidence for the infinitive being verbal, but that the status of for to/forto is not clear and that it is on occasion placed in AGRo as in (44) rather than in P as in earlier texts or in C as in Otho. For is in C in (52) and, as will be shown below, it occurs alongside the forto infinitive for some centuries (contra Lightfoot's 1979: 194 claim that the demise of the for to construction coincided with the intruduction of constructions such as (52)).

In Chaucer's Canterbury Tales written in the late fourteenth century, for to occurs to introduce subordinate verbs approximately 430 times as in (54) to (57), (59) and (60); in all the writings of Chaucer, it occurs 1260 times. It is not restricted to occuring with purposive infinitives as these sentences show, but complements verbs as in (60) and functions as subject as in (59). For to still precedes the VP and is not a C(omplementizer) because extraction of an object is possible as in (57) (unlike in Modern English where for in (58) is a C). For in (61) to (63) is in C; to is probably in I as in (9) (cf. van Gelderen 1993 where it is argued that around 1380, the I position is activated and that to occupies this). Infinitival endings still persist as in (60). In (64), for is in C:

54. Canterbury Tales, Knight's Tale, I, 1955

The Statue of Venus, glorious for to se

55. Idem, I, 1095

This prison caused me nat for to crye.

56. Idem, I, 1255

Som man desireth for to han richesse.

57. Clerk's Tale, IV, 533

This child I am comanded for to take.

58. *This child I expect for her to see.

For unto a povre ordre for to yive

Is signe that a man is wel yshryve,

`Because to give to a poor order

is a sign that a person is well-confessed'.

60. Idem, I, 12-3

Thanne longen folk to goon on pilgrimages,

`Then long people to go on pilgrimage

and pilgrims to seek foreign beaches'.

61. Merchant's Tale IV, 1615

Nedeth namore for hym to go ne ryde,

`Did not need anymore for him to ride'.

62. Miller's Tale, I, 3268-70

She was a prymerole ... for any good yeman to wedde,

`She was a primrose ... for any good yeoman to marry'.

63. Parson's Tale, X, 786

That oother manere is whan men or wommen preyen for folk to avauncen

`The other way is when man or woman pray for people to help

them only out of wicked devotion that they have for the person'.

64. Prologue, I, 77

For he was late ycome from his viage,

`because he had recently come back from his voyage'.

In sentences such as (59), for to `still' precedes the VP (it is unclear whether for to is in AGRo or in I) but it introduces a complement much more often than previously (e.g. than in Otho) and is introducing sentential subjects (cf. Quirk & Svartvik 1970: 399; 401 and Visser 1966: 956ff.) for instances of such verbs). In relation to the absence of such sentences in Caligula, it was observed that if for is a preposition and the infinitive a noun, their absence from subject position is expected (since the PP would be unable to check the Case in subject position). Once for and for to are in C, it is expected that they introduce infinitival subjects.

The tense in the infinitives complementing verbs is straightforwardly future since the verbs used are deontic ones. This indicates that the tense of the infinitive is anchored through the features associated with forto. It is interesting to compare a for to complement with one without for as in (65):

65. Chaucer, Boece V, pr. 3, 117

that every thing be ryght as science comprehendeth it to be. (Visser 2309)

The tense of the infinitive in (65) must be simultaneous with the tense of the main verb and is like Modern English (5). This is not so with for(to), e.g. as in (56), (57) and (63). Again, adapting Enç (1987), if for is a future marker, the tense in the infinitive (which cannot be independent) is linked to that of the future marker. Future complementizers are selected by a special class of verbs, i.e. deontic ones and this is the reason why up to the present day, for only occurs with verbs such as expect, want and love.

In addition to these changes with for to, a reanalysis of for as a C is taking place in Chaucer, as (61), (62) and (63) show. As mentioned, (52) above is one of the earliest instances of fourteenth century use, but Chaucer seems to use them frequently. I have found 14 of these in The Canterbury Tales. Most of the instances do not complement verbs, which Visser (1973: 2244-5) confirms (perhaps too categorically): "the idiom is hardly ever met with before the beginning of the twentieth century". For in (61), (62) and (63) is analyzed as other fors, introducing clauses, as in (64). The features associated with for are the same as those associated with for to, namely unrealized tense/future.

Since Chaucer, double object verbs such as (55) and (57) cease to be complemented by an overt complementizer. I have no explanation for this, but it is not only with tenseless complementizer as the occurence of the now ungrammatical (66) shows:

66. Reeve's Tale I, 4253

That makes me that I ga nat aright.

(cf. Visser: 2238)

In this paper, I do not focus on the position of to. I have argued elsewhere (van Gelderen 1993) that the I-position (which to occupies in (9) above) is activated/introduced around the time of Chaucer (see also Lightfoot 1979 for the introduction of do as in (67) and (68) and a new modal category). Hence, the claim above that to is in I in (61) to (63):

67. Chaucer, The Monk's Tale VII, 2431-2

His yonge sone, that thre yeer was of age,

Unto him seyde, Fader, why do ye wepe?

68. Chaucer, The Reeve's Tale I, 4025

And John also, how now, what do ye heer?

In conclusion to Chaucer, it is unclear what the position of for to is (AGRo or I), but for is in C (and to in I). There is a shift from infinitives only being used as purposives to being used as complements or subjects. In these cases, for(to) acts as a future marker that the non-tensed infinitive can take as reference point. The future is anchored in the tense of the main verb.

Gower's Confessio Amantis was written 1390-3 in a London dialect. It has many Southern characteristics, e.g. the use of hem for the third person plural pronoun. Forto introduces infinitives 693 times as in (69) and (70); for to 4 times. Of the infinitivals introduced by forto, 12 have endings. These infinitives have a purposive and/or future meaning. For is not split off from to in any instance that I can find. There are, however, a number of instances where the infinitival object precedes forto, indicating that forto is not (always) in C:

69. Gower, Prol 52-3

Thus I, ..., Purpose forto wryte a bok.

70. Gower, Prol 58-9

I thenke forto touche also, The world which neweth every dai.

The Chancery Texts written in London from 1417 to 1462 by a variety of people contain 90 instances of for to introducing infinitives as in (71). Except for 2, these have no inflection. Many are purposive; others are future as in (71), but all are compatible with having for to in C, or for in C and to in I. Evidence for the latter possibility is that there are 4 instances of for separated by the subject of the infinitive from to. These have a structure as in (9) above:

71. Chancery, letter 40 (1418), l. 22

if it like you for to here him

`if you'd like to hear him'.

72. Chancery, letter 129 (1435), l.4-5

hit is right nedeful for me to haue certaine personnes þere.

The Paston Letters, written from 1425 to 1495 by different people, contain 182 for tos introducing an infinitive as in (73); one of the infinitives has an -n ending, as in (74). Sentences (73) and (74) indicate that for to infinitives serve as complements with the purposive or future sense:

73. PL, letter 150 (1454)

My modere prayith yow for to remembre my suster.

74. PL, letter 131 (1449)

he desyryd for to speken wyth me.

Pray in (73) is a double object verb and you is the object of the main verb with the second (infinitival) object introduced by for in C. Evidence for the claim that for is in C is that there are 27 instances of for and to separated by an accusative subject of the infinitive, with future, purposive or subjunctive meaning as in:

75. PL, letter 281 (1473)

it were better for me to be owt off syght.

76. PL, letter 363 (1474)

it shalbe necessary for me to come vp to London.

Comparing the earlier letters to the later ones, there is an increase in constructions such as (73) and (74) but not of those such as (75) and (76). I do not examine the reasoning behind this.

Thus, in the fifteenth century, both for to and for are used and the evidence indicates that both are in C.

Leaving Middle English and skipping a century, Mulcaster's Elementary from 1582 has 4 for to infinitives and has 20 or so instances where for is separated from to by the infinitival subject. This shows the decline for the for to infinitive. The reason is that once to can be seen as separate from for, by being analyzed as modals and do situated in I, it is hard not to exclude the subject from Spec IP where for assigns Case under government to an otherwise Case-less infinitival subject (cf. Chomsky & Lasnik 1977; Chomsky 1981 for a discussion). Shakespeare's 1623 First Folio Edition of 6 plays contains no for tos introducing infinitives and has very few for complementizers introducing to infinitives. The only examples in the 6 plays that I have found are (77) to (79). These examples involve adjuncts. Thus, Shakespeare's use of for as a complementizer with future meaning is less extensive than that in earlier and later periods:

77. King Lear III, 4, 2-3

The tirrany of the open night's too rough For nature to endure.

78. 1 Henry 6, V, 3, 58

And hell too strong for me to buckle with.

79. As You Like It, III, 2, 180

it is a hard matter for friends to meet.

Texts of that time waver between using some fortos in addition to for separated by to. Francis Bacon's early seventeenth century Essays are similar to Shakespeare: no for to and some fors separated by to. Bunyan's late seventeenth century work has both. By the time of Swift and Hume (beginning and middle of the eighteenth century respectively), no forto is left, but constructions such as (77) to (79) are.

Examining earlier stages or dialects, it becomes clear that for and to have always been `floating' elements, prone to reanalysis. This is perhaps the nature of prepositions that are among the most general location and direction indicators. In some languages they are seen as indicators of Case (e.g. Old English); in other languages, they indicate tense (Modern English). In a Minimalist framework, this means that they have different intrinsic features and therefore are linked to different positions. In this section, I examine for and for to in some related languages and variants. The generalization (from Universal Grammar) is that when they are relevant to Case, they will be in P or AGRo but when they are relevant to tense, they will be in C (or I).

In a study of Belfast English, Henry (1995: 82) mentions the widespread use of for to introducing purposive infinitives as in (80):

80. I went to the shop for to get bread.

This is similar to the situation described in Middle English: for to is in C and connected to future. In one variant of Belfast English, however, (81) and (82) are grammatical. Henry (1992; 1995) argues that for moves to the right and cliticizes to to (in I):

81. I wanted Jimmy for to come with me.

82. I believe them for to have done it.

The constructions in this variant are comparable to the Middle English with the structure of (44). My analysis for Middle English differs in that I claim that both for and to are in AGRo. In Belfast English, there is no evidence that for to is responsible for Case. In addition, there is a difference in meaning. In Middle English, for carries the meaning of future or purpose but in the variety of Irish English that has (81) and (82), "for does not seem to have a semantic contribution, since any infinitive can occur with for" (Henry 1995: 12). In a framework such as Kayne (1994), a rightward movement of for the way Henry suggests is problematic; one might claim that for to is in I, connected with intrinsic non-tense features. The reason for to can be in I is that future is not connected to it.

In Dutch, the counterpart to for is om. Om in (83) is used to complement deontic verbs and to introduce adverbial infinitives, as in Modern English, indicating future, and is positioned in C. The same is true in Standard Afrikaans (84):

83. Ik sta erop om nog snel even hem te zien,

I insist on it for quickly him to see.

84. Hy bied aan om met hulle saam te werk,

He offers for with them together to work

`He is offering to work together with them'.

(from Ponelis 1968 [1973]: 125]

Du Plessis's (1984) facts seem to suggest that in a variety of Afrikaans the equivalent of for to, namely om te in (85), is in C. The sense om te carries is future and it can be seen as a unit especially since, in this variety, te is often doubled as in (86). In yet another variety (Ponelis 1993: 64), om te as in (87) indicates future and is in C but te is optional as in (88):

85. om te daar wegkom, Meneer.

for to there away get, Sir.

86. om te ok vir ons te help,

for to also us to help.

87. om te die waarheid sê,

for to the truth tell.

88. om die waarheid sê.

Other languages indicate the same `floating' quality of elements like for and to. Winford (1985) discusses a number of proposals (by Bickerton and Washabaugh among others) regarding the categorial status of fi `for' in creole variaties. Demske-Neumann (1993) argues that Gothic du `to' is in C and Louden (1992) discusses the complementizer status of fer in Pennsylvania German. More work is needed in these varieties on correlating Case features and AGRo; future/purpose and C; and tense features and I.

I have examined the structural position and function of for and forto as these elements grammaticalize. I suggest that both for and forto change their intrinsic features from Case to future/purpose features. As this change occurs, the category changes as well. For to is initially a Preposition connected to Case but as it acquires a future sense, it changes to C (with an intermediate stage as AGRo). This development is related to (and possibly caused by) the change of the infinitive from nominal to verbal. The relevance of these conclusions to Universal Grammar is that functional categories such as C and AGRo attract elements with future and Case features respectively; as the features change through grammaticalization, they change categorial position.

Primary

Bacon, F. 1601 [1904]. Bacons essaies; a facsimile edition. New York: Dodd, mead & Co.

Davis, N. 1971. Paston Letters and Papers. Oxford: Clarendon Press.

Eddy, W.A. 1933. Swift's Gulliver's travels. Oxford: Oxford University Press.

Klaeber, F. (ed). Beowulf. Boston: Heath and Company, 1950 edition of 1922.

Krishna, V. 1976. The Alliterative Morte Arthure. New York: Burt Franklin & Co.

Macauley, G.C. 1899-1902. The Complete Works of John Gower. Oxford: Clarendon.

Secondary

Abney, S. 1987. The English Noun Phrase in its Sentential Aspect. MIT PhD.

Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris.

Chomsky, N. 1995. The Minimalist Program. Cambridge: MIT Press.

Kayne, R. 1994. The Antisymmetry of Syntax. Cambridge: MIT Press.

Lyons, J. 1977. Semantics 2. Cambridge: Cambridge University Press.

Ponelis, F. 1993. The Development of Afrikaans. Frankfurt: Peter Lang.

9. CP

. C'

C IP

for . I'

[fut] her I VP

[tense] . V'

to V NP

see Zora

28. PP

. P'

P NP

for to . N'

[Case] N NP

habben al

43. CP

. C'

C AGRoP

for to . AGRo'

[fut] londes AGRo VP

^ NP V

t seche

44. AGRoP

. AGRo'

his folk AGRo VP

for to . V'

[Case] V NP

stewe t

THE FUTURE OF FOR TO

Elly van Gelderen

Summary

I examine the development of for to and for in several stages of English. I argue that as prepositions grammaticalize, they acquire certain intrinsic features and occupy special positions. In Old English, for to is a P, related to Case, and has some future sense (through an extension of the locative meaning). Verbs do not subcategorize for complements with for (yet), however. In Early Middle English, for to is used to introduce a complement with future meaning. Now, for (to) occupies C, which I assume is universally true for purpose/future indicators. In late Middle English, the situation solidifies and more verbs select a complement with for(to) indicating purpose and futurity. In Early Modern English, for to disappears but for separated from to takes over its function to introduce purposive adjuncts and future complements.

Notes